Datasets:

mimiklee

/

masterthesis_clean_text_20k

like 0

INTRODUCTION Social class gradients in health are ubiquitous across time and geography. An awareness that social status or position is related to individuals' health is evident in the writings of Hippocrates, and has been quantitatively described in populations since Petty and Graunt in the 17th century, and Farr in the 19th century counted and summarized deaths and death rates . From these earliest systematic collections of population health data was evidence that individuals and groups with greater power, wealth, and prestige typically have better outcomes than those with less or without. These patterns persist into the early 21st century both within and between nations for wide-ranging processes and outcomes including cancers, chronic diseases, infectious diseases, and neurodevelopment. Concern for social class variation among children engages the concern of scientists, policymakers and families for at least three reasons. First, children occupy a privileged space in most industrialized cultures, being both protected throughout childhood and valued for their future potential. Therefore harm done to children is less tolerated than similar harm done to adults. Second, children 'receive' their social class from their families, and as such they are not causally responsible for their status in life. Finally, rapid growth and developmental plasticity make children particularly sensitive to the stimuli of their environments, with lifelong consequences. Stimuli and exposures during critical developmental windows, or cumulatively across childhood can illicit disease and disease processes that persist into adulthood. These three actions of social class become clearer when placing them in the context of current theory exploring macrosystemic dynamics and processes as they relate to health. The allocation of, or opportunity to, acquire material resources action is paralleled by "fundamental cause" theory or the belief that social stratification of individuals within society leads to variation in the money, knowledge, power, and interpersonal resources necessary to prevent and manage illness, resulting in social disparities in health . A second theoretical approach dealing with distribution and life chances emerges from political economy, focusing not only on the social hierarchy of resources, status, and power, but on the political and social production of inequality . This perspective goes beyond asking about the health consequences of being poor or non-poor to asking why and how some groups come to have less than others . Political economy draws attention to the values, priorities, and structures of systems which produce the context in which children develop. A third theoretical frame focuses more on salient group identification and intergroup conflict, and the role of relative position within a social hierarchy as the driving force for social class health patterning . Thus, in this perspective, it is not an absolute level of material resources that produce health, but instead the presence of status inequality between groups, and the psychosocial stress resulting from awareness of one's location within a more or less unequal hierarchy. --- Theoretical foundations -bioecological model of human development The conceptual framework guiding this manuscript is fundamentally informed by bioecological theory of human development . A developmental theory is essential for understanding variation in children's health, like health disparities, because health is a manifestation of development. Regardless of whether the focus is on infant birth weight, motor vehicle-related childhood fatality among toddlers, or any number of specific conditions like diabetes or asthma; children's health risks and problems typically develop over time, thereby necessitating a developmental theory. The proposed model uses three core ideas from bio-ecological theory. First, our model adopts the concept of proximal process as the primary engine for child health. Bioecological theory contends that human development occurs through proximal processes, which are defined as progressively more complex interactions between the child and the persons, objects, and symbols in the child's environment. For example, the American Academy of Pediatrics recommendation for child feeding represents a codified proximal process. The AAP recommends exclusive breastfeeding for the first six months of life , with the gradual introduction of solid foods around six months of age, and the addition of healthy snacks around 9 months of age. In this case the proximal process is an interaction between the child and the parent that becomes increasingly complex and is presumed to contribute to optimal physical growth. A simple extension of this idea to child health more broadly suggests that poor health or morbidity results from stagnant or pathogenic interactions between the child and his or her environment such as those illustrated by a parent who provides no solid food to their child for the first year of life, or more extreme conditions such as abuse or maltreatment. A second feature of bio-ecological theory reflected in the proposed framework is the recognition that child development results from nested interactions. This feature is visually depicted in our model in terms of "child health" being nested within the Family and Home Microsystem, with the nesting represented by the same geometric shape used to identify "child health" and distinct elements of family and home. Further, the Family and Home Microsystem is nested within Children's Primary Mesosystems, a feature illustrated in the model by the placement of the microsystem within the three dimensional space of the mesosystem. Finally, both the Family and Home Microsystem and Children's Primary Mesosystems are nested within systems of social stratification represented in terms of social class in our model. The third and last feature of bio-ecological theory captured in this framework is the premise that proximal processes, the drivers of development , can be promoted or inhibited by more distal features of children's environments. This feature is reflected by the arrows flowing from left to right . Social class and its core actions are posited to operate through two parallel levels of children's environments. The first level influenced by social class is the mesosystem, which is conceived of as the institutions and establishments that children and their families interact with on a regular basis. Primary mesosystems in children's lives include schools , the health care delivery system, neighborhoods, and employment opportunities for parents. The actions of social class shape Children's Primary Mesosystems, albeit in complex ways which are illustrated by the "bent arrows". Just as light entering and exiting a prism is bent based on its naturally-occurring or engineered angles, likewise the effects of social class on children are "bent" or otherwise modified depending on naturally-occurring or engineered features of children's mesosystems. Children's Primary Mesosystems, in turn affect the Family and Home Microsystem and the subsequent proximal processes contributing to child health. A quintessential point of the proposed framework is that it steps beyond describing, toward explaining social class variation in children's health. That is, social class recreates itself through a multistep process. In the first step, social class contributes to the financial and material resources and the sociopolitical dynamics of Children's Primary Mesosystems, which, in turn, shape children's Family and Home Microsystems in the second step. In the third and final step, interactions within children's families and home set proximal processes relevant to health and illness into motion to create children's health outcomes. Social class can then be recreated because differences in children's health at birth and accentuated throughout childhood influence the social class children acquire as adults . --- EMPIRICAL REVIEW In order to highlight empirical evidence relevant to the theoretical foundations presented above, we briefly review literature illustrating current knowledge for the manner by which class patterns child health. Our review is limited to empirical evidence from high-income countries, recognizing that the manner in which social and economic stratification are experienced in low-and middle-income countries could be substantially different. An overarching objective is to view the extant literature through the lens of the multidimensional actions of class as characterized by Wright. Therefore we take this opportunity to characterize current evidence for how these actions of class might be embodied as child health, as well as pointing out areas where fuller examination of the mechanisms connecting class and health might be informative. To aid in focus and coherence in this section, we limit our review to child physical health outcomes. This in no way suggests that class is not relevant to cognitive, socioemotional, or mental health of children; on the contrary we believe these are inter-connected with physical health. We hope that the issues raised in the illustrations and discussions that follow can be readily translated beyond physical health. --- Mesosystemic forces that shape child's environment and family interactions Employment-Parental employment is a resource that is not evenly distributed in society, but yet has substantial health implications for children. Recent evaluations of the Great Recession provide clear evidence that job loss and unemployment was not equally shared, but rather concentrated in areas with high concentrations of racial and ethnic minorities and individuals with low educational attainment . Areas with elevated unemployment experience greater unintentional injury of children , elevated preterm births , greater initiation of marijuana by adolescents , and more child maltreatment . These results, connected with others wherein area unemployment is combined with other metrics of area socioeconomic deprivation and linked to health outcomes like morality suggest that jobs are a resource, that when scarce, impedes healthy child outcomes. Jobs are also a distribution channel for additional resources at both the macro and micro levels. At the macro level community job loss has effects for all children, even those whose parents retain their jobs because higher unemployment compromises tax revenues resulting in lower governmental allocations for local health departments as well as education . Additionally, jobs are also used to distribute financial resources like earnings and health insurance. Indeed, employment-based insurance is the primary source of insurance coverage for Americans and substantial evidence documents a close link between job loss and being uninsured . Other resources that are built into jobs, such as work schedules that can be "flexed" to accommodate children's needs and are advocated for promoting breastfeeding , are frequently less available to individuals in lower social classes . Substantially less research has considered the link of other actions of social class through employment to child health. Theory and research elaborated in the Parenting section of this manuscript documents the possibility of social class variation in parenting that may socialize children to better affiliate with salient occupational groups , and potentially serve as a source of intergroup conflict between these occupational groups. If the labor force continues to bifurcate with increasing value placed on science, technology, engineering and math fields, pressures placed on children to acquire the characteristics of STEM professionals may create fertile ground for negative health outcomes in both the short and long term . These are promising areas for future research. Health care system distribution, access, and quality-Access to and interaction with the health care system is shaped by family socioeconomic position and is likely to contribute to class-based inequities in children's physical health . Access to health care is measured in a variety of ways , including having health insurance, access to a usual source of care, or utilization of needed services . Children from families with lower incomes are less likely to have health insurance, less likely to have had a medical office visit or utilized a hospital outpatient center in the last year , and less likely to get all the recommended vaccines as compared children from families with higher incomes. Lower income children are also more likely than higher income children to have made trips to the emergency room, had problems getting a necessary treatment or had a referral to a specialist , more likely to have delayed care , experienced increased risk for hospitalizations , and have higher likelihood of ruptured appendicitis . These differences in access to, and utilization of, care exist despite the fact that, in the US, lower income children are eligible to enroll in public insurance through either Children's Medicaid or the Children's Health Insurance Program . As of 2013, an estimated 88% of income-eligible children were enrolled in the program . Beginning in January 2014, the Affordable Care Act extended Medicaid eligibility to low income adults in the 32 states that opted into the Medicaid expansion component and provided subsidies for lower-middle income families in all states to purchase insurance, thus potentially covering many more lower-middle income children. It is not yet clear whether having non-Medicaid/CHIP insurance will help ameliorate the inequities in getting timely preventive and medical care. Unlike Children's Medicaid and CHIP, often private medical insurance through employers or purchased through the ACA requires premiums, deductibles, and co-pays for office visits and medications that can act as a deterrent to seeking care for lower income families. In this context inequities in access to quality health care may stem primarily from classbased resource allocation processes-for lower middle and middle income children, the high monetary cost of co-pays and premiums are a cost barrier, and for children covered by public health insurance, the monetary cost of health insurance is not the only cost of health care visits. Additional costs include lost wages, transportation costs, and time costs. Other resource allocation pathways could include the shortage of health service providers that accept patients with public insurance or more broadly the presence of geographic health professional shortage areas . However at the intersection of race and class, other actions of class may be at play including inter-group conflict and salient group membership in accessing and delivering high quality healthcare. African American populations, who are disproportionately represented among those with low-incomes, have experienced repeated betrayals of trust by medical and research communities and continue to experience discrimination in the medical setting which likely contributes to higher levels of distrust of medical professionals seen among these populations . This distrust may lead to delaying preventative care or treatment. Strengthening causal inference for the relationship between access to health care and health outcomes requires methods or designs that can overcoming the major sources of confounding that exist between the insured and uninsured populations . Both the Medicaid expansion component and the provision of subsidies to purchase health care could be leveraged as natural or quasi-experiments in order to better understand the impact of access to care on health. A handful of studies have begun to examine the impacts of ACA on health outcomes. These have focused primarily on adult health since adults' insurance coverage is most affected; however, future studies should examine whether child health responds to changes in parents health coverage through the ACA as lack of insurance has been cited as a source of family stress and a barrier around providing optimal care for children . Neighborhood conditions-Because of historical processes of racial and economic residential segregation, children living in poor families often live in impoverished neighborhoods . This means that poverty is not only an individual or family experience, but to the extent that high poverty neighborhoods have fewer amenities and greater social discord, it is also a community experience. In the U.S. in particular, poverty and race go hand-in-hand to impact children's neighborhood contexts; whereas only 1.4% of white children live in both poor families and poor neighborhoods, 16.8% of black children experience this "double jeopardy" . Empirically, neighborhood environments have been linked to a wide range of child health outcomes including body mass index , overweight and obesity , asthma and and blood pressure. . Much work reports that women living in disadvantaged neighborhoods are more likely to deliver infants preterm or low birth weights , suggesting possible intergenerational effects of neighborhood context, whereby the influence of disadvantaged environments on child health begins in utero or prior to conception. A small body of research has examined the specific pathways by which early-life neighborhood exposures influence children's health. Chen and Paterson report that lower neighborhood socioeconomic status is associated with lower basal cortisol in adolescence, suggesting that neighborhood disadvantage may dysregulate the hypothalamic-pituitary-adrenal axis, which is in turn associated with regulation of many physiological processes and health outcomes. The 'neighborhood effects' literature has been critiqued on numerous counts including differentiation of selection versus causation, the challenge of identifying effects when there are few poor individuals living in salubrious environments , and measurement of contextual versus compositional features of neighborhoods . Beyond addressing the challenges of causal inference, there are opportunities for enriching the conceptualization of class-based neighborhood effects with respect to the proximal processes for child health. Much of the neighborhood effects literature implicitly or explicitly posits that the drivers of neighborhood health gradients are distributional in nature, focusing on the allocation of resources and services to neighborhoods. However neighborhoods also represent relational spaces where salient group identity and inter-group conflict may play out in ways that impact child health . An example of the importance of salient group identity in neighborhoods concerns the production of local social norms. Neighborhood, community, and school social norms influence health behaviors such as tobacco and alcohol use among adolescents . The notion of salient groups may also come into play as a moderator of "neighborhood effects" if individuals' interactions with their neighborhood's service, physical, or social environment differs depending on their affiliation with other salient class-based groups. For example, the role of school environments on children's BMI appears to differ by household-or neighborhood-level socioeconomic status . Much less attention has been paid to investigating the role of class-based antagonistic or inter-group conflict within neighborhoods and their effects on child health. For example, one line of inquiry could investigate how potentially conflict-generating processes such as gentrification and urban re-development, affordable housing, zoning, ordinances regarding property upkeep, policing programs , etc. affect the health of both the "beneficiaries" and "victims" of such programs. A few studies have attempted to explore the effects of neighborhood change and gentrification on birth outcomes , but this work is in its infancy and remains in need of advances in exposure measurement and replication. The relatively large literature on factors such as social capital, social control, and collective efficacy could represent actions relevant to both salient group identity and inter-group conflict processes of class. Because collective efficacy may be strengthened in areas of racial, ethnic, or class homogeneity, identification with salient groups could promote cohesion. On the other hand, cross-class efforts at social control-for example around expectations for behavioral or aesthetic norms such as yard care-could result in inter-group conflict. While some empirical work has found that social cohesion, social control, and/or collective efficacy are associated with better health in adults, little work has examined these characteristics with respect to children's physical health. Such work could follow the model of Donnelly and colleagues who found that adolescents who grew up in neighborhoods with high collective efficacy exhibited fewer depression and anxiety symptoms compared to those who grew up in neighborhoods with lower collective efficacy. Education System-Educational attainment is one of the strongest predictors of adult morbidity and mortality , and the expansion of compulsory education has been called among the most important health intervention instituted in countries around the globe . While the association between education and health and longevity is widely documented, debate about the causal direction of education and health exists . The lifecourse and transgenerational reproduction of education and health represent complex dynamics, and while health selection and residual confounding are not unimportant, the bulk of the evidence is consistent with a causal beneficial effect of greater education on health . Though the link between education and cognitive development in children is most evident, educational policies and institutions can also directly impact child physical health through policies of mandatory vaccination or physical exams, and programs promoting physical education and nutrition. For example, schools can facilitate the delivery of services to children who are food insecure and lack health and dental care. In the U.S., the National School Lunch Program provides free or subsidized lunch for families at or below 185 percent of the federal poverty line . Research has suggested that food insecurity is associated with obesity, lower energy, and more internalizing behavior problems in children ; therefore providing meals during the school months may reduce negative health outcomes, improve concentration and provide a reliable source of nutrition and food for low-income children.. Educational systems also afford opportunities for children to be tested for vision, dental, and health issues through various screening programs. Most of these screenings are mandated by states and provide an avenue for identification and referral for identified with conditions that may be problematic to their learning. Unfortunately, the effectiveness and the amount of services vary from state to state with many of the highest need schools providing fewer services. Even though the infrastructure exists for providing services across class groups, as discussed previously in the section on neighborhood effects, there are still stark differences in how these children fare in poorer schools with large income gaps in achievement and unmet needs still present . Beyond the service environment of schools, the quality of school infrastructure could also affect health in class-dependent manner. Poorly funded schools are often built in areas and of materials that may be high in toxicants that have consequences for achievement and behavior . There appear to be no safe levels of lead exposure for the developing child and lead can be found in old buildings and contaminated soil. Almost every major city in the U.S. is dealing with the issue of lead exposure in children and how both the home and school environment relate to exposure is important for understanding class disparities in children's health. Services such as school lunches and health screening represent one way that educational institutions and education-related policy seek to buffer or address disparities in the allocation of resources to children. The programs are successful to the extent that supply of needed services meet the distribution and intensity of demand. One emerging area of relevance for class and race-based disparities that arise from a different action of class is concern about the 'school to prison pipeline' and the disproportionate representation of poor black boys in school disciplinary events . One explanation for the disparate rate of school suspension of low income black and Latino children is conflicting expectations about 'normal' behavior and unequal tolerance for deviation from class-based norms . --- Microsystemic forces that shape children's interaction within family contexts Parenting and Family Resource Management- Kohn and Schooler were among the first to comment upon and document social class variation in parenting activities. They argued that parents in middle-and upper-class families valued promoting children's independence, decision-making and self-direction, whereas lower-class families valued or prioritized children's deference to authority, rule-following and conformity. Kohn and Schooler surmised these distinct parenting values were held to prepare children for the occupation they would likely occupy in the future. Subsequent research has used class-based differences in parenting style to explain associations between childhood socioeconomic status and child and adult health outcomes including BMI trajectories , and adult metabolic function . Lareau presented compelling ethnographic evidence suggesting that middle-class parents take a fundamentally different approach to parenting than working class parents, and that differences in parenting were manifest in children's own beliefs. "Concerted cultivation," the form of parenting manifest by middle class parents, was characterized in terms of purposeful nurturance of children's special gifts and interests through organized and structured activities. Children raised under this parenting model articulated feeling "special" because of the activities organized by their parents, and feeling "bored" if time was not sufficiently filled with structured activities. By contrast, "accomplishment of natural growth," the approach to parenting observed by lower class families, was characterized in terms of parents providing basic care and allowing the child to grow. Children raised under this parenting model had a more "go with the flow" everyday experience that responded more to emergent experiences than planned events. Analysis of the Early Childhood Longitudinal Study, Kindergarten Class reported that fully 50% of the variance in their measure of "concerted cultivation" was explained by multiple indicators of socioeconomic status, with the strongest effect leveraged by parental educational attainment . This logic is consistent with Wright's notion of subjectively salient experiences. That is, class generates children's subjectively salient experiences-one of which can be exposure to systematically different parenting styles or practices; this exposure helps steer children's identity as being located on the lower end of a system of economic stratification. Parenting is closely related to the topic of family resource management, which, according to the National Council of Family Relations, focuses on "decisions individuals and families make about developing and allocating physical, psychological and social resources such as time, money, material assets, and energy." Lareau's work on class variation applies here as well. "Concerted cultivation" parenting expends more financial resources and structured time; by contrast, maintenance of kinship networks with family, including liberal opportunity for emergent play and interaction with older and younger peers within the family is exercised more deliberately by parents embracing "accomplishment of natural growth." Although not written under the general framework of "family resource management," Wither Opportunities draws substantial attention to how socioeconomic status creates substantial differences in how distinct household resources are allocated. This logic aligns closely with Wright's view of class as generating an unequal allocation of resources and/or life chances. When the focus is placed on using financial resources to provide children with enriching experiences like sports or travel, individuals in the upper quintiles of household earnings spend between four and 10-times the amount of money than individuals in comparably sized families in the lowest quintiles of household earnings . In terms of the allocation of family time, mothers with higher levels of educational attainment spend more time engaging in literacy activities with their children and taking them to novel locations like museums . The increasing emphasis placed on allocating time and financial resources to enriching children's development has contributed to an expansion of the academic achievement gap between rich and poor and leads some to contend that "class differences in childrearing is on the rise" . No research we know of has explored the notion of class as generating subjectively salient groups when examining parenting practices and child health. The extant literature relating class-based parenting activities to children's health inequalities is more closely aligned with Wright's view of class as producing an uneven allocation of resources and life chances. One prime example is that parents of different social classes vary in their ability to ensure access to healthy foods and also engage in varied feeding practices; both are key factors shaping children's diet and development of healthy eating habits-two critical proximal processeswhich are then reflected in outcomes such as overweight or obesity . Overall, this literature is relatively nascent and deserves further attention; additionally, exploring class through the lens of subjectively salient groups would enrich future research on this topic. Family structure, stability, and interpersonal dynamics-Family structure -in terms of number and relationship of caregivers -is strongly associated with socioeconomic status, and has been consistently associated with indicators of child health and development. Children living in single or stepparent homes are generally in worse health , show elevated levels of health vulnerability , participate in fewer routine medical and dental visits , and have worse access to preventative health care than their counterparts living in two biological parent families. Child health is further stratified by parental cohabitation status. For example the risk of asthma diagnosis is highest among children of single mothers, intermediate among children of cohabiting mothers, and lowest among children of married mothers . A similar pattern is evident for birth weight: the highest weights at birth are found among infants born to married mothers, intermediate weights among infants born to cohabiting mothers, and lowest weights among infants born to single mothers . Much of the literature on families in the U.S. has transitioned away from simple dichotomies of biological parent families versus all other types , and now commonly distinguish families according to other criteria related to family processes and parental dynamics, such as parental stability, father involvement, and/or quality of parental relationships . For example, instability in parental relationships is associated with worse child health and family instability mediates at least a portion of the association between maternal depression and lower levels of childhood health . Father's multiple-partner fertility is also linked with poorer childhood health, and this association is mediated at least at in part by a reduction in the frequency of father's engagement with his child on a weekly basis . However higher levels of engagement of fathers who reside in the home-regardless of whether they are biological or "social" fathers -are positively associated with child health . No studies we are aware of have examined children's physical health across heterosexual and gay/lesbian families, but, there is evidence suggesting children's mental health is not significantly different across heterosexual parent families and same-sex parent families . Siblings are also an integral part of families-of-origin, and potentially make important contributions to children's health. Older siblings can function as an additional source of learning and socialization for younger siblings, and can directly and indirectly influence the behaviors and attitudes of younger siblings, which may in turn be linked with health. Sibling interactions can spur social-cognitive development while sibling conflict is linked with deviance and substance use ; presence or absence of high-quality sibling relationships, comprised of warmth, closeness, and lack of conflict, also mediate the association between single-parent family structure and children's substance abuse, sexual risk behaviors, and risk of contracting a sexually transmitted infection . Having multiple siblings is generally associated with a dilution of family resources ; this may also adversely impact a child's health. As with other microsystem processes, much of the literature on class, family structure and child health has been grounded in a life chances and/or unequal distribution of resources framework. In contrast, we suspect the class actions of subjectively salient groups could generate new insights. Social classes generate subjectively salient experiences such as growing up in an economically disadvantaged single parent family. These experiences in turn critically shape the schema people use to locate themselves and others within a system of economic stratification. As an example, we speculate social isolation and support at the family-level could reflect class-based processes generating subjectively salient experiences, such as multiple spells of unemployment that prompts a series of residential moves, putting family members at risk of social isolation. At a more basic level, social isolation and social support have powerful and enduring effects on health. The literature on adult health shows this very clearly [see for example ]. Some evidence along these lines has been shown for health early on in the life course. Low-income children whose mothers are socially isolated exhibit higher rates of serious accidents, injuries, and poisonings . Maternal perceptions of instrumental support predict higher levels of child health , whereas greater household chaos predict lower levels of child health . In sum, health is likely maximized among children who belong to a family in which both the parentchild and sibling relationships are characterized by high levels of closeness and warmth, as well as high levels of social support which may buffer the child from the full effects of stressful events. Housing-While family structure, dynamics, and resources are primary elements of the microsystem, the intimate physical context of family life is the home, and this environment is also part of children's microsystems. As a source of protection from the elements, animals and insects, and a place to safely store valuables and food, housing is-in Maslow's hierarchy of needs-a fundamental human necessity. A lack of housing, or, homelessness, is associated with higher levels of food insecurity and poor nutrition for children, as well as increased behavioral problems and developmental delays, relative to housed poor children . For housed children, the quality and stability of the built structure and utility services provided to households can also affect health. Quinn and colleagues found that children living in homes with more infrastructure and service stressors had lower self-rated health, reduced controllability of asthma, and more unplanned medical visits. Exposure to dust mites and indoor air mold appear to increase risk for the development and exacerbation of asthma in children . There is also strong evidence for the negative effect of specific toxicants in the home, such as lead, on children's health outcomes. Lead paint dust is well-known to cause severe and irreversible cognitive development problems and surveillance data indicates that dangerously high levels of lead are 12 times as common for low income children compared to higher income children . Crowding is an additional housing feature that is more prevalent among lower-income populations and has been associated with increased risk for childhood respiratory and gastrointestinal diseases , potentially due to a higher degree of infectious disease transmission among individuals living in crowded conditions . Crowding has also been associated with inadequate sleep among adults and lack of personal space is thought to contribute to social and emotional challenges among adults and children; however, few studies have looked at this specifically . Housing quality and instability can also indirectly affect children's health through their impact on family processes such as high levels of parental stress or depression and increased parental conflict . Overall, the literature on class-based differences in housing and child health rely on an allocation and distribution action of class as the motivating process. Whether salient group affiliation or intergroup conflict further transmits class differences to child health disparities is unclear but potentially enlightening. Embodying social environments-In summary there is varied evidence for the manner in which the actions of social class shape and affect the meso-and microsystem environments to which children are most exposed. To the extent that the class-based influence on these environments affects the critical proximal processes of child health and development, the social environment may be physically embodied as illness or wellness. These processes are evident across the lifecourse of individuals, but because of the transgenerational nature of class, may also be communicated from one generation to the next. Most efforts to examine the actions of class implicitly frame the question in terms of differential allocation or distribution of necessary resources. This is clearly a potent and intuitive action. However opportunities may exist for further understanding actions of classand possible unintended consequences of social policies -including the production of salient group experiences and antagonistic inter-group conflicts. If these sources of experience also serve as proximal processes they too become part of the reproduction of class and class-based health disparities. --- HIGH PROFILE TARGETS FOR FUTURE RESEARCH In the end, what contributions are made by this high level review focused on how social class "gets under the skin" to affect children's health, and what next steps are needed? First, although the idea of socioeconomic differences in children's health is not new , this paper provides one way to organize the broad domains of relevant research and theory across multiple disciplines into a coherent conceptual model. The fact that there is clear theoretical rationale and corroborating empirical evidence for many of the pathways proposed in the model speaks to its potential utility for building an integrated understanding of population variability in children's health. Second, despite its high level nature, this review points to several discrete ideas, such as the importance of the conceptual meaning of social class and its diverse modes of action that have unique and distinct implications for children's health. Another discrete and unique point uncovered by this review is the diverse means by which social class "gets under the skin". Psychosocial stress and the biologic sequelae are meaningful and deserving of attention. However, we also highlight other conceptually more direct and potentially actionable pathways, such as paraoccupational exposure to disease agents or differential exposure or vulnerability to environmental toxicants , aging infrastructure , and poor housing stocks. Finally, this review illustrates the complexity of child health and development and the corresponding need to work across boundaries separating disciplines and the worlds of "discovery" and "practice" to protect the health of all children. Contributions notwithstanding, this review highlights multiple areas for additional work. The upcoming areas highlight ways of operationalizing Wright's actions of social class and "high priority" areas for future basic and policy research. The identified items are considered "high priority" for several different reasons; either because the item addresses a basic critical gap in the knowledge base, because it would generate substantial impact, or because it would open doors that have historically been closed. --- Operationalizing Wright's Actions of Social Class The framework linking social class to child health and the evidence provided in the empirical review above highlight the importance of creating theoretical and empirical clarity about how the three different actions of social class influence child health. Thus, it is necessary to consider how one might operationalize the framework and what difficulties might arise in the process. The biggest lesson from our framework is that researchers should use different indicators of social class depending on the actions being tested. Some health pathways will call for distributional indicators such as income, wealth, or education; others will need to be more focused on salient groups or antagonistic conflicts using occupational prestige or latent classifications of groups as indicators. In addition, researchers need to be sensitive to the way that different spatial and temporal scales influence indicators of social class. For instance, social class might be operationalized at the level of individual persons, households, or spatially across neighborhoods, each reflecting a distinct manifestation of social class, and each measuring different aspects of the distribution of resources, status and power. Similarly, many measures are temporally dynamic throughout the life course of an individual, across generations within families, and through history of successive cohorts, necessitating attention to the timing of measurement . For example, technological progress has shifted the relative social position of people with different levels of educational attainment. Cohorts born in the early twentieth century in the United States had low high school and college completion rates of approximately 10 and 6 per cent respectively . The relative prevalence and earnings of each group has changed dramatically as more people attended secondary schools and universities. High school graduates have seen a relative decline in their wage premium compared to 8 th grade graduates from 35 percent in the early twentieth century to 20 percent or less by the end of the twentieth century. College graduates, on the other hand, have seen an increase in their wage premium versus high school graduates from 31 percent in 1950 to 62 percent in 2005 suggesting an increase in the relative status and resources returned from a fixed level of education . These variations over time are also matched by spatial variations in high school and college completion rates and the relative status and wage premium of degrees . Cohort and period influences on these various indicators of social class highlight the importance of contextualizing studies in time and space when analyzing how social class influences child health; explicitly examining cohort and period patterns in child health trends may contribute to our understanding of the dynamics of child health disparities. --- Basic Research The first high priority area for research is the need for more studies that use experimental or quasi-experimental designs. The vast majority of studies covered in this review are based on observational, albeit complex observational, designs such as those deployed in the National Longitudinal Study of Adolescent to Adult Health and Early Childhood Longitudinal Program studies that use rigorous sampling procedures and multimodal data collection techniques. Nevertheless, strong causal inference demands random assignment to change agents or designs that mimic these features. Experimental studies of overt attempts to manipulate some feature of socioeconomic position do exist. The Perry Preschool Project, for example, allocated educational resources and family wraparound services to promote parental labor force engagement in low-income families. The program essentially simulated an upward shift in social position through the provision of sustained high-quality childcare for low income children and resources to enhance human capital for parents. Long-term evaluations of the Perry Preschool Project demonstrate a substantial return on investment in terms of benefits to individual children and broader societal benefits . In contrast results from the Moving to Opportunity program, another simulation of an upward shift in social position by allowing impoverished families to move into better neighborhoods, are mixed. Specifically, results indicated greater rates of depression and other behavioral health outcomes among boys who moved, but reduced rates of depression and conduct disorder among girls who moved . More recent results suggest economic effects in young adulthood varied by the age of the child at the time of intervention, with positive economic outcomes in young adulthood for children who moved before age 13, but either null or detrimental effects for moves after age 13 . Together, these findings highlight the complex, and often unintended consequences of social interventions and the importance of their timing in the life course. Experiments like the Perry Preschool Project and the Moving to Opportunity program make a valuable contribution to the literature, but they also have shortcomings. A simple shortcoming is that "health" is an atypical outcome in many of these projects; consequently, research designed to delineate variation in meaningful and age-appropriate health outcomes across assignment groups is needed. Perhaps the greatest shortcoming of existing experimental work is the inability to attend to distinct putative effects of competing elements or actions of social class. Building on Wright's work we propose three distinct functions or modes of action, minimally, that require isolation to understand how manipulations to social position may produce health outcomes. Consider, for example, the complex and presumably counter-intuitive findings from Kessler and colleagues evaluation of the Moving to Opportunity program: residential relocation is ripe with stressors ranging from the simple moving logistics, to all the nuances of acquiring familiarity in your new community and the social stressors of losing contact with some acquaintances while trying to build new ones. This reality highlights a basic confound in the Moving to Opportunity experimental group: they all "benefited from resource allocation" but they were also all reallocated to a different social group and confronted with sources of group conflict. Data linkage initiatives are another high priority research activity stimulated by this review. Definitive studies designed to capture nuanced conceptual features of the actions of class, multiple pathways to child health, an array of developmentally appropriate health outcomes, and the myriad of possible behavioral and biological mediators are typically not feasible. Moreover, even if feasible, collecting these data in a single study would be inefficient because they already exist in several distinct areas such as income tax returns, school records, electronic medical records, personal health information maintained in proprietary applications , county land use records and other sources that will undoubtedly grow as technology expands. Consequently, the current problem is not necessarily the absence of data, but rather the ability to harvest and integrate existing data. Indeed advancements in bioinformatics and expansion of comprehensive health information exchanges provide substantial potential for linking studies of the social world with useful child health outcomes. Of course such tools are not a panacea, recognizing that medical records are not typically collected for research purposes, and that information maintained in health information exchanges is only relevant to users of health services. Nevertheless, recognizing that some health information exchanges like those by major insurers such as Kaiser Permanente boast greater than 90% of covered lives within specific catchment areas, harnessing the strength of these tools for research is necessary. --- Policy Research Research to identify potential competing or conflicting commitments within policies is needed. Policy makers interested in protecting or promoting children's health oftentimes have multiple commitments to balance. Self-evidently, the development of any given policy whose express purpose is to improve child health demonstrates a commitment to child health. However, in the real world riddled with limited resources, competing world views on who is responsible for child health, and what "health" really means, any policy maker will quickly confront other commitments. Possible competing commitments can be pragmatic , moralistic , ideological , or political . While some commitments may compete, others may clearly conflict. Consider, for example, federal policy supporting Head Start. The policy is clearly committed to children living in poverty. Nevertheless competing commitments related to accountability, such as grantees' ability to meet performance standards and requirements for active parent involvement can result in Head Start agencies screening out children who are most in need. Research or critical analyses of underlying conceptual mechanisms targeted in policies intended to benefit children's health is needed. As others have noted , very few policies have been evaluated for their putative effects on child and family health outcomes. One illustration of this need is the conflicting and counter-intuitive adolescent mental and behavioral health findings previously described from the Moving to Opportunity program . The findings that male adolescents receiving the opportunity to move to a better neighborhood experienced poorer mental and behavioral health outcomes than their peers in the control group does not make sense from a "Class as Inadequate Access to Resources" perspective. But, the results make more sense when viewed from a "Class as group affiliation" or "Class as group conflict" perspectives. Similarly, the putative value of various policy approaches to "school choice", whether through voucher systems to private schools or transportation options within district, such policies are predicated on a view of class wherein access to resources is viewed as the key criterion, whereas consequences resulting from disruptions to group affiliation and potential shifts in group conflict are given less attention. Conversely, policy attempts driven more by perspectives of class as group affiliation or attempts to minimize group conflict, such as strategies to build social capital have been presumed to yield greater health benefits than those focused on resource allocation . Researchers should evaluate ongoing and new policy experiments like state's willingness to adopt Medicaid expansion, school choice and the array of social welfare programs to determine whether or how they are affecting children's health . More importantly, such results-informed by analysis plans that capture variation in the distinct modes of action underlying social class-would ensure delivery of clear information to decision makers as to why specific elements of, or entire, policies help or harm children's health. In sum, explaining the ubiquitous finding of social class gradients in child health is complex and the possible causes operate across many levels and domains; yet the posited causal processes producing the gradient need not be seen as a 'black box' of invisible steps. Our conceptual model is certainly a vast over-simplification, but by bringing together the theoretical perspectives and empirical evidence of multiple disciplines it offers one possible view of testable pathways and mechanisms by which social class 'gets under the skin' to affect children's health and development. --- REFERNCES Alvarado SE. Neighborhood disadvantage and obesity across childhood and adolescence: Evidence from the NLSY children and young adults cohort . Soc Sci Res. 2016;57:80-98. DOI: 10.1016/j.ssresearch.2016.01.008 [PubMed: 26973033]

Social class gradients in children's health and development are ubiquitous across time and geography. The authors develop a conceptual framework relating three actions of class-material allocation, salient group identity, and inter-group conflict-to the reproduction of class-based disparities in child health. A core proposition is that the actions of class stratification create variation in children's mesosystems and microsystems in distinct locations in the ecology of everyday life. Variation in mesosystems (e.g., health care, neighborhoods) and microsystems (e.g., family structure, housing) become manifest in a wide variety of specific experiences and environments that produce the behavioral and biological antecedents to health and disease among children. The framework is explored via a review of theoretical and empirical contributions from multiple disciplines and high-priority areas for future research are highlighted.

Emerging adulthood is marked by a changing social context, with emerging adults decreasing their reliance on parents and beginning the transitional process of forming long-term romantic relationships . This developmental period is also seen as a high-risk time for type 1 diabetes management, as evidenced by high A1C, low self-care, high diabetes distress, and loss of access to health care and insurance , in part as a result of changing relationships . Although it is well established that parental monitoring and involvement in diabetes self-care is an essential component of optimal diabetes management during adolescence , sources of social involvement that facilitate diabetes management during emerging adulthood are only beginning to be understood. Recent research suggests that continued parental involvement remains beneficial even into emerging adulthood . Peer and romantic relationships evolve across the emerging adulthood period and become important sources of social support for diabetes management as they develop , yet little is known about precisely how peers and romantic partners are involved in diabetes management. Despite a vast literature regarding the benefit of social support in individuals with diabetes , findings have been mixed regarding peer support and its associations with diabetes management, with some evidence suggesting that support from romantic partners may be more important than peer support for type 1 diabetes management and psychological well-being . Romantic relationships during emerging adulthood vary from casual dating relationships to more committed relationships of differing lengths , with differences in these relationships potentially related to how partners are involved in diabetes management. Early adolescent romantic relationships are characterized by support, negative interactions, and turbulence . Relationships that are longer and more committed are associated with more positive support and a decline in negative interactions and thus may be more amenable to the type of involvement needed for type 1 diabetes management. As emerging adults develop long-term romantic relationships and reorganize their social support system, they may increasingly seek involvement from their romantic partner rather than from parents or peers. This prediction is consistent with what is known about adults in long-term relationships, as spouses are the most frequently cited relationship for type 1 diabetes support among adults . It is also possible that emerging adults use romantic relationships differently from relationships with parents or peers as sources of support . For example, emerging adults may rely on parents for involvement in instrumental tasks such as navigating insurance. However, they may turn to romantic partners for emotional support as they deal with diabetes distress and changing social contexts. It has been suggested that, for adolescents, parents may provide instrumental support, whereas peers and romantic partners may provide more emotional support . However, little is known about how emerging adults view helpful and unhelpful involvement, especially given the changing relationships with parents and the emergence of romantic relationships during this developmental period. Accessing emerging adults' views of the involvement of their parents and romantic partners in type 1 diabetes management will assist in understanding what roles they view as most helpful from these relationships. Because social involvement in diabetes management is not all positive, we also sought to understand how the involvement of parents and romantic partners may be perceived as not only positive, but also negative. Parents are sometimes perceived as involved via miscarried helping or in ways that are perceived as controlling . Similarly, the involvement of long-term romantic partners can also be perceived as negative in terms of the control they exert and their persuasive strategies . In fact, social support has frequently been described as a double-edged sword, with instrumental and emotional support perceived as both helpful and controlling or nagging . Among adolescents and young adults with type 1 diabetes, peers, coworkers, and romantic partners can be perceived as both facilitating and getting in the way of diabetes management . Furthermore, as the social context changes throughout this developmental period, emerging adults may face the additional challenge of learning to let others know about their support and self-management needs. Because emerging adulthood represents a unique developmental period with a changing social context, understanding who is involved and how these individuals are perceived by emerging adults is crucial to providing the best diabetes care for this population. This study sought to examine emerging adults' perceptions of social involvement in type 1 diabetes management through a qualitative research approach. We sought to understand from emerging adults themselves the form and extent to which their parents and romantic partners were involved in their diabetes management. We examined who was viewed as having the biggest impact on their diabetes management , what these individuals did that was helpful or unhelpful, and what information emerging adults with type 1 diabetes shared with those individuals. We expected that parents and romantic partners would be heavily involved in diabetes self-management, but that this involvement would depend on whether emerging adults were in a committed romantic relationship. We anticipated that emerging adults in a committed romantic relationship would perceive their romantic partner as being primarily involved in their diabetes management, as opposed to their parents. We also explored through qualitative coding what emerging adults found to be helpful and unhelpful about these individuals' involvement, thus providing information on whether parents and romantic partners may play different roles in supporting type 1 diabetes management at this transitional time . Second, we assessed via a survey how parents, romantic partners, friends, and others were involved in various aspects of instrumental and emotional support surrounding type 1 diabetes management. We explored whether the breakdown of parents' and romantic partners' involvement in the survey mirrored findings from the qualitative data. --- Research Design and Methods --- Procedure Participants were drawn from a sample that participated in a 4-year longitudinal study of late adolescents and emerging adults with type 1 diabetes . Individuals who had completed participation in the longitudinal study were invited to participate in the current study. At the time of recruitment for the current study, participants were 22-24 years of age and facing a period of multiple transitions, as emerging adults tend to move out of the parental home, attend college, and begin careers. Individuals were randomly selected to obtain a 1:1 ratio of males to females and a 4:1 ratio of Whites to racial/ethnic minorities. Of the 60 individuals identified for recruitment, 21 were not reached, 1 declined participation because of time constraints, and 38 agreed to participate; after agreeing to participate, 7 did not complete the phone interview and 2 were excluded from analyses due to technical errors. The final sample included 29 emerging adults with type 1 diabetes . The emerging adults provided consent to participate. They were then sent a link via e-mail to complete the pre-interview survey via a secure online platform, after which they completed a phone interview lasting between 9 and 40 minutes. Participants were compensated with either a gift card or a check for $25. Trained research assistants conducted semi-structured interviews, which were audio-recorded and transcribed. All study procedures were approved by the Institutional Review Boards of the University of Utah and University of California, Merced. --- Measures --- Relationship Status Participants responded to a single-item measure of romantic relationship status in the online survey. Response options were 1) no current romantic relationship; 2) dating, but not in a committed relationship; 3) dating for a period of time, but no long-term commitment; 4) in a committed, long-term relationship; 5) in a committed, long-term marital relationship; 6) other; and 7) decline to answer. Given prior literature noting the importance of long-term committed relationships versus more casual relationships, we collapsed these categories into a single dichotomous variable . Responses indicating committed, long-term marital and nonmarital relationships were coded as being in a long-term committed relationship. --- Demographics Participants completed a brief demographics survey, providing information on their residential status , who lived in their primary residence, their romantic relationship status, their education status , and their current occupational status . Additional demographic information was taken from the prior study's records, including participants' sex, age, time since diagnosis, and race/ethnicity. --- Semi-Structured Qualitative Interview Each participant was asked, "Overall, who is the person who seems to have the biggest positive effect on your diabetes, other than yourself?" For each person identified as positive, interviewers asked, "How does this person affect your diabetes?" To understand the frequency of positive involvement, we asked, "About how frequently would you say they affect your diabetes?" To understand whether emerging adults solicit support from their available social systems, we asked, "What kind of information do you tell them about your diabetes?" These questions were then repeated to understand who had the biggest negative impact on diabetes management. Of note, the language used throughout this article is consistent with the wording of interview questions and is not meant to imply causality. A coding team developed a qualitative coding system using an iterative process to identify broad themes related to social involvement in type 1 diabetes management using a subset of interviews . Although coders were knowledgeable about the extant literature on this developmental period, no a priori codes were used, as participant responses informed all codes. An initial system was used to code additional interviews until no new themes emerged and the codes were well defined . Through this iterative process, a final coding system was developed and was used to double-code all interviews, with an option of "other" being used to identify any emerging themes. Coding discrepancies and emerging themes were resolved via discussions with the full coding team until consensus was reached. All interviews were coded using NVivo, v. 12.1, qualitative coding software. To examine who had the biggest positive and negative impact on type 1 diabetes management and how those individuals were involved, we identified interview codes to qualitatively describe emerging adults' perspectives of who had the greatest impact on their diabetes management, as well as their type of involvement and whether this involvement differed based on relationship status. Second, we examined whether emerging adults' perceptions of social involvement in the survey supported findings from the qualitative interviews. --- Social Involvement Participants completed a nine-item survey modified to assess helpful or hindering people in the emerging adults' social network . The original WHOTO survey measure assessed to whom individuals turn when in distress or when trying something new and with whom they prefer to stay in close proximity . Given the focus of the current study on social involvement in type 1 diabetes management, the seven original items were modified to reflect diabetes-specific aspects of social involvement when individuals are feeling diabetes distress or are in new or unknown situations related to diabetes. For exploratory purposes, two additional items were created to identify social network members whose behavior was especially problematic for diabetes management. These items pertaining to disease management are shown in Table 1. Participants selected from the following options to report who was involved in their diabetes management: parents, friends, romantic partner, sibling, nobody, or other. --- Results --- Sample Demographics Of the emerging adult participants, 55% were female, 66% were Caucasian, and the mean age was 22.96 years . Fifty-five percent of our sample were working full-time jobs, and 37.9% were working part-time jobs. Thirteen were identified as not in a long-term relationship , whereas 16 were in a longterm, committed relationship . The majority of our sample was not living in the parental home, with nine living in the parental home, six in an apartment or house without roommates, 12 in a shared apartment or house, and two in a college dormitory. Table 2 Participants commonly reported that either their parent or romantic partner had the biggest positive impact on their diabetes management. When emerging adults did not describe a parent or a romantic partner, they described a range of individuals as having the biggest positive impact: health care provider , roommate n = 2), grandparent , coworker , or sibling . Because some participants mentioned multiple people, these numbers may exceed the sample size of 29. There were no significant differences in who was described as having the biggest positive impact depending on the sex of the emerging adult. Most participants reported that individuals who had the biggest positive impact on diabetes management did so on a daily or weekly basis, with some involved on a less than weekly basis . Themes related to positive social involvement included reminders and instrumental and emotional support. Table 3 provides examples. Emerging adults frequently stated that no one other than themselves had the biggest negative impact and occasionally mentioned general acquaintances and coworkers as having the biggest negative impact. Instances of individuals having the biggest negative impact on diabetes management were described as occurring fairly infrequently, with these individuals involved on a weekly or less than weekly basis. Themes related to negative social involvement included social context barriers and uninformed or misguided behaviors of others . See Tables 1 and3 for additional details regarding qualitative themes. To examine whether parent or partner involvement differed based on emerging adult relationship status, we descriptively compared qualitative interviews of emerging adults who were or were not in long-term relationships. When emerging adults were in a committed relationship, they primarily discussed their romantic partner as having the biggest positive impact . In contrast, emerging adults who were not in a romantic relationship or were casually dating discussed their parents as having the biggest positive impact on their diabetes management . Interestingly, even emerging adults who were casually dating but not in a committed relationship focused on parents as a primary source of help, focusing on parents' long history and knowledge of type 1 diabetes management. For example, one women said, "My parents, because they've known about it the longest and kind of know what to do and when to do it. They've [parents] been to the doctor with me in the past, so they know what kinds of questions to ask because they know exactly what it's like" . Participants also said that remaining on their parents' health plan positively affected their type 1 diabetes management. As described in Table 1, common themes emerged regarding the ways in which parents and romantic partners positively affected type 1 diabetes management, including by providing reminders , emotional support , and instrumental support . Parents and romantic partners were described as providing similar types of positive behaviors on a daily or weekly basis. Specifically, emerging adults discussed being able to vent about diabetes challenges to both parents and romantic partners and said that both parents and partners helped to keep them on track with diabetes management. For example, an emerging adult who was not in a romantic relationship and one who was married both reported that their parent and romantic partner, respectively, helped administer insulin when needed and would help count carbohydrates for meals and snacks . Emerging adults who discussed parents as positively affecting diabetes management also often cited their shared history with type 1 diabetes management as the primary reason. For example, one woman noted that, "she's [mother] kind of been through the learning process with me" , which did not come up in interviews in which romantic partners were described as having the biggest positive impact. Table 1 includes other representative quotes. Regarding the kinds of diabetes health information that emerging adults shared with parents and romantic partners, emerging adults frequently provided diabetes health information to those who had the biggest positive impact on their diabetes management. Specifically, emerging adults described sharing blood glucose levels, recent A1C values, and upcoming supply or prescription needs with the individuals who had the biggest positive impact on their diabetes management. Few emerging adults explicitly reported seeking support from these individuals, but acknowledged that asking for such support could be helpful. For example, one participant described an instance in which her pump malfunctioned on the way to dinner, but her friends were in a rush, so she was unable to change her pump site. She explained that explicitly soliciting support may have ameliorated this situation . Regardless of relationship status, the individuals identified as having the biggest negative impact on diabetes management were primarily individuals other than parents and romantic partners. Representative quotes can be found in Table 2. Specifically, friends and coworkers were identified as negatively affecting diabetes management, and the specific type of behavior that was perceived as having the biggest negative impact was uninformed/misguided behaviors. Behaviors were often unintentional, including individuals not understanding the difference between type 1 and type 2 diabetes and social context barriers such as social situations like taking an exam in college or eating out that make it challenging to take care of their diabetes. Participants mentioned that they shared more general information with individuals negatively affecting their diabetes, such as whether their blood glucose was low or high, rather than specific numbers. Furthermore, emerging adults described having to provide a lot of diabetes education to these people. One woman said, "I become a health professional" to explain what diabetes is and how she must take care of it. Others mentioned even trying to limit the information they share with these individuals, with one emerging adult stating, "When she asks me a question about it, like, I give her a one-. . . [or] two-word answer, just to kind of get her to stop the conversation" . romantic partners as sources of both positive and negative influence on diabetes management . Specifically, emerging adults primarily reported that they counted on parents and romantic partners to be available to help with diabetes, wanted parents and romantic partners to be around in new or unknown diabetes situations, and perceived parents and partners as supporting them in reaching diabetes goals. Friends were cited as the most negative, such that it was most difficult to manage diabetes in the presence of friends. No emerging adults selected romantic partners in response to the survey question "In whose presence is it most difficult to manage your diabetes," regardless of relationship status. Consistent with the qualitative interview, emerging adults in a committed romantic relationship counted on their romantic partner most for diabetes help, indicated that romantic partners were likely to try to do everything for their diabetes management and that partners supported them in pursuing diabetes goals, and indicated preferring to turn to romantic partners for comfort when feeling down about their diabetes. Emerging adults who were single or only casually dating counted on their parents most for diabetes help, reported that their parents were likely to try to do everything for their diabetes and that their parents supported them in pursuing diabetes goals, and preferred their parents to comfort them when feeling down about their diabetes. Interestingly, emerging adults often identified parents as sources of nagging and criticism regardless of relationship status. --- Discussion Parents and romantic partners are an important part of the social context during emerging adulthood and are seen as both positively and negatively involved in diabetes self-management during this developmental period. Both interview and survey results indicated that parents and romantic partners provide multiple supportive functions, as opposed to one individual providing solely emotional support or instrumental support. Because parents and partners were involved in diabetes management in similar ways, the key difference at this developmental stage is perhaps who is the primary support, rather than what that person is doing. This similarity may exist because individuals with type 1 diabetes may benefit from both instrumental and emotional support-both types of support that one would get from one's primary support person. Although emerging adults identified parents and romantic partners as the most trusted social resources for type 1 diabetes, these same relationships were the most common sources of criticism and thus were perceived as making diabetes management more difficult, suggesting that when others are frequently involved, they often slip into nagging and criticizing. Emerging adults may benefit from assistance in managing the harmful aspects of social involvement and in soliciting more helpful forms of social involvement, such as navigating the fine line between reminding and nagging. Romantic relationship status emerged as an important indicator of social involvement during emerging adulthood. Emerging adults who were not in a romantic relationship often cited their parents as most involved in their type 1 diabetes management, whereas those with a committed romantic partner often cited their partner. Although future longitudinal research will be necessary to fully understand how this social context develops from adolescence into young adulthood, such findings raise the possibility that parental involvement is maintained until emerging adults develop a committed, long-term romantic relationship. If so, relationship status may serve as an important indicator to health care providers of who is likely to be involved in type 1 diabetes management tasks. In particular, because romantic partners may not have the same shared history as parents who were likely involved since diagnosis, partners may benefit from diabetes education to enhance their diabetes knowledge and promote more of the positive aspects of involvement in diabetes management tasks. When asked about the type of information emerging adults share with those who have a positive impact on their diabetes management, emerging adults rarely reported that they explicitly solicited support from these individuals. This finding may suggest that emerging adults view disclosure of diabetes health information as a form of seeking out support but do not view this as explicit solicitation of support. For example, if emerging adults disclose information to people in their social network, they may receive more support for diabetes management . In addition, although those in a romantic relationship reported that they most prefer to turn to a romantic partner for comfort related to diabetes , emerging adults rarely discussed romantic partners as providing emotional support for their diabetes management. It is possible that this finding is the result of the unstructured, organic nature of the interview or that emerging adults do not perceive partners' overt acts of emotional support, but rather view them as a person to turn to when things go wrong. Given that parents are still among the primary people involved in emerging adults' type 1 diabetes management, maintaining their involvement across the transition to emerging adulthood may be important . Such involvement may also be beneficial as emerging adults transition from a pediatric to an adult care clinic. The findings that both parents and romantic partners were seen as involved in positive as well as negative ways is consistent with research across adolescence that parents can be both supportive and controlling and that romantic partners can both be supportive and cause conflict . Somewhat surprisingly, friends were infrequently cited as having the most negative or most positive impact on type 1 diabetes management. When mentioned, however, friends were frequently described in more negative terms, although, as noted, these instances were often identified as unintentional. This finding is consistent with the existing literature on peer involvement in type 1 diabetes . The fact that family and friends can be involved in both helpful and harmful ways is also found among adults with type 2 diabetes . Emerging adults may benefit from assistance with optimizing these relationships, as both negative and positive aspects of relationships have been associated with poorer self-care and A1C over time . Interventions such as the FAMS program, which was developed for those with type 2 diabetes, may be beneficial in increasing the positive and decreasing the negative aspects of social involvement . Through phone-based coaching, adults in this program learn skills to engage in assertive communication to ask for support or reduce negative involvement and enhance collaborative problem-solving with others. The current study has limitations that must be considered. First, the sample was likely highly motivated, given that they remained in a 4-year longitudinal study and agreed to participate in future research. Combined with the small sample size and restricted age range , it is possible that the results of this sample may not fully transfer to the full age range of emerging adults. Thus, future research to replicate these findings with a larger sample is warranted. In particular, exploration of the nuances among subsets of participants in the current study were limited, and future research with larger samples may provide fruitful information regarding differences in perceived social involvement among groups . In addition, we were unable to determine whether emerging adults ask for help from others, and if not, why not and how others know to be involved. Although this study serves as a starting point from which the field can continue to examine these questions, future research with a larger sample should include examination of how emerging adults seek out social support. In addition, although interviewers instructed participants to provide open and honest answers to interview questions, it is possible that the reason many emerging adults cited "no one" as having the biggest negative impact on their diabetes management was because they were unwilling to say negative things about others or because they were highly independent in their type 1 diabetes self-management. Finally, we did not assess emerging adults' insurance status, which limits the generalizability of our findings to emerging adults who may lose access to care or supplies during this developmental period. Future research should incorporate this crucial aspect of the high-risk emerging adulthood period to better understand the social support systems that are either in place or needed for emerging adults. In summary, it is clear that romantic relationship status has important implications for who is involved in type 1 diabetes management during this time and that parents likely remain involved in diabetes management into emerging adulthood. Providers working with this population should consider assessing social involvement in individuals' diabetes management, with particular attention to whether emerging adults perceive others as positively and/or negatively affecting their diabetes management. Furthermore, romantic partners who are newly on the scene for diabetes management may benefit from additional diabetes education to provide optimal support. Because it is possible that romantic partners may engage in more frequent negative involvement with increased diabetes knowledge , education efforts for romantic partners should include content targeting ways to optimally support their partners with type 1 diabetes without overinvolvement or control. Likewise, because the same individuals may provide helpful and unhelpful involvement, emerging adults may benefit from assistance in managing their social networks to optimize support for type 1 diabetes tasks. ---

During the emerging adulthood of people with type 1 diabetes, long-term romantic partners may be involved in diabetes management in ways that supplant parental involvement. We examined the perspectives of involvement in diabetes management of the parents and romantic partners of 29 emerging adults with type 1 diabetes, using qualitative interviews and an online survey. When the individuals with diabetes were in long-term romantic relationships, their partners were heavily involved in managing diabetes and providing support; however, when the individuals with diabetes were in short-term relationships or not in a relationship, their parents were described as having the biggest positive impact on their diabetes management. Emerging adults described the involvement of their parents and romantic partners in both positive and negative ways. Romantic relationship status is an important but understudied variable in understanding social involvement and its effects on type 1 diabetes management during emerging adulthood.

Introduction Information and communication technologies are deeply rooted, widely accepted, and developed in today's society [1]. The field of education, in particular, has not been left out of this technological projection, so it was nurtured, and it benefited from the continuous advances produced [2]. The inclusion and development of educational technology has allowed the emergence and evolution of various resources that serve to enhance and improve teaching and learning processes [3]. In this sense, new technological means have made it possible to generate new learning environments and methodologies [4,5]. In addition, they have promoted the development of various digital and innovative resources that complement and stimulate the training action carried out among educational agents, assuming new roles and functions of both the teacher and the students [6,7]. As a consequence of technological development in the field of education, the digital competence of teachers and students is presented as a relevant factor for the development of good practices with ICTs [8][9][10]. This competence is of great relevance for the effective realization of an instructional process in a digital plane, far from the classroom, as is the case of e-learning [11]. On the other hand, recent studies verify that current teachers do not have the skills and abilities necessary to carry out teaching and learning processes in digital environments [12][13][14], which makes it difficult to carry out digital teaching [15]. E-learning can be found within the broad spectrum of innovative learning methodologies. This training approach is conceived as a virtual instruction, carried out through content management platforms to carry out the teaching and learning process in a digital environment [16]. This training model emerged in 1996, creating consortia between universities to offer a digital educational alternative and, since then, this type of learning has proliferated in recent years [17], especially today, as a result of the impact of COVID-19 on the educational field [18]. E-learning, understood as a distance teaching method, allows the possibility of deploying a training plan of both a synchronous and asynchronous nature [19]. There are several advantages that e-learning has over other, more traditional teaching methods [20]. E-learning is understood as an instructional process focused on the student. In this case, the teacher's role focuses on guiding student learning in digital environments [21]. In addition, the availability of the teaching materials and resources offered on the learning platforms allow students to deploy their training anywhere, and at any time, even cooperatively [22]. It is considered a type of adaptive learning, since it adapts to the particularities and learning rhythms of students. In this sense, this mode of instruction gives flexibility to the teaching and learning process [23]. This type of teaching allows greater communicative interaction between educational agents, by being able to emit messages through the platforms established by educational centers [24]. In addition to the advantages it presents for students, for the teaching community it also reports a series of potentialities. Teachers have specified in a digital space all training and resources [25]. They have all the tasks and information of the students [26]. Likewise, teachers can establish fluid channels of communication through private messages, discussion forums and even videoconferences [27]. Despite the potentialities reflected, e-learning also has limitations or disadvantages [28]. This innovative approach reduces physical contact between people [29]. It needs a set of technological requirements for its effective development [30]. In addition, this training modality requires certain digital skills [31]. Moreover, e-learning requires that all of the teacher's teaching materials be converted to digital format in order to be able to use them through the established platform [32]. Likewise, it requires great autonomy and responsibility from the students to follow a distance training process [33]. After analyzing the literature on e-learning, it was observed that there are no studies that have analyzed the construct in a comprehensive manner. This is, from a bibliometric perspective, based on the scientific mapping of published studies. This problem, which has arisen in the impact literature, is intended to minimize with this work, presenting the importance and evolution of the concept of e-learning, specifically in higher education spaces. This research also offers the themes that e-learning runs through, as well as the way forward in coming years, revealing trends that are on the rise. --- Materials and Methods Due to the importance of the subject today, the purpose of this work is to know the evolution of e-learning in the higher education stage in the scientific documents registered in the Web of Science database. In this research, the concepts "e-learning" and "higher education" in WoS are studied. The analysis of this work was developed through scientific mapping [34][35][36][37]. The purpose of this research is to reveal the evolution of ELHI through the analysis of WoS export documents. In addition, it focuses on knowing how scientific production performs and evolves, as well as specifying the most influential topics and authors. The research methodology used is bibliometrics. Bibliometrics allows to analyze different bibliographic characteristics in documents and the analysis of diverse features of scientific activity. This kind of analysis is important to scientific publications as a tool that allows qualifying the quality of the knowledge-generating process and the impact of this process on the scientific background [38]. Specifically, a co-word analysis has been developed [39]. In the same way, various indicators and indices frequently used in this type of study have been analyzed, such as the h, g, hg, and q 2 index [40]. This is to design maps that contain nodes that represent the subdomains of the concepts linked to ELHI. All of this will allow the study of its thematic evolution. The research procedure has been carried out in several stages: selection of the database to be analyzed , considered the most important scientific database; choice of keywords under study ; construction of the search equation AND "higher education" ). Several inclusion criteria were also applied: year of publication ; language ; knowledge area ; type of document ; institutions ; authors ; source of origin ; countries ; citation . These actions reported 1305 publications. The cut-off values for the criteria are established with those results whose nodes are formed by more than one document or value. Publications indexed in 2020 and repeated or poorly registered documents were established as exclusion criteria. Moreover, 1261 publications conformed final analysis matrix. This figure was classified into three-time intervals that have been prepared under the criterion of documentary similarity. All presented analysis was carried out with the citation reports tool of WOS and SciMAT software. --- Results --- Diachronic Analysis and Scientific Production The set of documents related to ELHI in Web of Science is 1261. The start of production dates back to the year 2000. From that date until 2010, the growth in the volume of documents was high . In 2011 the scientific production decreased compared to the previous year, but in 2012 and 2013, it increased again. From that same year until 2017, the production was constant . However, in 2018 and 2019, production fell to levels close to 2008. Peak production occurred in 2015 . developed [39]. In the same way, various indicators and indices frequently used in this type of study have been analyzed, such as the h, g, hg, and q 2 index [40]. This is to design maps that contain nodes that represent the subdomains of the concepts linked to ELHI. All of this will allow the study of its thematic evolution. The research procedure has been carried out in several stages: selection of the database to be analyzed , considered the most important scientific database; choice of keywords under study ; construction of the search equation AND "higher education" ). Several inclusion criteria were also applied: year of publication ; language ; knowledge area ; type of document ; institutions ; authors ; source of origin ; countries ; citation . These actions reported 1305 publications. The cut-off values for the criteria are established with those results whose nodes are formed by more than one document or value. Publications indexed in 2020 and repeated or poorly registered documents were established as exclusion criteria. Moreover, 1261 publications conformed final analysis matrix. This figure was classified into three-time intervals that have been prepared under the criterion of documentary similarity. All presented analysis was carried out with the citation reports tool of WOS and SciMAT software. --- Results --- Diachronic Analysis and Scientific Production The set of documents related to ELHI in Web of Science is 1261. The start of production dates back to the year 2000. From that date until 2010, the growth in the volume of documents was high . In 2011 the scientific production decreased compared to the previous year, but in 2012 and 2013, it increased again. From that same year until 2017, the production was constant . However, in 2018 and 2019, production fell to levels close to 2008. Peak production occurred in 2015 . The language, mainly used by researchers, is English. All other languages are far removed from it . The studies on ELHI are collected mainly in the knowledge area Education Educational Research, being far from the other areas . Both papers and research articles are the main types of documents used by researchers to show their findings. The number of papers stand out, confirming the fact that this field of study is of a recent appearance in the scientific community . The Universitat Oberta de Catalunya is the leading institution in ELHI-related studies. The fact that the most prolific universities are Spanish stands out . The author with the highest level of scientific production on ELHI is Pelet, J.E. The rest of the authors are at a close distance, which shows the variety and number of authors who study the field under analysis . The main sources collecting research on ELHI are two books of conference proceedings, with EDULEARN as the main reference. This fact coincides with the data offered previously on the source of production. In this case, the main journal that publishes studies on this subject is Procedia Social and Behavioral Science . Two countries are the reference countries in this field of study in scientific production. Spain is in first place, closely followed by England . In ELHI's field of study, the most widely cited work has 234 citations [41]. The second work has 205 citations [42] . --- Reference Citation [41] 234 [42] 205 [43] 133 [44] 114 --- Development of the Structure and Thematic Axis of ELHI The evolution of keywords shows the number of keywords collected in a given period. In addition, it also shows the keywords that go out and come in a certain period. In this study, the level of keyword matching between contiguous periods is low, and it increased from its beginnings until 2019. This shows that the line of investigation is not established, although it is beginning to be established . The main sources collecting research on ELHI are two books of conference proceedings, with EDULEARN as the main reference. This fact coincides with the data offered previously on the source of production. In this case, the main journal that publishes studies on this subject is Procedia Social and Behavioral Science . Two countries are the reference countries in this field of study in scientific production. Spain is in first place, closely followed by England . In ELHI's field of study, the most widely cited work has 234 citations [41]. The second work has 205 citations [42] . --- Reference Citation [41] 234 [42] 205 [43] 133 [44] 114 --- Development of the Structure and Thematic Axis of ELHI The evolution of keywords shows the number of keywords collected in a given period. In addition, it also shows the keywords that go out and come in a certain period. In this study, the level of keyword matching between contiguous periods is low, and it increased from its beginnings until 2019. This shows that the line of investigation is not established, although it is beginning to be established . The academic performance presents the most valuable subjects in each of the established time periods. These values are based on bibliometric indicators of various kinds . Academic performance is shown according to established time periods. In the first interval , the subject with the highest bibliometric value is "education". In the second interval , "e-learning" is the most important. In the third interval , it is "extension" and "students" . Thematic diagrams provide data on the relevance of the different topics in a given time period. The coordinate axis shows the level of centrality and density. The former analyzes the relational strength of external links. The second analyzes the relational strength of internal links. The study of the three diagrams shows that no single theme is repeated as the driving force in the three periods. Only "satisfaction" stands out, which appears in the second and third periods . Thematic diagrams provide data on the relevance of the different topics in a given time period. The coordinate axis shows the level of centrality and density. The former analyzes the relational strength of external links. The second analyzes the relationa strength of internal links. The study of the three diagrams shows that no single theme is repeated as the driving force in the three periods. Only "satisfaction" stands out, which appears in the second and third periods . In the first interval , the driving themes are "perceived-usefulness", which is related to "impact", "technology-acceptance-model", and "user-acceptance"; "e-readiness", which is based on "measuring-e-readiness", "cultural-aspects", and "IT"; and "on-line" which is oriented towards "qualitative-differences", "knowledge-work", "teaching-quality", "technology", "satisfaction", "conceptions", "teaching/learning-strategies", and "face-toface". In this period, although ELHI is analyzed, it seems that the research interest is more oriented towards distance learning than e-learning. In the second interval , the driving themes are "used-acceptance", which is related to "perceived-ease", "extension", "national-culture", "information-technology", "model", "behavioral-intention", "self-efficacy" and "technology-acceptance"; and "satisfaction", which focuses on "information-systems-success", "continuance-intention", "instructor", "critical-success-factors", "web", "management", and "web-based-e-learningsystem". In this period, the evaluation and acceptance of the e-learning method, mainly oriented to technological resources, takes on an important nuance. In the third and final interval , the driving themes are "DeLone" which is related to "McLean-model", "information-systems-success", "attitudes", "e-learning-system", and "management-systems"; "extension", which is oriented towards "unified-theory", "determinants", "students-behavioral-intention", "information-technology", "technology-acceptancemodel", "user-acceptance", "usage" and "variables"; "intention" which is based on "acceptance", "information-systems-continuance", "trust", "computer", "model", "adoption", "self-efficacy" and "perceived-ease"; "satisfaction", which relates to "e-learning-success", "students-perspective", "public-administration-education", "quality-of-e-learning-system", "critical-success-factors", "performance", "context" and "use"; and "behavioral-intention" which focuses on "technology-acceptance", "TAM", "system-characteristics", "perspective", "preservice-teachers", "structural-equation-modeling", "university-students", and "perceivedusefulness". In this period, the e-learning method is more relevant, the behavior of the subjects when faced with the pedagogical method and the relationship between the people involved in the pedagogical act. Furthermore, in this period, the themes "MOOC", "gamification", "blended-learning", and "Moodle" must be taken into account, given that their location in the diagram places them as unknown themes. In other words, the topics may become the next driving force, or they may disappear from the scientists' lines of research. --- Scientific Evolution of ELHI The scientific evolution provides information on the relationship between the themes of the recognized contiguous periods. To establish these relationships, the Jaccard index has been taken into account. The relationships can be of two types: continuous line and discontinuous line. The thickness of the lines shows the strength of the relationship between the themes. The scientific evolution of ELHI presents a conceptual gap. This is due to the fact that there is not one theme that is repeated in the three periods. Even so, two lines of research can be glimpsed, on the one hand that established by "model-user/acceptanceextension/behavioral-intention" and-on the other-that of "students-e/learning-students". In other words, studies in this field of study are mainly oriented towards the acceptance of the pedagogical model and in the students themselves. Furthermore, as can be seen in Figure 4, there is a shift from elements of distance learning to aspects of e-learning. It can also be seen how terms such as social networks, learning for life, and learning platforms appear, which shows the variety of resources and the purpose of this type of teaching method. It should be borne in mind that there are more conceptual than non-conceptual connections. --- Most Relevant Authors The author considered as the driving force in the ELHI study is Gullu, F. , because of his position in the diagram. purpose of this type of teaching method. It should be borne in mind that there are m conceptual than non-conceptual connections. --- Most Relevant Authors The author considered as the driving force in the ELHI study is Gullu, F. --- Discussion The ELHI field of study is relatively young, as it began in 2000 according to WoS. From that date to 2019, the level of production was irregular, with a great deal of incidence by the scientific community from 2000 to 2010, and then remaining constant over time. Since 2018, there was a downward trend in production, although with the health crisis caused by the COVID-19, it will probably rise again, given that it is currently the model used to develop all types of education, both university and non-university levels. The studies on ELHI are written in English and presented in conference papers. The area of knowledge where research is based on is Education Educational Research. The most important institution in this field of study is the Universitat Oberta de Catalunya , with Spanish institutions being the most productive. Pelet, J.E. is the author with the largest number of productions, although Gullu, F. stands out as the most relevant in this field of study. The main source of publication for the conferences is EDULEARN proceedings, while the journal source is Procedia Social and Behavioral Sciences. Spain has the largest scientific production on ELHI. The most cited manuscript is [41], with 234 citations. If we look at the co-word analysis, we can see that this field of study does not have a solid line of investigation recognized, given that the level of coincidence of key words between the established periods is low. Even so, it can be seen that a line of research is beginning to be established in this field of study, given that, over time, the number of key words coinciding between periods is increasing. Academic performance reiterates and confirms this fact, given that there is not one theme that repeats in all three periods, not even in two. This analysis shows that the topics dealt with in this field of research are oriented towards education, online teaching, and students. With regard to the most relevant topics in this field of study, it is shown that there is an evolution over time, with varying interest and focus on the part of the researchers. In the first period, interest is focused more on distance learning than on online learning it- --- Discussion The ELHI field of study is relatively young, as it began in 2000 according to WoS. From that date to 2019, the level of production was irregular, with a great deal of incidence by the scientific community from 2000 to 2010, and then remaining constant over time. Since 2018, there was a downward trend in production, although with the health crisis caused by the COVID-19, it will probably rise again, given that it is currently the model used to develop all types of education, both university and non-university levels. The studies on ELHI are written in English and presented in conference papers. The area of knowledge where research is based on is Education Educational Research. The most important institution in this field of study is the Universitat Oberta de Catalunya , with Spanish institutions being the most productive. Pelet, J.E. is the author with the largest number of productions, although Gullu, F. stands out as the most relevant in this field of study. The main source of publication for the conferences is EDULEARN proceedings, while the journal source is Procedia Social and Behavioral Sciences. Spain has the largest scientific production on ELHI. The most cited manuscript is [41], with 234 citations. If we look at the co-word analysis, we can see that this field of study does not have a solid line of investigation recognized, given that the level of coincidence of key words between the established periods is low. Even so, it can be seen that a line of research is beginning to be established in this field of study, given that, over time, the number of key words coinciding between periods is increasing. Academic performance reiterates and confirms this fact, given that there is not one theme that repeats in all three periods, not even in two. This analysis shows that the topics dealt with in this field of research are oriented towards education, online teaching, and students. With regard to the most relevant topics in this field of study, it is shown that there is an evolution over time, with varying interest and focus on the part of the researchers. In the first period, interest is focused more on distance learning than on online learning itself, given that it uses terms that are not typical of the e-learning method. In the second period, this fact changes radically, given that interest is focused on the assessment and acceptance of e-learning by those involved in this teaching process. In the third period, the line of the second period is maintained a little, although the researchers extend their field of interest to the intention of those involved in the training process. Over the next few years, the themes of "MOOC", "gamification", "blended-learning" and "Moodle" should be kept in mind when researching the use of e-learning in higher education, and could be the next driving force. The thematic evolution once again confirms what was indicated previously-that is to say, there is no solid and stable line of research in this field. Moreover, there is a conceptual research gap related to ELHI. Even so, there are two lines that have focused more interest on the part of the scientific community. These timelines are "model-user/acceptanceextension/behavioral-intention" and "students-e/learning-students". In other words, studies in this field of study are mainly oriented towards the acceptance of the pedagogical model and in the students themselves. --- Conclusions It can be concluded that research on the use of ELHI does not have an established line of investigation. This bibliometric study indicates that the subject is quite new in the Web of Science database. That is to say, studies on this line of research have begun to be registered recently. The bibliometric analysis specifies that the research is oriented towards knowing the level of acceptance and application of the pedagogical method in the teaching and learning processes. This information offered here can guide researchers and professors in their work. The prospective of this research consists of offering the scientific community new trends in this field of study. In addition, the aim is to show the educational community the latest trends on this method of teaching and learning. As for the limitations of this research, we must say that there are several. Firstly, generating the database in WoS requires a great deal of effort for researchers. This is because the entire scientific production must be read to determine whether it follows the inclusion criteria set out in the PRISMA protocol. Secondly, the time distribution has been carried out, taking into account the criterion of equity. An attempt has been made to collect an even number of manuscripts in each of the established time periods. Thirdly and finally, the criteria for inclusion have been established at the discretion of the researchers, which have sought to present results that are in line with each other in terms of size and relevance. In this case, the data shown in this study should be considered with caution. The change in the inclusion criteria may slightly change the data shown in this research. Future lines of research include analyzing the e-learning method in other educational stages, such as compulsory secondary education, baccalaureate, or vocational training. --- Data Availability Statement: Data is contained within the article. --- Author Contributions: A.S.-R.: conceptualization, formal analysis, investigation, writing-original draft preparation, writing-review and editing, visualization. J.L.-B.: methodology, formal analysis, investigation, writing-original draft preparation, writing-review and editing, visualization, supervision. A.-J.M.-G.: software, formal analysis, investigation, data curation, writing-original draft preparation, writing-review and editing, visualization. M.-E.P.-G.: formal analysis, investigation, writing-original draft preparation, writing-review and editing, visualization. All authors have read and agreed to the published version of the manuscript. ---

E-learning is conceived as a purely virtual training approach. Different learning styles have been proliferated in recent years, especially now, due to the impact of COVID-19 in the educational field. The aim of this study is to discover the evolution of e-learning in higher education (ELHI) in scientific literature indexed on the Web of Science. Co-word analysis and bibliometric analysis was performed. A total matrix of 1261 documents was analyzed through SciMAT software. The results revealed that studies on ELHI are written in English and presented by conference papers. The main source of publication for the conferences is EDULEARN proceedings, while the journal source is Procedia-Social and Behavioral Sciences. Spain is the country with the highest volume of production. It is concluded that research on ELHI use does not have an established line of research, due to its recent creation and the lack of related research. The bibliometric analysis specifies that the research is oriented towards knowing the level of acceptance and application of the pedagogical method in the teaching and learning processes.

INTRODUCTION Coproduction, participatory research, peer research and citizen science are some of the terms used within a large body of emerging concepts and models that advocate public participation in research and require new skills and ways of producing and communicating knowledge. 1 2 The benefits of citizen participation in health research have been increasingly recognised for both the research and the public who participate. Citizen participation in research can give health interventions more relevant objectives, make their content more appropriate to local needs, generate more feasible implementation strategies and lead to more beneficial outcomes. [3][4][5] For those 'researched', it can promote the human right to participate in processes concerning them. 6 7 On the other --- WHAT IS ALREADY KNOWN ON THIS TOPIC ⇒ Coproducing research with young people is gaining recognition in health research and intervention development, with various models proposed. However, there is a lack of understanding of the micro-level aspects of this collaborative approach, particularly in the Global South. --- WHAT THIS STUDY ADDS ⇒ Our autoethnographic analysis revealed an immersive coproduction process characterised by flexibility, trust in the model and a commitment to nurturing the partnership. Care practices, addressing not just technical but also social and personal needs, enhanced synergy, equity and transparency. --- BMJ Global Health hand, this way of doing research may require more skills and resources, including financial resources, relational skills and time. 8 9 In this article, we will adopt the term coproduction, defined as a way of doing research in which 'researchers, practitioners and members of the public work together, sharing power and responsibility from the start to the end of the project'. 10 It is expected to drive more egalitarian, democratic or transparent research processes, which more effectively address the needs of patients, service users and/or marginalised citizens. 11 In a co-production process, different stakeholders bring their skills, life experiences and social roles to the table. Together they develop the various stages of research, such as idea generation, funding acquisition, study design, management, data collection, analysis, evaluation and dissemination. Coproduction is an increasingly popular approach in health research; yet, while there is a growing body of published work in the Global North or high-income countries, there is comparatively little documented in the Global South or low and middle-income countries . 12 When it comes to the coproduction of research with young people, a systematic literature review identified studies conducted exclusively in European countries, the USA and Australia. 13 The available evidence is even more limited when considering research on mental health interventions coproduced with young people; only two studies were found during a systematic review of literature on the codesign of mental health services. 14 Codesign of digital mental health tools with young people seem more frequent, with 25 original articles identified in a recent review, although none of these was from the Global South. 15 In both the Global North and the Global South, the meeting of heterogeneous actors is known to pose a number of challenges, especially those related to power dynamics. [16][17][18] Researchers have emphasised the importance of building and maintaining reciprocal and trusting relationships, sharing power and valuing different skills, knowledge and perspectives, to overcome these challenges. 10 19 Nevertheless, there is a need to better understand the microprocesses that unfold in the context of research coproduction and how teams overcome challenges in practice. This article aims to expand our knowledge about coproduction processes in research on youth mental health, by bringing together the voices of young people and academic researchers. The setting for this study was Brazil, where coproduction methods have been used to drive innovation in public services 20 and public policy design 21 22 but not in the field of youth mental health. In recent years, the country has faced deepening social inequality, children's rights' violations and democratic fragility. 23 Adolescents have been particularly badly affected, with a sevenfold loss of income between 2014 and 2019, surpassing the average loss for traditionally excluded groups-illiterate individuals, those of Black ethnicity and residents of the North and Northeast regions of the country. 24 Their quality of life was further worsened by the COVID-19 pandemic. 25 During the pandemic and even after social distancing restrictions were lifted in Brazil, adolescents in high school were the group that struggled the most with mental health issues. 26 A case that featured on national TV gave an emblematic example of this: twenty-six students in a state school had a simultaneous crying fit, along with shortness of breath and body tremors. The students were rescued by the Mobile Emergency Care Service, which reported a 'collective anxiety crisis', 27 prompting a debate around the need for school-based interventions. In this paper, we, a team of academic researchers and young people, offer an in-depth autoethnographic account of our experience of coproducing a digital capacity-building intervention to support Brazilian young people's participation in promoting good mental health, optimised for use in schools. The intervention was based on digital storytelling and consisted of a 'chat-story': a virtual experience in which a narrative unfolds as users interact with fictional characters on a text-messaging platform through dialogue, audio recordings, videos and memes. The narrative was based on real-life stories of Brazilian adolescents collected during a mapping phase. Developing this tool required the symbiotic integration of multiple kinds of expertise, including academic/ scientific experience, lived experience and technical and creative skills. Our team experienced several challenges during this process, and collectively and iteratively developed solutions to them. The purpose of this article is to reflect on these challenges, realignments and unplanned learnings, providing a rich description of how coproduction might unfold in practice. We hope that our analysis inspires new ways of promoting coproduction with young people, especially in LMICs. --- Autoethnography Researchers participating in a social process that is being studied find in autoethnography a methodological tradition for structuring immersive experiences into shareable knowledge. The main pillar of this approach is the personal biography, an author's personal experience, which serves as the primary source of information. 28 The term autoethnography first appeared in research belonging to the poststructuralist paradigm, where the researcher was part of the group being studied. 29 However, autoethnography has evolved into diverse methodologies, all of which feature the open inclusion of the self, in effect the researcher's own biography, in the investigation of social and cultural processes. 30 31 The authors 'scrutinize, publicize, and reflexively rework their own self-understandings as a way to shape understandings of and in the wider world' . This paper is the result of 'collaborative autoethnography', 32 whereby authors work collectively to observe and collect --- BMJ Global Health personal experiences and interpret the data to understand a particular phenomenon. Participants' identities are revealed, challenging their default position of anonymity traditionally held in social science research. --- METHODOLOGY Patient and public involvement This autoethnography analyses the process of research coproduction, which is a type of patient and public involvement, from the first-person perspective of a team of academic researchers and young people. To better define our aims, we begin by describing our coproduction environment. As recommended by Das and colleagues, 33 we outline three critical elements that characterise coproductive initiatives: ► Who?-comprising the context in which the project was carried out and its actors. ► …did what?-referring to the resulting product. ► How?-referring to the nature of the collaboration and the way it unfolded. --- Co-production context and actors The coproduction took place as part of the Engajadamente Project. 34 The project was the initiative of a Brazilian researcher at the University of Oxford, UK, GP, who invited a researcher at the University of Brasília , Brazil, SGM, and the technology company Talk2U to partner up on a project aimed at developing a chat-story enabling young people to participate in the promotion of good mental health. After securing funding from the British Academy Youth Futures Programme in collaboration with Oxford researcher IS, and establishing an international agreement, GP and SGM selected two postdoctoral researchers to join the team as well as five young people , who took the place of peers in the research. The selection criteria for the postdoctoral students included the following: academic expertise in young people's mental health, intersectionality and/ or social participation; international experience and the ability to work with non-academic partners. The young researchers were undergraduates from the UnB, selected on the basis of their age ; interest in or experiences related to mental health and community engagement; and experience of digital innovation and creative activities. The selection procedures for all researchers were carried out online and included two phases. In the first phase, all applicants submitted resumés and letters of motivation, and the young applicants additionally submitted an original 1 min video aimed at teenagers about mental health and well-being. The shortlisted candidates were invited to an individual interview or a group discussion on mental health and youth participation , where the ability to work collaboratively as part of a team was also assessed. Our core coproduction team, therefore, consisted of four academic researchers and five young people . The academic researchers-henceforth called 'adult researchers'-had backgrounds in psychology, public health, communication and ethics. They were involved in projects promoting good mental health and young people's rights and best interests, with academic careers ranging from 10 to 22 years in these fields. The young people-henceforth called 'young researchers' -were aged between 17 and 20 at the time of recruitment and were studying political science, social sciences or psychology as undergraduates. Both groups included a range of gender identities, sexual orientations and ethnic and socioeconomic backgrounds, including individuals with lived experience of poverty and mental health challenges. All the researchers were Brazilian and native Portuguese speakers. The core coproduction team collaborated closely with Talk2U, a company that pioneered the concept of 'chatstories' and has in its portfolio several such tools, covering issues related to migration, climate change, violence and mental health. 35 Talk2U co-developed the chat-story script and led the technological and audio-visual development. Several other stakeholders, including young people, education professionals, policymakers and professionals in the creative industries, were also recruited, constituting an extended co-production network. --- Co-produced results The main product derived from the coproduction process was a chat-story, a gamified narrative intervention aimed at supporting young people to promote mental health in Brazilian high schools. In particular, the tool was designed to build young people's skills when it came to supporting their peers and to engage in collective action to promote mental health . 36 The codesigned tool, titled Cadê o Kauê? , was produced and disseminated via a social media campaign during the project. Cadê o Kauê? lasts approximately 90 min, during which time users are led to make choices that support the development of skills for peer support or collective action. 37 We also developed a guide for teachers, to enable them to facilitate use of the chat-story in schools and promote relevant discussions regarding mental health. 38 Both the digital intervention and the teachers' guide were based on an understanding of mental health as a result of social determinants 39 and of youth participation as a protective factor for adolescent well-being. 40 Pluralistic coproduction in an extended network We adopted a horizontal way of working, following principles of inclusion, reciprocity and mutual respect. 10 We used a pluralistic model, 33 where young people had a --- BMJ Global Health voice and played an active role, and they and the adults shared control. The young researchers and the adult researchers collaborated on the study design, literature review, data collection and interpretation, design of the digital tool and dissemination of research results. Contributions from the young researchers were largely based on lived experience expertise, and contributions from the adult researchers were mainly grounded on academic/scientific expertise. Both groups had equal ownership of the work, including coauthorship of all scientific articles and conference presentations arising from the project. Authorship order was discussed and jointly agreed on. The audio-visual production for dissemination of findings was mainly the responsibility of the young researchers alongside one adult researcher. Participant onboarding, data analysis, paper writing, development of an educators' guide, management of the advisory groups and organisation of events were mainly the responsibility of the adult researchers. The project management and governance, including contracts, ethics approvals, reports to funders, management of external partnerships, financial management and legal aspects, were largely the responsibility of two adult researchers , representing the University of Oxford and UnB. All team members had access to the full documentation about the project and jointly monitored its progress and timeline. All team members contributed to planning, designing and executing activities; the main outputs and budget expenditures followed the plan outlined in the original proposal, as approved by the British Academy. The work was conducted predominantly online, owing to the pandemic and the physical distance between our respective places of residence . The adult and young researchers held approximately 40 joint online meetings via Zoom. Most of the meetings were held weekly, for reflection and decision-making and lasted 3 hours 40 min excluding breaks. In addition to these meetings, different groups met to carry out the tasks assigned in the weekly meetings or for further discussion; sometimes there was a mix of adult and young members and sometimes the adults and young researchers met separately. Of the total number of meetings recorded, around 60% had both young and adult researchers present. In addition to the remote meetings, approximately once every 3 months, we had in-person meetings in Brasília. Throughout the project, the team also kept an active WhatsApp group for communication between meetings. Meetings with collaborators and partners were held almost exclusively via Zoom, except for the dissemination workshops that closed the project in Brasília. Formal commitment to the project varied from 10 hours a week to full time. However, all team members managed their time flexibly according to project needs and other competing responsibilities. Both the young researchers and the adult researchers were remunerated monthly, according to the guidelines of their respective universities. --- Observational method and recordings Autoethnographic observation and analysis by the core coproduction team took place during the full project cycle . All the team members consented to taking part in the autoethnography and coauthored this piece . The collective autoethnographic process included project and personal records as well as dedicated meetings to reflect on the coproduction process, often held in relaxed environments and using arts-based methods. Challenges and solutions that arose during the coproduction process were recorded and systematised throughout the project, through analysis of emerging material and group discussions led by an adult researcher in collaboration with two young researchers . The team discussed and agreed on the main challenges and solutions during joint meetings, with refinement of key themes conducted by SGM and GP. RRAdS oversaw the final readings and refinement as a young researcher member. Given the personal nature of the research and the impossibility of full anonymity, careful discussions were held about the authors' thoughts and feelings about the inclusion of different practical examples and quotations; all authors reviewed and approved the final content selected for inclusion. --- RESULTS The core coproduction team was formed in September 2021. From there onwards, we participated as peers in an extended network of professionals and partner institutions. Even though the project was originally designed to last 21 months, contextual barriers inherent to the establishment of the international partnership shortened this timeline to 14 months, resulting in substantial time pressure for the team. There was a constant need to coordinate work and adjust expectations in the face of time constraints. The project was carried out in six stages, through which the coproduction team functioned in different ways and combinations. Table 1 summarises the challenges experienced and the corresponding solutions collaboratively agreed on by the team, across each of the project phases. Two videos led by the young researchers, sharing reflections on the co-production process, are provided in online supplemental videos 1 and 2 to illustrate some of our results. The conception of the project and the proposal submission, which preceded Stage 1, took place prior to full team recruitment, and were, therefore, not coproduced. However, youth voices were represented there via informal consultations with members of youth advisory groups worldwide. Below we detail each stage and illustrate it with anonymised quotes extracted from meeting notes and recordings across the project. Quotations are labelled according to respective identity groups . Adult researchers don't know the specific language and codes that are common to young people and their creative process. --- Education and training strategy for youth leadership in focus groups Investment in interpersonal knowledge of team members. The focus group/interview guide is coproduced by youth and adult researchers; youth researchers lead creation of a vignette in video format, with a character that would captivate the attention of the interviewees. Focus groups/interviews are co-delivered by pairs composed of youth and adult researchers. 2-Our differences, our learning community The entry of a new group of media professionals, influencing the coproduction dynamics, brought new learning challenges and more relational differences to assimilate. Innovation Lab promotes knowledge exchange: media professionals involved in the project present the chat-story programming model; researchers present results of the mapping and concept definitions; young researchers translate concepts into artistic and cultural references relevant to Brazilian young people. 3-A secure base from which to connect 'Not everything is young people's responsibility' appears in the results of our mapping study, but the project does not include the voice of adult stakeholders who might support adolescent participation in the setting where the chat-story will be implemented . The incorporation of multiple advisory groups adds complexity to the practice of coproduction. Policy advisors on education and adolescent health suggest creating an advisory committee with teachers and education professionals to anticipate challenges and facilitate the implementation of the intervention in the school context. The School Community Committee and Chat-Story Advisory Committee are formed, managed by adult researchers. Periodic meetings are jointly facilitated by young researchers and adult researchers for specific inputs. Cohesion within the core team provides a secure base to expand and manage increasing complexity, as difficulties are talked through and managed together. 4-Care, transparency and accountability Balancing expectations of all agents involved in the chatstory writing process, including the creative collaborators and needs identified during the mapping stage. The creative professionals express overload from numerous revisions of the chat-story script. Young researchers note limits to the listening sensitivity of peers and partners, work overload, as well as the need to develop greater 'internal transparency' of our shared work processes. Adult researchers invite young researchers to express problems and difficulties they perceive in coproduction and decision-making, as well as pointing to the lack of punctuality, presence, responsiveness in WhatsApp communication, and the fact that cameras are often turned off during meetings. This problem did not have a satisfactory solution for everyone. More than 10 versions of the chat-story were created, with collective revisions and recurrent negotiations regarding characters and structure of the story. The story needed to be revised in the test phase. Meetings to renegotiate deadlines and division of work, and to resolve differences in creative perspectives and schedules. --- Strengthened coproduction structure We tested a routine of sharing the achievements and non-achievements via WhatsApp, with emojis of clinking glasses and pineapples. We also tested group activities-remote or face-to-face-to build common understandings of co-production and our process, using online padlets and cardboard papers, and including leisure activities at waterfalls and in parks in Brasilía. --- BMJ Global Health Stage 1: a level playing field The first stage launched the co-production process between adult and young researchers. Our first goal was to conduct a qualitative mapping of Brazilian young people's perceptions of barriers to and enablers of good mental health as well as strategies they used to enact peer support and promote mental well-being. 41 The results were used to guide the design of the digital tool Cadê o Kauê?. --- Taking collective form The process of working together made differences in levels of knowledge and experience more salient , such as the technical knowledge to conduct focus group discussions or interviews, more present among adult researchers, and the lack of knowledge on the part of the adults of the specific language and communication codes of young people in their creative processes. Added to that was a lack of familiarity with one another, and with the coproduction model, on the part of most group members. This meant that our initial meetings often resembled traditional adult-led environments, with adult researchers more likely to speak, set direction and have cameras turned on. Both adult and young researchers expressed feelings of 'otherness', emphasising ways in which the two groups differed, as illustrated by the following quotations from initial meetings: I feel distant from the reality of what it's like to be an adolescent today . I am a bit scared of talking to the researchers . I don't have a graduate degree, and I have barely started my undergrad studies . In order to deal with these challenges, we adopted two lines of action: one technical and one interpersonal. To address inequalities of knowledge, one of our adult researchers, a specialist in qualitative research, led a training session with the young researchers to help them develop skills for conducting focus groups and thematic data analysis. Although the young researchers did not offer reciprocal formal training to the adult researchers, they supported the latter in tasks they had less experience with, such as the dissemination of recruitment information via Instagram. As for the interpersonal element, we invested time in discussing coproduction, teamwork and the values we wished to enshrine within the project. This was achieved, for instance, through workshops to discuss shared values and teamwork skills. We also focused on sharing knowledge about ourselves as people, so that generational differences were not perceived as barriers but as resources. This process made our similarities as human beings more salient, supporting the creation of a single team identity. This type of interpersonal knowledge is not to be confused with technical skills or what it is necessary to know for efficient production. Rather, these efforts --- Project stage --- Coproduction challenge Care/solution 5-The team as a living organism Tiredness in the collective construction of the chat-story generates the perception that it is better to invest less in everyone's synergy in all tasks due to the lack of time. Adult researchers had more time available for the project than young researchers, as well as more skills for production and dissemination of the generated knowledge via research papers. Young researchers had overall higher video-making skills. The groups had little time to train each other in those skills. Work partitioning: young and adult researchers took turns in leading scientific dissemination and academic conferences. Some decisions were made by representatives of each group rather than full team. The audio-visual production for conferences became a mixed task: some of the recording, scriptwriting and video editing was executed by young researchers; some was executed by an adult researcher and an external company. Instagram content creation, however, was almost exclusive to young researchers. Young and adult researchers were distributed across different research papers, with adult researchers leading the process and sharing tasks. There was also a reorganisation of schedules with extension of deadlines, to make better use of the co-production format. 6-Friendship for knowledge and care The radical participation in each other's lives was clear, with team members disclosing personal vulnerabilities at a level that is not commonly seen in research environments, at least from our previous experience. The openness to surprises, to the unforeseen, and the attention to suffering not necessarily derived from the research, contributed to the emergence of friendship in the group, who took ten hours to say goodbye at the end of the project. One of the first creative exercises we did to get to know each other was to tell the story of our hair. Each adult or young person told their story from the point of view of their hair. Vulnerabilities were felt and challenges were faced by long, short, straight, curly, braided or black power hair. We heard stories of ancestry, power and madness, of love and hatred; we learnt that some of our team's hair had been the victim of prejudice, while that of others not so much. The diversity of our hair and its life trajectories symbolised the diversity present within our coproduction collective. Through this interpersonal process, we noticed the two groups getting closer and more committed, sharing and recognising the value in our different voices; and young researchers building their own 'visibility' in meetings, metaphorically and literally, and in creative ways that were not expected or planned by the adult researchers. For instance, young researchers took the initiative of producing a recruitment video as well as a visual vignette that was incorporated as a discussion starter for the interviews and focus group discussions. Together we cocreated a topic guide and implemented youth-led interviews and focus group discussions, with adult researchers taking a supportive role. Cocreation during the mapping phase ensured that the concepts covered were relevant from a theoretical viewpoint and also spoke to the lived experience of adolescents, and that interviews and focus group discussions were comfortable, engaging and accessible to young participants. --- Challenging assumptions As we worked together, young researchers and adult researchers gradually built a new identity for themselves in the coproduction team, but not without backand-forth movements of self-discovery in the team as a whole . We became more aware of 'presumed responsibilities' within the group, such as 'leading an online meeting' or 'managing social media' , which helped us to address this. Efforts were made to redistribute responsibilities across team members, sometimes following direct requests from team members . In some cases, tasks were managed exclusively by the young researchers, who then called for more involvement by the adults. The problem is the assumption that [adult] researchers make that we know how to easily produce a video. I really like to work on the project's Instagram, but it is not easy A new working routine emerged, with young researchers and adult researchers taking it in turns to lead the weekly meetings. Moments of reflection were incorporated at the end of each meeting , to allow team members to express their feelings about the meeting or the coproduction process. Both groups felt increasingly comfortable making requests of each other, such as asking others to take on responsibility for timekeeping or maintain transparency with regard to tasks and budgets. Being able to make and respond to such requests made the group a more open and authentic space, where difficulties were aired in a frank yet respectful manner. Stage 2: our differences, our learning community In the second stage, we instituted the Digital Innovation Lab, which aimed to outline, in an intensive week of work, the core elements of our chat-story intervention and the structure of the narrative, on the basis of the findings from the mapping. In this phase, the core coproduction team was joined by the technology company Talk2U, which included professionals in the fields of cinema, communications and technology. We worked more directly with three creative professionals specialising in storytelling and scriptwriting. Our main aim was to create a compelling chat-story that incorporated the most relevant lessons from the mapping and supported young people's participation in the promotion of good mental health. As described in table 1, the solution we found to the challenge of combining multiple sources of expertise was openness to mutual learning. The creative professionals were willing to explore the research data and theoretical framing that we presented as foundational bases for the intervention; the adult and young researchers were willing to learn more about cinematic narrative resources, such as the 'hero's journey'. Both the creative professionals and the adult researchers were open to learning youth culture and were introduced to a range of memes, slang expressions and common situations or phrases relevant to young people's mental health. Stage 3: a secure base from which to connect We integrated into our network several advisory stakeholder groups to support the chat-story in a consultative capacity. Two policy consultants provided recommendations on how to align the chat-story with public policies in education and adolescent health. Our policy consultants recommended bringing the voices of education professionals into the project, a suggestion which was accepted by the research team and subsequently resulted in the creation of a School Community Committee to --- BMJ Global Health support the introduction of the chat-story into school settings. The SCC comprised teachers, school psychologists, educationalists and educational advisers, totalling 10 education professionals in Distrito Federal . These professionals helped to anticipate the challenges of using the chat-story in the school environment and to increase its chances of success. Finally, in order to maximise the cultural relevance of the chat-story at a national level, a Youth Chatstory Advisory Committee was formed. This group was composed of 24 high-school adolescents from all over the country, who provided input into the content and language of the chat-story. Adult researchers and young researchers jointly hosted remote meetings with each of these groups by videoconference, in addition to conversations via WhatsApp for specific input, such as the name of the chat-story. 29 Figure 1 depicts the pluralistic model employed in the Engajadamente project, which brings together individuals and institutions on both a national and an international level. The coproduction process was evidently made more complex by involving a variety of external partners, with different technical specialisms and life experiences, and occupying different spaces, both geographically and in terms of their role in the public sphere: education professionals, public policy experts, artists, technology experts and students. To tackle overload and increase efficiency, we decided that responsibility for managing external groups would be divided among the adult researchers, with at least one young researcher attending meetings. Contributions were documented and actions agreed on at core team meetings. The input received from different advisory groups was substantial and sometimes conflicting, and it occasionally necessitated unexpected work. 'Errância' was a word used repeatedly by the team throughout the project to describe the process, reinforcing its positive aspect: the ability to keep on moving, even when the path taken was not completely in the plans or not planned at all. Sometimes I feel that we are achieving something, and I am quite satisfied, especially after this has been built on the basis of conversation. … A continuous effort. It is no use getting frustrated and saying "I won't do it." How to deal with frustration: work to deal with frustration . The cohesion and trust that existed within the core coproduction team were experienced as an asset, a secure base that supported engagement with a complex and multilayered system of partners and advisors. The core team's aspiration to 'stick together' and engage in active and empathetic listening to solve problems was perceived as a protective factor. --- BMJ Global Health I feel that I am always trying to take the perspective of everyone involved. It is a continuous exercise … you learn over time. The coordination process has been more complex than I imagined; it is difficult to align sometimes. And it is OK to have this complexity because this circle [adult and young researchers] is protected. We accept that there are difficulties along the way, and that we will work through them . Collectivity [in the larger network] does not always have positive moments … The toughest and most tiring moments become lighter when we are together Stage 4: care, transparency and accountability In the fourth stage, coproduction of the chat-story took place. This consisted of the development of the chatstory's narrative script and resources such as videos and cards. This was the longest stage in our research timeline and brought us the most frequent and complex challenges. Cowriting a 72-page script, aimed at fulfilling the dual role of an engaging story and an intervention, in a team of 12 was no mean feat. Challenges related to ensuring the inclusion of all involved were added to other issues typical of complex work done with collaborators and advisors, with several expectations needing to be balanced. Emerging challenges included overload; the need to listen sensitively and incorporate everyone's voices; the ambiguities of work segmentation and the inequality of resources, such as the time available to each member for the project. Two sets of solutions were sought. --- Strengthened coproduction structure The work process with new collaborators revealed the difficulty of integrating the groups, sometimes generating dissatisfaction . Among the solutions created were having many conversations about the importance of working together and valuing the expertise each stakeholder brings . Critically, we created accountability processes, whereby we would, for instance, systematically collate all feedback received, internally and externally, on a particular topic/ scene. The document would later be updated with justifications regarding the incorporation or rejection of each piece of feedback into the script. New roles emerged, somewhat spontaneously. One of the adult researchers took on the task of invigorating relationships between young and adult researchers, while two others invigorated relationships with collaborators, supporting mutual understanding as well as timely and fair resolution of eventual conflicts. We also encouraged core group members to reach out to anyone they trusted in the team to express any difficulties to increase mutual support. Great effort was invested in developing assertive relationships. Within the core team of adult and young researchers, solutions were developed to increase transparency and support collective organisation and efficiency. We began to list weekly work goals that were jointly achieved and those not achieved by the team, via WhatsApp, with emojis of clinking glasses and pineapples . We also started voting on decisions related to changes in the chat-story script where no clear consensus was reached, to ensure that everyone's voices counted equally. We also decided that some of the meetings and small-scale decisions did not require all of us to attend, as long as both groups were represented . Some of the disagreements that emerged through the process were resolved through extensive discussion and negotiation. An example of such an instance was the joint decision-making around the profile and tone of the speech of one the most controversial characters in the story. The young researchers and adult researchers disagreed with the professional scriptwriters as to whether the character should be portrayed as an aggressive 'bully' or a likeable classmate. After several consecutive meetings and script revisions, the group reached the decision to portray the character as a funny troublemaker who lacks mental health literacy. A video/image portfolio of social media posts containing examples of similar behaviour was compiled to consolidate a common understanding of the character's personality. --- Caring for each other As a care strategy for the core team in the face of overload or disagreements, we mixed work evaluation with leisure activities together, near waterfalls and in parks in Brasília. We used remote and in-person activities to share our feelings about coproduction and our different understandings of the coproduction process, jointly designed and facilitated by two young researchers and an adult researcher. One such activity involved coconstructing an online board where members shared images or songs illustrating how they currently felt about the coproduction process. Another activity, facilitated in person in a green space suitable for relaxation, consisted of a collective drawing made by adult and young researchers that depicted barriers to or enablers of the chat-story coproduction. As shown in figure 2, coproduction was expressed through the drawing of swirls, showing that the process perceived as messy, and affective memories. The backpack in the drawing refers to a moment in which the creative collaborators and education professionals consulted disagreed about the inclusion of a scene in the chat-story script, in which the user The coproduction was also portrayed as a two-way arrow pointing to a university graduate and someone without a degree, symbolising the partnership of young and adult researchers; an open umbrella represented the model's rich potential for innovation and creativity. The co-production process is a storm of ideas; we are always accepting and rejecting ideas and building something together. Our potential is endless. … We are always reaching new heights, like going up to infinity The huge creative potential and everyone's commitment to the project meant that the chat-story narrative was continually improved, with over 10 drafts generated. To remind ourselves to overcome collective perfectionism and keep our efforts at a manageable level, we often used a quotation from Isabel Allende, who said of her own writing process: 'You never finish a novel-you just give up'. The exercise of intertwining knowledge and care fostered emotional security. Group members became more likely to share details about their personal life and to introduce humour . Enhanced emotional security increased the space for free and authentic sharing of difficulties, not always related to the project, and sometimes of a personal or delicate nature. For instance, a young researcher shared feelings of burnout and reduced accomplishment as a result of returning to campus after COVID restrictions were lifted, which involved a 4-hour daily commute . When someone does not turn on their camera, does not show up on time, says the whole weight is on them, this can be a cry for help and demands another level of attention and understanding. Yes, it is important to invest in discussing transparency and the role of each person, but also to care for our emotional needs and ascertain the support network The core team became more sensitive to changes in each other's behaviour or eventual disengagements of team members from the coproduction process and potential associated emotional and contextual challenges. Our coproduction process took on new dimensions --- BMJ Global Health when peers started recognising each other as a support network. Greater proximity generated new responsibilities or a duty to care for each other's health and wellbeing. This feeling was particularly salient among the young researchers in relation to each other, and among the adult researchers in relation to the youth group. Meetings between team members were held to discuss issues of a personal nature and signpost resources. I keep thinking about the responsibility that, in the project, we researchers have towards you, the younger ones. … What is the limit to work the two dimensions: the feelings that need to be experienced and expressed and the care that should be offered to those who do that Stage 5: the team as a living organism In the fifth stage, the main project goal was the dissemination of the chat-story and project findings. Dissemination of the digital intervention was led by Talk2U via a social media campaign, which integrated input from the core coproduction team, reaching over 6400 users; a digital influencer with a mental health difference also supported dissemination via Instagram, 42 with a video totalling over 20k views. Adult and young researchers aimed to disseminate the chat-story and research findings via academic conferences, scientific articles and YouTube videos, as well as workshops for teachers, researchers and policymakers. At this stage, our biggest challenge was handling the overload caused by the overlapping tasks, combined with national and international travelling. The solution found was to divide the work: It is important to be aware of one's own feelings and to think of rearrangements, to cover tasks, impossibilities, emotional and physical problems that will arise, not always as a result of the project, sometimes yes. One needs to take another person's task and the team needs to function as a living organism The coproduction model meant that all members felt a sense of responsibility and ownership over all outputs. Roles became relatively interchangeable at this point, as everyone felt confident to speak to any aspect of the project. Delivery of presentations was distributed across team members , ensuring broad dissemination across the fields of psychology, health tech, youth mental health and public health as well as the education sector. Inequalities in certain areas emerged as a challenge, with technical knowledge and time for writing articles being greater among the adult researchers and filmmaking skills better among the young researchers. As for writing research articles, problems were alleviated by the allocation of pairs composed of an adult lead and a young researcher to lead on each article being prepared. For videos, a group of young researchers and an adult researcher took the lead on different outputs, with support from the team. Overload was handled by collective renegotiation of the schedule and responsibilities, according to personal and contextual limitations. Our closing event, jointly chaired by an adult and a young researcher, included researchers, members of parliament, health ministry representatives and local policymakers, most of whom were being introduced to coproduction for the first time. Praise for the working model was effusive. There was no way the chat-story could go wrong, because of the co-production model . Stage 6: friendship for knowledge and care At the end of the project, the perceived impacts of the coproduction process were shared over an evaluation lunch, which evolved into afternoon coffee then dinner, totalling almost 10 hours of relaxed and open conversation. Members of the core team expressed the symbolic and material value of inclusion in the academic field, both for young people and adults. I am grateful for this international cooperation as it has always been very difficult to do research in Brazil with so few resources . The project gave me the conditions to go ahead with my course: I left a complicated environment, I was able to rent my own place and have an income … This project saved me, and made my graduation possible . The radical participation in each other's lives was clear, with team members disclosing personal vulnerabilities at a level that is not commonly seen in research environments, not at least from our previous experience. At one point in my life, I thought I would not have been able to continue as a researcher, after suffering a very stigmatising mental health intervention. My doctor recommended that I apply to work with you … she thought I would be the right person for it and that this collaboration would be powerful . I have always felt I did not belong at university, mainly because of my skin colour and all the sacrifices I had to make to experience university. Whenever we worked together this feeling disappeared . Because of my life trajectory, I needed to work effectively on my own, to get out of very precarious living conditions and expand my networks. It was very special to work among peers, not to be alone. This is my best paper yet . I am going through a unique moment of grappling with our finitude, as I a care for someone nearing the end of life in my family environment. And that mixes cycles of my life . Group members' verbal contributions were interspersed with exchanges of glances between different people, sometimes tearfully, at a round table in a café in Brasília, where young and older researchers spontaneously held hands. The circle at the café closed the project, whose continuation now depends on forming But perhaps being moved by each other's stories is part of the process of pluralistic coproduction, in times of vulnerability, as the research progresses along its trajectory. This cultural clue left us pondering about what defines and structures radical forms of coproduction, where friendship between heterogeneous groups may be an unplanned discovery, but a bond that is possible and perhaps necessary. --- DISCUSSION We have presented an autoethnographic account of the challenges and solutions adopted in the process of coproduction of a chat-story intervention in Brazil, where adults and young people shared power and control, within a network of creative collaborators and allies across education, policy and communication. Differences or heterogeneity in our group, especially in terms of age and professional experience, created blind spots and made it challenging to build a common perception of the whole and the parts involved. The autoethnographic work allowed us to observe the following: continuous investments to improve our collaboration that considered technical, social and personal needs, resulting in greater transparency and power sharing; the emergence of internal trust and cohesiveness, which supported engagement with a complex network of partners; the creation of formal and informal mechanisms for accountability and fair inclusion of multiple voices; reorganisation of responsibilities and care practices to handle fatigue and overload and promote equitable participation and joint care invested in reorganising our work process in each challenging situation, giving rise to synergy and intergroup friendship. Our experience of coproduction contrasts with decades of literature that uses the metaphor of a ladder to characterise participation, especially that of children and young people. The model proposed by Arnstein 43 includes eight steps that go from superficial citizen representation to effective citizen power. This model, as adapted by Hart, 44 maintains the ladder metaphor: going from resistance or impediment to youth participation, through manipulation, decoration and 'tokenism'-which includes consultation with adolescents but 'with minimal opportunities for feedback'-to higher stages of participation, with young subjects who are accountable or are autonomous protagonists themselves. An obvious contrast lies in the idea of a linear, or rather ascending, trajectory, from a lower degree of participation to a higher degree over time. In our project, some challenges could not be promptly met and some of the solutions gave rise to new challenges, for instance, when efforts to amplify participation generated overload or when interpersonal closeness gave rise to duties of care. Consistent with our experience, Chung and Lounsbury 17 described participation in a research project as 'neither linear nor unidirectional; rather it zigzagged up and down as actors negotiated issues of power, process and relationships'. Yet, the authors mention the possibility of a continuum from lower to greater coproduction quality, which we were also able to observe, with progressive ethical adjustments to facilitate synergy and equalise decision-making power. 'Messiness', or non-linearity, is further amplified in complex social, political and economic scenarios, settings where awareness of coproduction is usually low, or where young people's right to life and health is not guaranteed. In our case, instability was compounded by: challenges derived from the COVID-19 pandemic; the negationist approach to science prevalent in Brazilian society; rampant economic insecurity and an epidemic of poor mental health, among other factors. Time was a key resource affected, with time constraints generating tight deadlines and necessitating agility to execute a project that had been designed in another global reality. More empirical research into the practice of coproduction in LMICs, or vulnerable settings, is necessary to understand how larger political and social factors influence the process. This will help inform recommendations on how to strengthen capacity for coproduction in LMICs 12 45 The costs of coproduction have been discussed by researchers in the field, including costs for the process and products of coproduction as well as the personal and professional costs for researchers and stakeholders. 46 However, others have argued that these costs result from structural and organisational failures to accommodate and promote emancipatory research and its outputs, rather than being flaws inherent in this research method. 11 In our experience and that of others, 47 48 the main costs involved in generating a favourable context were related to time, material resources, symbolic context and relational context . In order to anticipate and minimise the personal wear and tear, expected in coproduction projects, a prior assessment of the resources available in the context or project is necessary. This analysis might help determine whether and how coproduction is feasible and capable of creating synergy, that is, 'the power to combine the perspectives, resources, and skills of a group of people and organisations'. 49 This is especially relevant for coproduction that makes use of a broad network of partnerships, with varying needs and priorities. If the context is favourable, coproduction from 'phase 0', when priorities in the research agenda are defined and the research project conceived, can foster successful implementation. 47 However, this is not always possible. In our project, recruitment of the complete research team was only possible after grant funding was secured and an international agreement was signed. The absence of co-production from 'phase 0' was partially circumvented by ensuring that all members had a comprehensive understanding of the project as a whole, including access to full documentation, and the ability to influence --- BMJ Global Health key research decisions beyond the core commitments made in the original proposal. These efforts gradually increased team members' sense of ownership over the project. Once a project is set up, sustaining relationships of trust and organisational structures that maintain partnerships is fundamental to allowing future coproduction that begins from agenda setting. 50 51 Lessons learned In our experience, three key factors were important for the success of our coproduction process. First, we continually sought solutions and adjusted our process to meet challenges encountered along the way, for instance, by reassigning responsibilities and revising timelines. Such flexibility, previously described as a critical ingredient to coproduction success, 18 52 helped achieve inclusive and equitable participation within a team that was heterogeneous in availability, knowledge and resources. This practice required recognising the difference between 'taking responsibility for' and 'taking away control from' the other, the latter being a very common form of paternalism usually found in the relationship between adults and children. 53 Christensen and Prout 54 suggest working from the concept of 'ethical symmetry' in coproduction with young people, acknowledging that every right accorded to adults in research has a counterpart for young people. This recognition of symmetry does not mean ignoring individual needs and demands. It means, rather, remaining sensitive and flexibly adjusting to such needs. Second, we sustained collective trust in the model-a shared understanding of coproduction as capable of surviving mistakes, unplanned successes and errors. Challenges were managed through a shared commitment not only to the stages of production but also to preserving bonds among the researchers and with external partners. All team members made efforts to cultivate a positive relational context, by regulating emotions, valuing different perspectives, validating each other's negative feelings and seeking to reach solutions that would suit the group's needs and priorities. Third, such commitments were operationalised through active organisational adjustments and care movements, coconstructed and agreed between the heterogeneous poles. These movements were initially led by adult researchers, and later cocreated by the two groups. Strategies included: mutual capacity-building ; investment in building interpersonal knowledge and cohesion ; dedicated time for reflection on the coproduction process ; transparency structures ; accountability processes ; adoption of multiple methods to ensure inclusion of all voices ; mutual support and strategies to reduce overload . By coinventing care, we deconstruct participation metaphors that represent the process of coproduction as a single axis-young people-who linearly and progressively accumulate participative experience up to the point of full autonomy. Rather, we aim to favour an understanding of the social, caring and productive bond that allows innovative outcomes, only possible because of the intersection of differences. --- Methodological considerations The collaborative autoethnographic method used proved to be a powerful methodology, affording our group enhanced control over the research process, without an 'intermediary' researcher representing our voices. 55 The method allowed for a shared narrative of our coproduction experience. Consensus through collective discussion validated our observations; however, as with other participatory research models, it did not 'negate differences in perception and experience'". 56 . It is likely that the autoethnographic process itself influenced the results, given that the opportunity for collective reflection strengthened our mutual understanding and cohesion, as observed in previous case studies. 57 The stories revealed were deeply personal and vivid, but they also went through collective maturation, balancing what the group considered relevant and safe to share. It is possible, therefore, that different conclusions would have been reached, or insights gained, by an 'external' researcher conducting participant observation or anonymous, individual interviews with group members. Rather than providing replicable, generalisable results, the goal of our analysis is to offer a practical case study that stimulates reflection and conceptual insight. --- CONCLUSION Our autoethnographic experiment was applied to a pluralistic coproduction project involving young people and adults working together closely to build a digital intervention for mental health in Brazil. That is a combination that is not naturally homogeneous: on one side were researchers linked to traditional research institutions; on the other, young citizens who experience youth environments first hand. The challenges we experienced collectively reveal a non-linear path of coproduction, with progressive ethical adjustments for inclusion, equity and transparency. The autoethnographic method, supported by creative methods that enabled conversations about feelings and difficulties, resulted in a story of coproduced care. The cross-generational friendship, built alongside the willingness to speak frankly and freely, opens new possibilities for radical approaches in coproduction research involving young people. Perhaps these would entail coproduction that does not conform to 'ladder' models, that is suitable for vulnerable contexts, and that intertwines knowledge and mutual care, drawing on the authenticity of each group member and that shows --- Supplemental material This content has been supplied by the author. It has not been vetted by BMJ Publishing Group Limited and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations , and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Introduction Coproduction of mental health research and interventions involving researchers and young people is increasingly common. However, this model raises challenges, related, for instance, to communication, power and control. This paper narrates-from a collective firstperson perspective-the lived experience of coproduction of a digital intervention by institutional researchers and young citizen researchers in Brazil. Method This study employed a collaborative autoethnographic methodology, utilising autobiographical data such as meeting recordings, individual notes and collective guided reflections on the coproduction process. Our analysis focused on challenges and solutions that arose during the process. Results Throughout the project, we created formal and informal mechanisms for accountability, transparency and fair inclusion of multiple voices. We engaged in mutual capacity-building, invested in building interpersonal knowledge, and implemented practices to reduce overload and promote equitable participation. Through ongoing reflection and readjustment in response to challenges, we progressively embraced more democratic and egalitarian values. The collective care invested in the process fostered synergy, trust, and intergroup friendship. Conclusion Our experience points to the value of creating a space for multiple research identities: the citizen young person and the institutional researcher, both of whom critically reflect on their roles in the research process. Our focus on coproduced care calls into question participation metaphors that represent the process via a single axisyoung people-who linearly progress from minimal participation to full autonomy. Instead, our analysis highlights the importance of a social and caring bond that supports the radical co-production of innovative health solutions in contexts of vulnerability.⇒ Our project offers a fresh perspective on coproduction, emphasising the role of social bonds in driving innovative outcomes. This paves the way for more immersive, radical collaboration possibilities between researchers and young people in the development of health research and interventions.

INTRODUCTION Adolescent sexual and reproductive health services for adolescents are often underserved to adolescents in many societies. [1] Currently, there are approximately 1.2 billion adolescents globally who are assets to countries. [2] Adolescents are characterized by a series of physiological, psychological, and social changes that expose them to unhealthy sexual behaviors such as early sex experimentation, unsafe sex, and multiple sexual partners. These put them at high risk of sexual and reproductive health problems. Such problems include early marriage, teenage pregnancies, unsafe abortion, sexually transmitted infections , HIV and AIDS, and other life-threatening SRH problems. For many of these sexually active adolescents, reproductive health services, such as the provision of contraception and treatment for sexually transmitted infections, either are not available or are provided in a way that makes adolescents feel unwelcome and embarrassed. As a result, adolescents are more likely to rely on resources outside the formal health service provision system, such as home remedies, traditional methods of contraception, clandestine abortion, or medicines from shops or traditional health practitioners. [5] Adolescents also go through financial barriers, long waiting times, inconvenient working hours, and lack of parental support when accessing healthcare services. [6,7] Other barriers include lack of adolescent-friendly resources in health facilities, [8] unfriendly and perceived negative attitudes of providers, [9,10] and poor quality of care. [7] Despite the clear need for access to sexual reproductive health services , [11] coverage rates are low. Data from five countries in Sub-Saharan Africa with high rates of new HIV infections found that 7% to 31% of boys and 9% to 58% of girls aged 15 to 19 years had been tested for HIV and received their results. [12] Fewer than half of young men in SSA reported using condoms at the time of the last sexual intercourse, and rates were even lower among young women. [12] In SSA, as many as 68% of adolescents have an unmet need for contraception. [12] Rates of skilled birth attendance is a critical intervention to reduce maternal and newborn mortality of approximately 55% in developing countries; however, coverage is low among adolescent births, despite the higher risk related to young maternal age. [12] Efforts in recent years have focused not only on ensuring health service availability but also on making its provision adolescent-friendlythat is, accessible, acceptable, equitable, appropriate, and effective. [13] In Ghana, the utilization of ASRH services by adolescents remains poor. [14] Community participation in the National Adolescent Health and Development program has been weak, and less improvement in the various adolescent health indicators was reported. [14] The 2022 Ghana Demographic and Health Survey indicated that 15% of girls aged 15 to 19 have ever been pregnant, including 11% who have had a live birth, 4% who have had a pregnancy loss, and 2% who are currently pregnant. [15] To address these issues, several initiatives have been developed by the World Health Organization and implemented that have made it easier for adolescents to obtain the quality health services that they need, by making health services "adolescent-friendly." However, a report by WHO argues that healthcare for adolescents remains highly fragmented, poorly coordinated, and uneven in quality. [11] For instance, the age of adolescents around the globe does not allow them access to health information, coupled with poor provider attitudes, lack of privacy, and weak parental support to access available healthcare services. [16] Sustainable Development Goal 3 focuses on ensuring healthy lives and promoting the well-being of all ages, including adolescents. [11,17] To achieve this goal, it is very important to consider the health and well-being of adolescents in development agendas. [18] Quality improvement is necessary in every sector given the high demand for quality goods and services and the strong competition that characterizes the corporate world. Quality assurance in healthcare is a necessity and should not be compromised, because human lives are at risk. [19] In addition, the health sector, like any other sector, is subject to competition, especially with the fast-growing private health sector. Hence, patients expect nothing but quality healthcare. Giving a precise definition of quality is difficult because of its subjective and intangible nature. Setting and assessing quality in healthcare is harder than other disciplines because new definitions for healthcare quality are emerging. [20] The Institute of Medicine defined healthcare quality as the degree to which healthcare services for individuals and populations increase the likelihood of desired health effects and are coherent with current professional knowledge. [19] This means that healthcare delivery must be in line with professional criteria and principles and must match the expectations of patients. The Institute of Medicine's domains of healthcare quality are effectiveness, efficiency, equity, patientcenteredness, safety, and timeliness. This study adopted the Donabedian model of healthcare quality assessment. Avedis Donabedian, a doctor and a health services researcher at the University of Michigan, developed a healthcare quality model in 1966. Donabedian defined healthcare quality as the application of medical services and engineering in a way that maximizes its benefits to health without correspondingly increasing the risk. Donabedian is well recognized for his structure, process-outcome quality model, which he believes is based on a systems approach to thinking about healthcare quality. [19] Donabedian argued that quality could be assessed based on three main domains, namely: structure, process, and outcome. According to Donabedian, structure refers to the characteristics of the health facility in which maintenance is delivered and accessed. Models are amenities, equipment, human imagination, and organizational structures. The structure may also include the health provider's skills, operating hours of the facility, and convenience in scheduling appointments. Donabedian defined process as what is done in the giving and receiving of health concerns. He classified processes into clinical and interpersonal processes. Clinical process refers to the clinical guidelines and criteria that must be discovered by health providers. The interpersonal process refers to the interaction between the patient and the provider. [22] The process extends to diagnosis, treatment, prevention, patient education, patient and family activities, access to care, and healthcare utilization. It is likewise the ability to build a relationship of faith, empathy, and understanding with patients, and showing humanism and sensitivity to patient needs as well. [22,23] The procedure extends to discussing or explaining the patient's conditions to them and taking them in decisions involving their care. Thus, the process comprises all the activities between structure and outcome. [21] Outcome or the product of care refers to the effect of the care on the health status of the patient and the population. [21] It includes improvements in a patient's knowledge, changes in behavior, and patient satisfaction; that is, a reduction in mortality, morbidity, disability, and improvement in patients' perceptions. Donabedian argued that there is a relationship among structure, process, and outcome. He emphasized that quality assessment is possible simply because good structure increases the likelihood of good process, and good process increases the likelihood of good outcome. [21] Construction and process can influence outcome, either immediately or indirectly. --- METHODS Ethical approval was obtained from the Ghana Health Service Ethical Review Committee of the Research and Development Division of the Ghana Health Services. Permission was obtained from the selected health facilities and respondents. Parental consent was sought from adolescents who are younger than 18. This study used a facility-based descriptive crosssectional study design to assess the structural and process factors available for delivering quality adolescent sexual reproductive health services in the Volta, Eastern and Oti regions of Ghana from November 2021 to April 2022. The target population for this study included all adolescents , and all health facilities that provided ASRH services in the selected three regions. The three regions were purposefully selected due to the high rate of adolescent pregnancy and fertility rate [15,24] compared with other regions within the southern zone of Ghana. A simple random sampling by balloting was used to select 158 health facilities across the three regions for the assessment of the structural factors. A total of 158 adolescents who used ASRH services were also randomly sampled to assess the process factors available for the delivery of quality ASRH services using an existing interview guide. The WHO global standards for quality healthcare services for adolescent assessment were adopted and used to collect data from the respondents. [11] Only the sections that elicit information on ASRH services were adopted. The Donabedian model of quality assessment was adopted, modified, and used to assess the structural and process factors at the selected existing health facilities. The Statistical Package for the Social Sciences version 20.0 was used to analyze the data collected and the findings presented in the tables. --- RESULTS --- Facility Background Information Most of the facilities assessed were in the rural communities, and 37 were in urban communities. Most of the facilities were owned by the government of Ghana. A total of 32.91% of facilities were community health and planning services, 52 were health centers, and 26 were hospitals, respectively, as shown in Table 1. --- Structural Factors in Delivering Quality ASRH Services --- Facility characteristics The study assessed the structural factors available to ensure the delivery of quality ASRH services in Ghana. The results showed that most health facilities had a clean and welcoming environment. A proportion of 85 facilities did not have separate space for adolescents, and a proportion of 73 facilities had a separate space for adolescents. Most health facilities had signposts indicating services available and the working hours for adolescents, and 68 of the facilities did not have signposts to indicate services/working hours for adolescents . All 158 health facilities had the National Health Insurance Scheme covering contraceptive/ family planning services for adolescents. Although most facilities provided privacy , 32 of the facilities did not ensure privacy for adolescents. In most of the facilities, there were ASRH teaching and learning aids available to help service providers. Also, most of the facilities had standard guidelines, protocols, and policies available to guide service provision. The average waiting time for adolescents to be attended to by service providers was 30 minutes or less . --- Adolescent health literacy As shown in Table 3, the study also assessed the adolescent health literacy materials available to ensure the delivery of quality ASRH services in Ghana. Most facilities had available educational materials on ASRH, and a proportion of 30 of the facilities had no educational material on ASRH for adolescent education. However, most facilities that had educational material on ASRH education had them in only one language , whereas only 21 of the facilities had them in different languages. In all the facilities that had educational materials on ASRH, they were not available for take home by adolescents in most of the 102 facilities. Only 26 had educational materials for take home by adolescents. Adolescents can only access services offsite in 63 facilities assessed in Ghana . --- Process Factors in Delivering Quality ASRH Services Equity and non-discrimination Table 4 shows that most of the respondents did not require parental, spouse, or guardian consent before using ASRH services in health facilities, whereas a proportion of respondents required parental, spouse, or guardian consent to use ASRH services. When asked whether ASRH service was rendered based on age race, gender, marital, educational, or economic status, 156 respondents said "No." A proportion of respondents said they were once denied services because they did not have active NHIS or did not have money to pay for it . --- Community support The study showed that there were no cultural/religious values in the respondent's community that did not support ASRH services. Most of respondents responded "No" to not having any cultural/religious values in their communities that did not support ASRH services. --- Provider competencies and attitude Findings from the study showed that ASRH service providers respected the opinions, decisions, and choice of services of adolescents. Service providers allowed adolescents to ask questions and ask for clarification on service availability. Service providers treated adolescents respectfully and were friendly. Most respondents believed that their information would be kept confidential by the service provider. When asked whether there was any interruption by others when the provider was attending to them, Most responded "Yes," attesting to the fact that privacy was not provided in most of the facilities when the service provider was attending to adolescents . --- Adolescent participation The study assessed whether adolescents were involved in the planning, monitoring, and evaluation of health services and decisions regarding their care, as well as in certain appropriate aspects of service provision by health facilities. The study showed that most facilities did not involve adolescents in the planning, monitoring, and evaluation of health services and decisions regarding their care, as well as in certain appropriate aspects of service provision. The study, however, found that most adolescents believed they were better positioned to suggest ways in which facilities could improve the quality of ASRH service delivery. --- Appropriate package of services In assessing the health facility's ability to provide a package of information, counseling, diagnostic, treatment, and care services that fulfilled the needs of all adolescents as well as how ASRH services were provided in the facility and through referral linkages and outreach systems, results of the study showed that all 158 facilities accessed were able to provide services to adolescents and were not referred to other facilities for the intended services. --- Data and quality improvement Facilities' capacity to collect, analyze, and use data on service utilization and quality of care, disaggregated by age and sex, to support quality improvement by policies and protocols information storage and disclosure was assessed. Findings, however, show that adolescents do not know of any written policies or protocols on adolescent information storage and disclosure in the facility. --- DISCUSSION --- Structural Factors in Delivering Quality ASRH Services The study found some structural barriers that affected the delivery of quality ASRH services in Ghana. The results showed that a proportion of 85 of the facilities assessed did not have separate spaces for delivering services for adolescents. This finding is similar to the finding in a study carried out in South Africa, Ethiopia, and Uganda. [25][26][27] The studies indicated a lack of a dedicated space for young people at the facilities. [25,26] In a study from Ethiopia, one of the participants indicated the lack of separate youth clinics saying, that designated space for the provision of Youth-Friendly Sexual Reproductive Health services has been mentioned numerous times as a barrier. Even where youth clinics exist, participants report a lack of privacy for SRH services and/or a sense of belonging. "When you go to hospitals for services, you may meet your parents there. I remember my friend who met her mother in a clinic." [28] However, even though there was a lack of separate space for the provision of ASRH services, the study found the majority 126 of facilities provided privacy in delivering ASRH services. This finding is, however, contrary to the findings of Rukundo et al. [26] and Wakjira and Habedi, [29] who indicated a lack of privacy in providing ASRH services. Many operational barriers in health facilities also impact access and utilization of ASRH services, such as inconvenient operating times, lack of transportation, and high cost of services. However, all the health facilities assessed had the NHIS covering contraceptive/FP services for adolescents in Ghana. This made it easy for adolescents in Ghana to access and afford the cost of services delivered. Three studies reported that adolescents and young people mostly preferred low-cost or no charges at all when seeking SRH services from youth centers. [27][28][29] However, studies in Uganda, Nigeria, and Kenya [30][31][32] showed contrary results, as in 19 of the 20 focus group discussions , adolescents noted that ASRH services were not free, and the cost was not affordable to them. --- Process Factors in Delivering Quality ASRH Services The study assessed ASRH service equity and non-discrimination in service delivery. The findings indicated that most respondents did not require parental, spouse, or guardian consent before using ASRH services. The study also found that respondents were not denied access to ASRH services based on age, race, gender, marital, or educational status. However, a proportion of respondents said they were once denied services because they did not have active NHIS or did not have money to pay for it. Studies in Uganda, Nigeria, and Kenya, [30][31][32] however, showed that in 19 of the 20 FGDs, adolescents noted that ASRH services were not free, and the cost was not affordable to them. Findings from the study showed that ASRH service providers had received adequate training and were wellskilled to provide ASRH services. The study indicated that service providers respected the opinions, decisions, and choices of services of adolescents. Service providers allowed adolescents to ask questions and ask for clarification on service availability. Most respondents believed that their information would be kept confidential by the service provider. Contrary to this finding, a study in Tanzania indicated that only 37.2% of the service providers received training in ASRH information and counseling, which is significantly very low. [33] Also, studies from South Africa, Uganda, and Ethiopia found inconsistencies in the perceptions of provider attitudes toward adolescents. [27,30,34,35] Negative attitude of health workers as per the case in one of the studies indicated that 30% had negative attitudes toward the youth in Ethiopia. [34] From FGDs in a study done in Uganda, 18 of 20 participants indicated that experiencing healthcare providers' negative attitudes toward providing SRH services affects the utilization aspects among adolescents. [30] Health worker attitudes can also significantly hinder adolescents' utilization of reproductive health services. Services need to be provided in a youth-friendly environment with health workers who are welcoming and supportive toward adolescents seeking care. [36] Community norms and traditions have a powerful influence on health. They can and, in some places, do, promote progressive and pro-social actions. [37] In many places, however, especially about ASRH, norms and traditions hinder rather than help. [37] The study, however, showed that there were no cultural/religious values in the respondent's community that did not support ASRH services. Contrary to this finding, studies [31,38,39] identified that social-cultural factors were greatly associated with some services mainly FP, voluntary counseling, and testing, and counseling services. It was established that some cultures and parents in a community cross-sectional study done in Kenya and Ethiopia prohibited the youth from using ASRH services as this was brought out when descriptive, chi-square, and odds statistics all showed significant relationships. [38] Some participants in a study done in Malawi indicated that parents expressed negative opinions of youth using FP and parents could prevent youth from accessing FP services and said youth younger than 18 are not old enough to be sexually active. Therefore, the youth did not need FP and should focus on completing their education and not engage in sexual activities. [39] The study showed that facilities did not involve adolescents in the planning, design, implementation, monitoring, and evaluation of health services and decisions regarding ASRH services. However, adolescents believed they were better positioned to suggest ways in which facilities could improve the quality of ASRH services delivered. WHO has called for the participatory engagement of adolescents, supporting programs and policies that are "partnership-driven, evidence-informed, gender-responsive, human rights-based, sustainable, people-centered, and community-owned." [11,40] Facilities require information about adolescent clients, and they must be able to keep this information confidential. Findings, however, show that adolescents did not know of any written policies or protocols on adolescent information storage and disclosure in the facility. Findings of a study indicated that less than half of facilities in Mali and Niger collected information to improve adolescent healthcare and 30% of facilities in Guinea did. [41] Limitations Many of the limitations encountered point to the pressing need for further research on how to best deliver quality adolescent SRH services and determine which components are most effective. Also, this study was limited to only three regions of Ghana of a total of 16 regions. Thus, the study was limited in terms of the sample size implying the generalizability of findings. --- Recommendation Health facilities need to make available ASRH educational material for adolescents. These materials will equip adolescents with the needed knowledge, skills, attitudes, and values that will empower them to realize their health, well-being, and dignity; develop respectful social and sexual relationships; consider how their choices affect their well-being and that of others; and understand and ensure the protection of their SRH rights throughout their lives. Because this is limited to selected facilities in only three regions of Ghana, further study to include facilities in all regions of Ghana are needed to verify if the findings apply to other facilities in the country. --- CONCLUSIONS The study found structural and process barriers that affected the delivery of quality ASRH services in Ghana. To improve the quality of ASRH services, standards for ASRH services for adolescents should include considerations related to adolescents' health literacy, community support, appropriate packages of services, healthcare providers' competencies, facility characteristics, equity and non-discrimination, highquality data, and adolescent participation.

Introduction: Sexual and reproductive health services are often underserved to adolescents in many societies. For many of these sexually active adolescents, reproductive health services such as the provision of contraception and treatment for sexually transmitted infections, either are not available or are provided in a way that makes adolescents feel unwelcome and embarrassed. This study assessed the structural and process factors available in delivering quality adolescent sexual and reproductive health (ASRH) services in health facilities across three regions in Ghana. Methods: A facility-based descriptive cross-sectional study assessed the structural and process factors available for delivering quality adolescent sexual reproductive health services in 158 selected health facilities across three regions (Oti, Eastern, and Volta) of Ghana. A simple random sampling by balloting was used to select the health facilities and a total of 158 adolescents who used ASRH services in the selected facilities were sampled for an existing interview. The Donabedian model of quality assessment was adopted and modified into an assessment tool and a questionnaire to assess the selected health facilities and respondents. The Statistical Package for the Social Sciences (SPSS) version 20.0 was used to analyze the data collected and the findings presented in the tables. Results: The study found some structural and process barriers that affected the delivery of quality ASRH services in Ghana. A proportion of 85 (53.50%) of the facilities assessed did not have separate spaces for delivering services for adolescents. All 158 health facilities had the National Health Insurance Scheme (NHIS) covering contraceptive/ family planning services for adolescents. Most (128, 81.01%) facilities had available educational materials on ASRH but were not made available for take home by adolescents. The findings indicated that most respondents did not require parental, spouse, or guardian consent before using ASRH services. The average waiting time for adolescents to be attended to by service providers was 30 minutes. Conclusions: The study found some structural and process barriers that affected the delivery of quality ASRH services in Ghana. ASRH services, particularly contraceptive/ family planning services, were well integrated into NHIS to improve access and utilization by adolescents.

It is common in research ethics codes, and in the procedures employed by Institutional Review Boards or Research Ethics Committees , now operating in universities and other institutions in many countries, for certain categories of research participants to be defined as vulnerable, with heightened protections placed upon research projects involving them. 1 Typical categories of the vulnerable listed include: young children, people with learning difficulties, sex workers, prisoners, people in poverty, and those suffering from serious illnesses. However, there is scope for disagreement about who should, and should not, be included in this list. For instance, some commentators challenge treating children as vulnerable because this implies that they lack competence or are powerless. 2 Other commentators have pointed out that groups not normally considered as such may nevertheless be vulnerable, including even members of political elites. 3 And at least one writer has challenged the very appropriateness of the concept of vulnerability in the context of research ethics. 4 Even writers of standard texts in the field often indicate their concerns about this concept in the way that they write about it: for example, Wiles 5 refers to "Research with children and other 'vulnerable groups' who are viewed as lacking the 'capacity' to give consent […]." In this paper we examine the concept of vulnerability, primarily in the context of social research, with a view to clarifying some of the issues raised in the literature. 6 As will become clear, we believe that this concept does have an important role to play in research ethics, but that there are many complexities surrounding it, and that its ramifications are considerably wider than often recognized. --- Who is vulnerable? In the broadest terms, vulnerability can be interpreted as an inability, for whatever reason, to protect one's own interests. 7 In the context of ethical regulation, the focus has been primarily on those who may not be able to provide informed consent. Van den Hoonaard 8 writes that "The concept of vulnerability has been the keystone test in medical research when researchers had to know whether a research subject had the capacity to understand and give consent to being researched." And, as he goes on to note, in social research, too, "vulnerable" is often taken to mean "potentially incapable of providing genuine informed consent," whether through an inability to understand the information 1 See, for instance: Economic and Social Research Council . Van den Hoonaard reviews what a wide range of ethics regulatory bodies, in several countries, say about vulnerable groups and how they should be treated by researchers. See also Bracken-Roche et al. . A very large number of social categories have been listed as vulnerable at one time or another, see Sieber : 93. For a brief history of the origins of the concern with vulnerability and subsequent interpretations of the term, see Levine et al. . Liamputtong offers guidance for research with "the vulnerable." 2 Morrow and Richards ; Farrell ; Wright . 3 Traianou . 4 Van den Hoonaard ; see also Levine et al. . 5 Wiles : 31. 6 For attempts at clarification in the field of bioethics, see DeBruin ; Schroeder and Gefenas ; Rogers et al. ; Lange et al. ; Wendler ; Boldt ; Gordon . 7 Feinberg . 8 Van den Hoonaard : 305. provided or to exercise autonomous decision-making. However, in ordinary usage, the usual meaning of the word "vulnerable" is rather different from this focus on informed consent. A common definition is: "susceptible to attack or injury, physical or non-physical." 9 Given this, of the various ethics principles listed in codes and texts on research ethics, minimizing harm is the most relevant to this second definition of vulnerability, though it would also need to include the protection of privacy and prevention of exploitation by researchers or others. While they overlap, these two interpretations of the term are not isomorphic: it is the case that those deemed incapable of providing informed consent may be more susceptible to harm than others, but the converse is not necessarily true. In this paper we will focus primarily on the second interpretation of vulnerability as susceptibility to the threat of harm. It is important to emphasize that vulnerability, in both the senses we have discussed, is a matter of degree. 10 Thus, it has often been pointed out that the notion of fully informed, entirely free consent is a mirage. 11 Indeed, in practice, participants may often gain relatively little understanding of what will be involved in the research process even when they sign a consent form. This can be for a variety of reasons ; and they may not be entirely free to consent or refuse consent, for instance as a result of kin group, peer group, or organisational constraints. 12 Similarly, the likelihood and severity of harm are also matters of degree. 13 We are all vulnerable to some threat of harm, this arising in part from the fact that we are "rational dependent animals." 14 So the key question is not who is and who is not vulnerable, but rather what are degrees of vulnerability. 15 Where vulnerability is taken to indicate the need for extra protection, some threshold must be assumed above which this is required. Thus, judgments have to be made about how vulnerable the participants in a study are. And in this it needs to be remembered that individual members of any category deemed vulnerable will vary in their level of vulnerability, so that these decisions have to take account of all the relevant characteristics and social relations of the particular people involved. 16 Part of what is involved here is intersectionality: people are members of multiple categories, and judgments about their vulnerability may differ depending upon which category is prioritized. While IRBs and RECs typically treat some categories of people as significantly more vulnerable than others, the sort of pre-emptive ethical regulation that has come to be established in many Western countries, and elsewhere, effectively assumes that all research participants are sufficiently vulnerable that they must be protected by the screening of research proposals, any of these deemed unethical being modified or blocked. Furthermore, this implies incapacity on the part of researchers to act ethically, unaided, or at least to be relied upon to do this. 17 The alleged source of this incapacity is that they have a vested interest in pursuing research since it brings them career advantages. As van den Hoonaard18 points out, it appears that ethics committees frequently regard researchers as "powerful, potentially uncontrolled and dangerous,"19 as well as assuming that research participants are not capable of protecting their own interests. It is on such oversimplified assumptions that damaging policies are frequently based. This pre-emptive procedure initially arose in medical research, in response to abuses which caused or seriously threatened physical harm to patients. The regulatory system that resulted, initially in the United States and later spreading to some other countries, was certainly justified, though its subsequent development has created serious problems even in the medical field. 20 The extension of this system to social research is much more open to question, given that the character of investigations there is very different. 21 While a number of ethical controversies have arisen in this field, these have generally presented considerably lower level of potential harm than is involved in much medical research. 22 After all, surgical or pharmacological treatments that can pose quite severe risks to physical health are absent from most social research. 23 Here, often, no research intervention is involved, people are simply asked to fill in a questionnaire, interviewed, or observed in their normal activities. Of course, risks of harm can still arise -to reputations, economic interests, or mental health -but social researchers' activities rarely seriously threaten participants' wellbeing, by comparison not just with medical research but also when judged against other factors in their environment. 24 Of course, this low level of risk might be judged irrelevant if the primary concern is not harm but respect for the autonomy of research participants. Perhaps this is why "vulnerability" is frequently treated as an incapacity to provide sufficiently informed, and sufficiently free, consent. But there is a danger here not just of informed consent being used as a means of passing on the responsibility for minimizing harm to participants themselves, but also of those judged incapable of informed consent being excluded from research participation, and their perspectives remaining unrepresented as a result. Furthermore, as we have noted elsewhere, the frequently recommended strategy for overcoming an incapacity to provide informed consent involves a contradiction: allowing others to provide proxy informed consent does not respect a person's autonomy. 25 Equally, though, the driver behind the pre-emptive regulatory system may be neither a commitment to protecting participants from harm nor ensuring that their autonomy is respected. Critics have suggested that a primary concern -on the part of funding bodies, universities, and other institutions 26 -is with the risk of litigation, financial penalty, and/or damage to public reputation, should problems arise in research they have sponsored. And this is a concern that is likely to be reinforced by the companies providing them with insurance against these threats. Here "vulnerability" takes on a new meaning, both in terms of who is being treated as vulnerable, and what they are being treated as vulnerable to. 27 We have suggested that, generally speaking, the likelihood of severe harm to participants in social research is low, when judged against much medical research and the background sources of potential harm that operate in people's lives. We have also indicated that we believe, with van den Hoonaard, that IRBs/RECs tend to exaggerate the vulnerability to harm of social research participants, and we have suggested an explanation for this. However, this does not mean that social researchers can ignore potential vulnerabilities to harm on the part of the people they study. The key point, instead, is that any general labelling of people belonging to a particular category as vulnerable can be no more than indicative, suggesting the need for awareness that there may be distinctive threats of harm that need attention. 28 This label should be a starting point for ethical reflection, not immediate grounds for imposing extra safeguards. Moreover, judgments about this will need to take account of the particular character of the situations and of the participants involved: in short, these judgments must be contextually sensitive. --- Who is vulnerable to what? Who is responsible for what? While it is often treated as an abstraction, "vulnerability" is a relational term: to say that people are vulnerable is to imply that they are susceptible to particular kinds of threat, whether these are specified or not. And, if we examine the social categories typically listed under the heading of "the vulnerable," it becomes obvious that they relate to different sources and types of potential harm, ranging through physical injury, psychological damage, material loss, and tarnished reputation, as well as obstruction to ongoing activities. 29 For instance, babies are vulnerable to physical dangers and emotional distress, but probably not to embarrassment or damage to reputation. Similarly, the 25 Traianou and Hammersley . 26 Rustin ; Dingwall . 27 Sluka . 28 See Gordon . 29 For further discussion of types of harm, see Hammersley and Traianou : chapter 3. --- 5/16 vulnerabilities of adults suffering from terminal illnesses are likely to be different from those in poverty . Indeed, once we start thinking in these terms, it becomes clear that people outside of the categories routinely labelled vulnerable may be susceptible to specific threats of harm that require attention from researchers. To take an example we mentioned earlier, members of political elites can be more susceptible than other people to damage to their public reputations if they are identifiable by readers of research reports, and they may be more easily recognizable than others. At the same time, these people are probably much less vulnerable to exploitation by the researcher -indeed, they often take control of the research relationship. 30 One implication of this is that vulnerability to particular harms must be monitored by researchers in relation to all participants, not just those belonging to "vulnerable" categories. However, this opens up another question: for what threats of harm are researchers to be held responsible? It should not be assumed that researchers have an all-encompassing responsibility to keep participants safe from harm. We have already suggested that there is a background threshold beyond which the small likelihood and/or low likely severity of a harm means that it can be ignored by the researcher. Equally important, there are some kinds of vulnerability that, arguably, researchers can legitimately treat as outside of their responsibility to control. This is recognized in some ethics codes, and by IRBs/RECs, when they insist that researchers have a responsibility to report crimes or abuse that they discover in the field. Such reports will, after all, cause harm to those whose actions are being reported. Here, vulnerability to arrest and criminal prosecution or to other kinds of penalty, on the part of lawbreakers or abusers, is excluded from the responsibility of researchers. 31 There are some other types of harm that are often treated as beyond the responsibility of researchers as well, for example the impact on key decision-makers when discrepancies are documented between stated policies and what actually happens on the ground. What this makes clear is that the limits to researchers' responsibility must also be given attention. Of some relevance here is the distinction between threats that are internal and those that are external to the research process. As we have noted, discussions of vulnerability in the context of social research recognize the possibility that the researcher will witness or hear about abuse by other actors in the setting, or beyond the setting, being investigated. This may be by parents or siblings in a family, carers within an organization, or even by research participants who are themselves designated as vulnerable. 32 Researchers clearly do not have direct responsibility for this harm, but do they have an obligation to intervene to curtail or prevent it? The normal responsibilities of a citizen or person are involved here, rather than ones arising specifically from being a researcher. We might even ask whether being a researcher can involve suspending such normal responsibilities to some degree? By contrast, when researchers have latent professional identities that carry additional obligations, they may need to intervene, for example by reporting what has happened. Even here, though, judgments must be made about the seriousness of the abuse involved, about the extent of the researcher's responsibility to prevent or report it, about what lines of action could be taken, and about what are likely to be the consequences of these, for the vulnerable people involved and for others, as well as for the research. The last of these considerations should not be underplayed, and the likely value of the research must be judged, not in terms of its benefits for the career of the researcher, but rather for its contribution to collective knowledge: such knowledge is a public good. Difficult decisions are involved in dealing with external sources of harm to participants, then, about which there can be reasonable disagreement. In the case of crimes committed by research participants , researchers may feel a responsibility to protect participants by not reporting them. 33 This is likely to stem not so much from a universalistic commitment to minimizing harm or respecting autonomy, but rather from what they regard as obligations coming out of the relationships built up as part of the research process. This highlights the fact that the range of considerations that researchers need to take into account in making practical research decisions is much wider than the set of ethical principles usually included in ethics codes and texts on social research ethics. Like all of us, they must recognize particularistic obligations, arising from relations of trust, friendship, and so on. 34 A further complication is that a researcher usually has closer relations with some people in the field than with others, and, given that there could be conflict between individuals or groups, careful reflection, and perhaps negotiation, may be required about what obligations have been incurred. 35 Another caution that needs to be sounded is that social researchers are never in total control of the situations in which they operate; nor are they all-powerful in relation to participants. This can be true, for instance, when they are carrying out interviews with informants who come from high status social groups or powerful elites. 36 Similarly, researchers have very limited control when doing ethnographic research in settings that are the domains of others. Given that "ought implies can," there may be harms that arise for 33 A controversial example is provided by Alice Goffman's research in a low-income black neighbourhood in Philadelphia: she not only did not report crimes she heard about, or witnessed, but admitted actively assisting action on the part of participants that could have resulted in murder. For the arguments of one of her critics, see Lubet : chapter 8. 34 Goodin argues that these particularistic obligations are frequently exaggerated at the expense of a broader commitment we have to protect the vulnerable, whoever they are. However, we believe these obligations are nevertheless very important, in the context of research and beyond. Exponents of a feminist ethics of care would agree -for a discussion of vulnerability from this ethical perspective, see Dodds . 35 Hammersley and Atkinson : chapter 4. Another issue that may arise from working with participants who are vulnerable to harm is that the researcher may feel an obligation to provide assistance. See, for example, van Dijk . 36 Neal and McLaughlin . 8/16 participants, whether directly from the research or from other sources, that researchers simply cannot control. Obviously, some initial assessment must be made of the risks, to determine whether the investigation is justified, but uncertain judgments are necessarily involved, and the best decisions that can be made rely on detailed knowledge of the particular people and situations being studied, along with assessments of the value of the specific research project. --- Researchers as vulnerable Following on from this, we should note that in much discussion of social research ethics it tends to be assumed that researchers have relatively high social status and power, while participants are vulnerable because they are relatively low status and powerless. While this model matches some research, it is at odds with a significant portion of it. 37 The researcher may be a postgraduate student or a junior member of staff on a temporary contract, they may belong to a minority or oppressed group, while the participants could be relatively high-ranking members of an established profession, or managers in a large organization. 38 And the latter may seek to use their power to serve their own interests, for instance asking the researcher for confidential information as a quid pro quo for facilitating access to data. 39 Indeed, we should note that, to a considerable extent, researchers are always dependent on the cooperation of research participants to get access to data: gatekeepers may block entry to sites where observation could take place, or to key informants; people in the field could refuse consent to be observed or interviewed, or may actively obstruct the research. As Kim 40 notes, researchers do not usually have the power to force gatekeepers and participants to cooperate with their research, even if this were legitimate; and this is true even in the case of the relatively powerless in society, including those deemed "vulnerable." The fact that researchers are not all-powerful also indicates that they, too, can be vulnerable. Some attention has been given to this in the literature. 41 Most of what we have said about participant vulnerability also applies to researchers: the primary issue is the threat of harm, but this is a matter of degree, and there are different types of harm. Projects vary considerably in how risky they are for researchers, and in how serious is the harm that could be involved, as well as in the types and sources of harm. Furthermore, while reasonable assessments can be made about these matters, here again perfect prediction is not possible. For example, in planning research with sex workers in Guatemala, Warden 42 was aware that she would face some danger, but this did not prepare her for the "existential shock" of witnessing and fearing extreme violence. Worse than this was the fact that she struggled to adapt to what the women she was studying had to cope 37 Kim . 38 See, for instance, Ozga and Gewirtz ; Grek ; Grek . 39 See, for an example, Alcadipani and Hodgson : 136. 40 Kim . 41 Lee ; Nordstrom and Robben ; Behar ; Lyng ; Lee-Treweek and Linkogle ; Downey et al. ; Bloor et al. ; Luxardo et al. ; Chevalier ; Laar ; Sampson . 42 Warden : 152; Nordstorm and Robben : 13. 9/16 with all the time. She writes that "The fragility of my body and the ease with which life was destroyed in Guatemala was a grim actuality to normalize." 43 But the most serious harm she experienced was not physical attack but the trauma of leaving the field while knowing that the women whose lives she had studied remained in great danger. She felt that she had been "only a tourist to their troubles." 44 She writes that "after leaving the field I could not turn off my emotional adaptation to Guatemala," and 45 she points to "the vulnerability that accompanies empathy." 46 The effects continued after she had arrived back in Scotland, indeed in some ways they became worse: I felt it incredibly difficult to phone my colleagues in Guatemala at the organization I worked with because of a mixture of survivor's guilt and my own avoidance strategy for fear of reliving my connection with Guatemala that sparked involuntary feelings of panic, but mostly I was afraid to hear if someone had been murdered while I had been safe in Scotland that would worsen my guilt and send me into a shame spiral. 47 Not surprisingly, she found it hard to analyze the data she had collected because this triggered the post-traumatic stress disorder that resulted from her fieldwork experience. However, most research projects do not involve this level of threat of severe harm for the researcher. A key question about the risk of harm to researchers concerns who is to decide what is excessive risk, and on what basis. One might think that this should be down to individual researchers themselves, and we believe that this is generally true. But complexities arise in the context of research teams, where junior researchers may feel obliged to take on risky assignments against their better judgment, for fear of losing their jobs or damaging their future careers. Slightly different problems arise where the researcher is a postgraduate student, since their supervisors, and the academic departments to which they belong, may feel that they have a duty of care. Aside from this, here again, the potential legal liability of institutions can result in researchers' willingness to put themselves in jeopardy being curbed. Whether for good or bad, this is properly a matter of judgment in particular cases. Clearly it is important that researchers try to inform themselves about any serious risk of harm they may face in the field, and they should take whatever precautions are available against this. A distinction is sometimes drawn between "ambient" and "situational" danger. 48 The former refers to variation in background level of threat across situations: as Warden 49 points out, Guatemala City is one of the most violent places in the world. Sampson and Thomas 50 report on the distinctive dangers associated with women doing research on board cargo ships, male-dominated environments where 10/16 one is "trapped in the field" for considerable periods. But they also detail the various strategies they employed to reduce these dangers, and to deal with them when problems arose. By contrast, situational or occasional danger irrupts in a situation over and above any predictable level of threat, even when it is prompted by the presence or actions of the researcher. This is much more difficult to anticipate or prepare for. --- Conflicting values and the vulnerability of research participants Much criticism of the notion of vulnerability in the context of research ethics has pointed out that the measures recommended to deal with "vulnerable" research participants may contravene other values that researchers ought to respect. 51 An obvious conflict arises from the fact that protecting people implies that they are not able to protect themselves: that they lack capability or competence in this respect. In other words, it may reinforce stereotypes, 52 and could even actually contribute to rendering people incapable in relevant respects by depriving them of the opportunity to learn what is required for this. Closely associated is the complaint that the concept of vulnerability implies an incapacity to exercise autonomous, or at least rational, decision-making, especially in providing informed consent. Mackenzie et al 53 comment on "the danger of using discourses of vulnerability and protection to justify unwarranted paternalism and coercion of individuals and groups identified as vulnerable." In these terms, to label someone as vulnerable may be at odds with respecting their autonomy. 54 Indeed, the measures used to provide protection will often actually prevent people from exercising autonomy: for example, they may be excluded from a project on the grounds that it is too risky for them to take part; 55 or it may be insisted that someone else provides informed consent on their behalf, or in addition to their own decision about whether to participate. Similarly, in the case of external threats, not only could reporting abuse lead to an increased, rather than reduced, threat of harm to the person concerned, it could also breach the commitment to protect privacy, as well as signalling a lack of respect for participants' autonomy, or the belief that they lack resilience, or are incapable of defending themselves. While they may, of course, be vulnerable, here again we are dealing with matters of degree, about which necessarily uncertain judgments must be made, rather than all or nothing certainties. There may also be conflict between a concern with protecting the interests of vulnerable groups and the effective pursuit of social research. One of van den Hoonaard's 56 criticisms of the preoccupation with vulnerability is precisely that it prevents research of particular kinds being done on groups designated "vulnerable," or leads to it being done in ways that are less likely to be successful. 57 This is one aspect of a more general 51 Van den Hoonaard . 52 Levine et al. . 53 Mackenzie et al. : 2. 54 See, for example, Pickering's discussion of heroin users. She also provides an excellent account of the complexities of autonomy and consent in this context. 55 Juritzen et al. : 647. 56 Van den Hoonaard . 57 See also Pickering . 11/16 point: that, paradoxical as it may sound, there are dangers associated with being too ethical, with giving ethical issues too great a priority. 58 The risk here is that it is always possible to talk up the likelihood of harm, and/or its severity, exaggerating the dangers involved. Indeed, sometimes it seems to be insisted that there should be no risk of harm to participants involved in research, that researchers must ensure that this does not occur; or, similarly, that people's autonomy must be fully respected. But these are unrealistic expectations, and if taken seriously they can only lead, ultimately, to the abandonment of social research. 59 It is the distinctive responsibility of a researcher to pursue worthwhile knowledge, in other words that which is of general value, as effectively as possible within appropriate ethical limitations; and risks of harm, to participants or researchers themselves, must be weighed against this. Furthermore, there are often side-benefits of research for participants, from having someone they can talk to in confidence to the provision of various services. 60 To repeat our key point: what is justifiable is necessarily a matter of situated, and uncertain, judgment; which is not to deny that there are better and worse decisions about this. --- Conclusion It should be clear from our discussion that the concept of vulnerability is complex and controversial. There have been disputes not just about whether or not particular categories of research participants should be treated as vulnerable, but even about the legitimacy of the concept itself. We noted that there is a significant fissure in its conventional meaning. In the context of the pre-emptive form of ethical regulation of social research that now prevails in many countries, it is typically taken to refer to the capacity of people to provide informed consent to participate in a research project. But there is a broader, more common-sense, meaning, relating to differential susceptibility to harm. We argued that, in both these cases, vulnerability is a matter of degree, and that the threat of harm in most social research is much lower than in medical research, where ethical regulation was initially, and justifiably, established. Furthermore, it is rarely more serious than background threats that people live with routinely in social life. Vulnerability remains an important concept and researchers must exercise wise judgments in making decisions about how to treat their participants. But it is a feature of the type of regulatory system now in force that it tends to exaggerate the prevalence of the problems with which it deals, partly as a result of the fact that one of the main drivers behind it is the concern of organizations and institutions to protect themselves from legal as well as reputational challenges or financial penalties. We focused our discussion primarily on vulnerability as susceptibility to harm, noting that the latter is not only variable but can also take many different forms. This is illustrated by the wide range of groups that are commonly treated as "vulnerable": they are vulnerable in different ways. We also recognized that researchers themselves could 58 Hammersley and Traianou : Conclusion. 59 Bronfenbrenner : 452. 60 See Hammersley and Atkinson : 68-71; van den Hoonaard : 315. 12/16 be vulnerable to harm while in the field, and beyond. 61 And we highlighted how a concern with protecting people from harm can be in conflict with respecting their privacy or autonomy. We insisted that minimizing serious harm should be the primary ethical concern of social researchers, but that there are limits to their responsibilities even in this respect. Furthermore, there are other sorts of consideration that researchers must take into account in making practical decisions about how to pursue their inquiries, including how research can be pursued most effectively. We underlined the fact that the knowledge produced by research is a public good. We believe the sort of pre-emptive ethical regulation that is currently in operation in many countries is not fit for purpose in the case of social research, and indeed that it can have damaging consequences. 62 In relation to vulnerability, decisions must be sensitive to the particular people and circumstances involved; they cannot be determined by abstract principles or procedures, important though these may be as guides. Indeed, blanket labelling of particular categories of participants as vulnerable undermines good practice in the field: it discourages proper assessments of degrees of vulnerability, as well as of how threats of harm or to privacy should be weighed against respect for competence and autonomy, and against researchers' duty to pursue their work effectively. --- Funding: None --- Conflict of Interest: The author declares no conflict of interest. --- License: This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

This paper examines the concept of vulnerability in the context of social research ethics. An ambiguity is noted in use of this term: it may refer to an incapacity to provide informed consent to participate in a research project, or it may imply heightened susceptibility to the risk of harm. It is pointed out that vulnerability is a matter of degree, and that there are different sources and types of harm, which must be taken into account in any judgment about whether additional precautions are required to protect particular categories of research participants. Furthermore, such judgments must be sensitive to the particular context in which research is taking place. This is one of several considerations that raise questions about the desirability of the sort of pre-emptive ethical regulation that has become institutionalized in many countries over the past few decades, a form that is more appropriate to medical rather than to social research. However, this is not to deny that a concern with the vulnerability of research participants is necessary on the part of social researchers. Furthermore, it must be recognized that researchers themselves may be vulnerable to harm in the research process. Finally, some discussion is provided of the way in which a concern with vulnerability can conflict with other considerations that researchers need to take into account in doing their work. The key point is that vulnerability is a complex and controversial concept, and it requires careful handling in thinking about social research ethics.

Open Access --- AbstrAct Objectives Stigma against mental illnesses is one of the significant obstacles faced by mental health service users and providers. It can develop at a young age and is also influenced by culture. Youths in Southeast Asian countries are under-represented in mental health research, thus this study aims to explore the dimensions of stigma and social tolerance and examine its correlates in the younger, multiethnic population of Singapore. Design An online survey collected data with sociodemographic questions, the Attitudes Towards Serious Mental Illness Scale, Social Tolerance Scale and an open-text question on words or phrases participants associated with the term 'mental illness'. Principal component analysis and multiple regression models were conducted to investigate the factor structure of the attitudes and social tolerance scales and their sociodemographic correlates. Participants Participants included 940 youths aged 14-18 years old who were residing in Singapore at the time of the survey and were recruited through local schools. results About a quarter of the students reported participating in mental health awareness campaigns while nearly half associated pejorative words and phrases with the term mental illness. The Attitudes Towards Serious Mental Illness Scale yielded five factors while the Social Tolerance Scale yielded two. Ethnicity, gender and nationality were significantly correlated with factors of both scales. Chinese youths showed higher sense of 'physical threat' and lower 'social tolerance' than those of other ethnicities. Females showed more 'wishful thinking', 'social concern' and 'social responsibility' towards the mentally ill than males. conclusions The dimensions of stigma and social tolerance are different in Asian cultures compared with Western cultures. Sociodemographic differences in attitudes towards the mentally ill were found among youths living in Singapore. Misconceptions and negative attitudes towards mental illness are common, demonstrating a clear need for effective stigma reduction campaigns. --- IntrODuctIOn Link and Phelan 1 defined stigma as existing 'when elements of labelling, stereotyping, separating, status loss and discrimination co-occur in a power situation that allows these processes to unfold'. Stigma against mental illnesses is one of the significant obstacles faced by mental health service users and providers and can be described as prejudice and discrimination against individuals with mental illness due to lack of knowledge, ignorance or misinformation. Negative attitudes also give rise to the desire for social distance, a form of behavioural discrimination which involves reluctance to interact with persons with mental illness. 2 Individuals with mental illness often feel a sense of low self-esteem/well-being as they are rejected and discriminated against by others due to stigma. 3 Accordingly, they are reluctant to be identified as having a mental illness, resulting in high rates of treatment avoidance. 4 Thus, Open Access stigma is one of the greatest barriers to seeking treatment for those with mental illness. 5 Negative attitudes towards mental illness are influenced by culture and affect people's behaviour differently depending on their cultural background. Singapore is a multiethnic island city-state in Southeast Asia with a population of 5.61 million in 2016. The population comprises three main ethnic groups: Chinese , Malay and Indian , while 3.2% belong to other ethnic groups. 6 A nationwide study of the local population in 2009 titled the Singapore Mental Health Study 7 highlighted the significant treatment gap for mental illnesses in Singapore. Only 31.7% of people with mental illness were found to have sought help. For disorder-specific help seeking, as many as 96.2% of those with alcohol abuse, 90% of those with obsessive compulsive disorder and 59.6% of those with major depressive disorder did not seek help. 8 A follow-up study called the Mind Matters study explored the potential reasons for the large treatment gap. 9 Mind Matters was a nationwide study of mental health literacy conducted in Singapore in 2014 by the Institute of Mental Health with a total sample of 3006 Singapore residents. The findings of this study showed low mental health literacy and high stigma among those aged 18-65 years old. However, the study showed that younger age was associated with better mental health literacy and attitudes towards the mentally ill. 9 10 While this finding is encouraging, this is not to say that younger people have no stigma towards the mentally ill. Attitudes towards various aspects of mental illness, including its conceptualisation and stigma towards the mentally ill are thought to form at an early age. 11 A review of children's attitudes towards the mentally ill suggested that children as young as 5 years old view those with mental illness more negatively. 12 The review showed that younger children show negative attitudes and have less sophisticated conceptualisations of mental illness than older children or adolescents. While older children had better understanding of mental illness as emotional and psychological disturbances, the review suggested that negative attitudes increased with age in both children and adolescents. Studies have also shown that youth are reluctant to interact closely with those with mental illness by indicating desire for social distance. 13 14 Furthermore, a review of the epidemiology of child and adolescent psychiatric disorders reports that many psychiatric disorders can manifest early on in life and negatively affect several aspects of a young person's life 15 including poor well-being, self-esteem, social relationships in and out of school and academic achievement. Stigma further compounds these problems by reducing well-being and acting as a barrier to help seeking. 16 Therefore, mental health stigma affects young adults and youths of schooling age during an important phase of development in their life. Few studies have explored the attitudes of Asian youths living in Asia, with even fewer including Malay and Indian youths. Although previous research has largely focused on youths in Western countries, their findings are limited due to the lack of generalisability to Asian cultures. Asian values of collectivism are associated with higher levels of stigma 17 and cultural factors may affect desire for social distance. 18 Young Asians also live in changing times and cultures and antistigma approaches used for adults or Western youths may not be relevant to them. Studies of Chinese youths in Singapore showed that some youths believed in the Asian physiological explanation of mental illness or attributed mental illness to religious and supernatural influences. 19 These beliefs could be related to some youths' preference for seeking help from Traditional Chinese Medicine physicians. 20 Although participants in these studies also showed beliefs in psychological causes of mental illness and preference for seeking help from mental health professionals, the findings suggest that one's cultural background may influence one's views of mental illness. However, stigma and social distance have not been well studied among youths in Singapore. More importantly, no studies appear to have included youths from the other two main ethnic groups in Singapore-Malay and Indian-and they are under-represented in mental health literacy research. Thus, the aims of the current study were to explore the factor structure of two scales measuring youth stigma, the Attitudes Towards Serious Mental Illness-Adolescent Version Scale 21 and the Social Tolerance Scale. 22 It also aims to examine their correlates in the younger, multi-ethnic population of Singapore and determine the amount of exposure local youths have to mental health awareness campaigns. --- MethODs --- --- Open Access Procedure Approval was obtained from the National Healthcare Group Domain Specific Ethics Review Board and the Ministry of Education to approach youths for the study. Ten schools registered with the Ministry of Education were approached using a convenience sampling method and six agreed to participate in the study. The participating schools included both single-sex and mixed-sex schools based in the North, North-East and Central Regions of Singapore and were a mix of government-run, government-aided and independent schools. To reach our target age group, the schools involved were secondary schools and junior colleges . Of the participating schools in the North Region, one was a mixed-sex junior college and one a mixed-sex secondary school . The participating North-East schools comprised one girls' secondary school and one mixed-sex secondary school . In the Central Region were one girls' secondary school and one mixed-level school comprising both a boys' secondary school and mixed-sex junior college in the southern area. Approximately 2500 students were informed of the study via school notification boards, e-mail blasts and presentations at the schools. Written informed consent was obtained from both the participant and their parent/ guardian before participants were recruited for the study. Consent forms were distributed to students via the schools so that anonymity and confidentiality were maintained. A link to the online survey was sent to participants via their preferred e-mail address, and they were allowed to complete it in their own time in a place that they were comfortable in. E-mail reminders were sent to participants who had volunteered but not completed the survey and those who completed the survey were reimbursed with SG$15 iTunes vouchers. The online survey was designed using the online survey tool, QuestionPro, which allowed the survey to end automatically when the quota of 1000 completed responses was reached. It was launched and completed in 2016 with 1016 surveys started and 1000 surveys completed. The survey consisted of sociodemographic questions as well as scales pertaining to attitudes towards the mentally ill. It also included an open-text question where participants could list words or phrases they associated with the term 'mental illness'. 23 The survey took 10 to 20 minutes to complete. --- Measures --- Sociodemographic data Questions relating to sociodemographic background were included to gather information on age, gender, ethnicity, education level and nationality. Nationality was divided into two groups: Singapore citizens and permanent residents who are defined as citizens of other countries but are permitted to live and work in Singapore. --- ATSMI-AV 21 Scale The ATSMI-AV Scale is a validated 21-item self-report scale that measures attitudes towards mental illness. Responses to statements are based on a five-point Likert scale where one indicates 'completely disagree' and five indicates 'completely agree'. The scale explores perceptions of violence, social avoidance, embarrassment if one were diagnosed as having a mental illness and personal invulnerability to mental illness. Previous research identified five factors comprising threat, social control/ concern, wishful thinking and categorical thinking and out of control. 21 A factor-based scale score ranging from 1 to 5 may be calculated for each factor, with higher scores indicating higher levels of stigma towards mental illness. The first factor titled threat refers to the fear of direct harm to oneself or one's reputation due to contact with mentally ill individuals and consists of six items. The second factor titled social control/concern consists of five items and pertains to concerns about being diagnosed with a mental illness and subsequently labelled by society. The third factor titled wishful thinking refers to unrealistic thoughts towards recovery from mental illness and consists of four items. In contrast, the fourth factor titled categorical thinking refers to all-or-nothing thought patterns towards the concept of mental illness and consists of four items. The fifth factor titled out of control consists of two items and relates to the association between mental illness and deviant behaviour. As the previous study was conducted on a Western population and stigma against mental illness is known to vary across cultures, it was necessary to examine the psychometric properties of the ATSMI-AV in an Asian population. --- Social Tolerance Scale 22 The Social Tolerance Scale measures social tolerance through desire for social distance and social responsibility for mental health issues . Items are rated on a five-point scale ranging from 'strongly agree' to 'strongly disagree'. Higher scores indicate higher levels of stigma towards mental illness. According to the scale developers, the Cronbach's alpha for the scale was 0.87. However, a factor analysis was warranted to verify the factors for the Social Tolerance Scale in the present sample. --- Words and phrases associated with 'mental illness' An open-text question asked participants to list at least three words or phrases that are associated with the term 'mental illness'. For example, a participant may indicate words/phrases such as, 'crazy', 'violent' or 'very dangerous'. A series of multiple regression models were performed to examine the sociodemographic correlates of the ATSMI-AV and Social Tolerance subscales. Data were analysed with the Statistical Package for Social Sciences V.23.0 with statistical significance level set at 0.05 for all procedures. Basic content analysis was used to analyse the qualitative data from the open-text question. 23 The data were coded twice by one researcher to identify the common themes. A chi-square test was used to examine the relationship between self-reported exposure to mental health awareness events and endorsement of the different qualitative themes. --- results The sociodemographic breakdown of the sample is shown in table 1. The mean age of the sample was 15.9 years and 443 participants were female which is comparable to the overall student population targeted . 24 Of the 940 students, only 212 said that they had taken part in a mental health awareness event. Analysis of the qualitative data gathered from participants asked to 'list words they think of when they hear the words 'mental illness'' yielded three main themes: pejorative associations; names of disorders and sympathy. Of the respondents 418 listed at least one pejorative word or phrase such as 'crazy', 'weird', 'scary', 'stupid', 'should avoid' and 'dangerous'. Local slang such as 'siao/gila ' were also found. Names of disorders including 'anxiety', 'OCD' and 'schizophrenia' were listed . Sympathy towards the mentally ill such as 'pitiful', 'sad', 'need love/ care' were also expressed . The top five most commonly used words are listed in table 2. No significant differences in the types of words listed were found between participants who self-reported partaking in mental health awareness events and those who did not. The groups did not differ in pejorative associations =2.25, p=0.13), naming mental illnesses =0.89, p=0.35) or in expressing sympathy =0.15, p=0.70). The first PCA was conducted on the 21-item ATSMI-AV. Results of the PCA with varimax rotation yielded five factors with eigenvalues greater than 1. Two items were weakly correlated and did not load onto any factors; these were excluded from subsequent analyses. The factorability of the remaining 19 items was supported by KMO value of 0.82, Barlett's Test of Sphericity and Cronbach's alpha . The five factors accounted for 49.07% of the variance. Question items and corresponding loadings are presented in table 3. The four items that loaded onto the first factor were related to the perception that mentally ill individuals are threatening . Thus, the first factor was labelled as 'physical threat', Cronbach's alpha . Four items loaded onto the second factor Open Access Responses to the survey questions regarding attitudes towards the mentally ill are shown in table 4. Of the participants, 29.8% felt 'that there really isn't anything called mental illness' and 34.3% believed that there are medications that can help those with mental illness. Seven per cent of the participants had some missing data in the ATSMI-AV. The second PCA was conducted on the 11-item Social Tolerance Scale. Results of the PCA with varimax rotation yielded two factors with eigenvalues greater than 1. The factorability of the 11 items was supported by KMO value of 0.86, Barlett's Test of Sphericity and Cronbach's alpha . These two factors accounted for 53.34% of the variance. Question items and corresponding loadings are presented in table 5. Six items loaded onto the first factor labelled 'social distance', Cronbach's alpha , which was related to negative reactions associated with being in close physical proximity with mentally ill individuals or forming close relationships with mentally ill individuals . The second factor had five items related to providing social support to mentally ill individuals . Therefore, the second factor was labelled as 'social responsibility', Cronbach's alpha . 2.7% of participants had missing data for the Social Tolerance Scale. For the sociodemographic correlates of the ATSMI-AV subscales, only ethnicity was found to be significantly associated with 'physical threat'. Specifically, Chinese students had a higher sense of 'physical threat' than non-Chinese students . Gender and nationality were observed to be significantly associated with 'wishful thinking'. Females had more wishful thoughts than males while permanent residents had more wishful thoughts than Singaporeans. Gender and nationality were also found to be significantly associated with 'categorical thinking'; males endorsed more categorical thoughts than females while permanent residents had more categorical thoughts than Singaporeans. Only gender was significantly associated with 'social construction/concern' where females were found to have greater social concerns than males . None of the sociodemographic variables were associated with 'label avoidance'. Table 6 presents the results of the multiple regression analyses for the sociodemographic correlates of ATSMI-AV and Social Tolerance. For the sociodemographic correlates of the Social Tolerance subscales, only ethnicity was found to be significantly associated with 'social distance'. Specifically, Chinese students had lower tolerance for contact with mentally ill individuals compared with non-Chinese students . Gender and ethnicity were both observed to be significantly associated with 'social responsibility'; males and Chinese students had lower tolerance to engage in supportive action towards others with a mental illness compared with females and non-Chinese students, respectively. --- DIscussIOn Overall, a large proportion of the participants appear to have misconceptions towards mental illness. The PCA of the ATSMI-AV yielded five factors, 'physical threat', 'wishful thinking', 'social concern', 'label avoidance' and 'categorical thinking'. The PCA of the Social Tolerance Scale yielded two factors, 'social distance' and 'social responsibility' though the items in each factor was not exactly the same as that used by Koller et al. 22 Significant associations were found between subscale factors and sociodemographic factors. While nearly a quarter were able to volunteer names of mental illnesses and express sympathy for the mentally ill, negative views were most predominant with almost half the sample associating negative words with mental illness. The proportion of participants who listed negative words was marginally smaller in those who reported attending mental health awareness campaigns , but the differences in pejorative associations, naming of mental illnesses and expressing sympathy were not statistically significant. As it is unclear what kinds of mental health awareness campaigns were attended by self-reported attendees, further research into the efficacy of youth mental health campaigns is necessary. 'Depression/Depressed' was the most commonly listed word which indicates that youths were familiar with depression as a mental illness. The words 'crazy' , 'different' and 'weird' were the next most commonly listed words, supporting the notion that youths still make negative associations with mental illness and educational campaigns are needed to address this area. The fifth most listed word was 'autism/ autistic' which also highlights that youths were familiar with autism as a mental illness. With negative views of mental illness, it is not surprising that about half of the respondents said they would be embarrassed if they were diagnosed with mental illness. Nearly a quarter said they would not want others to know if they had a mentally ill relative. Around a third also said their friends would see them as weak if they had a mental illness. This suggests that mental illness is seen as a mark of shame among local youths and something that their peers would stigmatise. This is of concern as fear of stigma is thought to play a key role in the large treatment gap found in the adult population, 8 and this could also present as an issue in the youth population. Besides avoiding treatment, youths may also lack social support if they find mental illness to be an embarrassing or taboo topic. This fear of others knowing about one's mental illness may be linked to collectivist Asian values 17 and while changing cultural values is not plausible, changing the view of mental illness as a mark of shame may be important for interventions. Despite this, more than 80% of students said they 'would visit a classmate in hospital if they had a mental illness' , 'tell a teacher if a student was being bullied because of their mental illness' and 'stick up for someone who had a mental illness if they were being teased' . Only 3.5% had the misconception that mental illness is contagious and approximately half were open to volunteering for mental health-related causes. While these positive findings are encouraging, there is room for improving mental health knowledge and reducing stigma among youths-possibly with the help of campaigns targeted towards youth that are age and culture appropriate. A review by Corrigan et al 25 suggests that education-based interventions are more effective in changing attitudes and behaviour in adolescents than contact with a mentally ill person, though both methods have significant effects. In-person contact appears to be more effective than video contact as the former yields significant changes in both attitude and behavioural intention while the latter only creates change in attitudes. However, this difference may be negligible as another review by Mehta et al 26 suggests that social contact in interventions only affect short-term outcomes and show little to no effect in the midterm or long term. The review found that mental health awareness interventions generally had a medium-sized effect on knowledge outcomes and a small effect on attitudinal outcomes in the midterm to long term. 26 Considering the conservative local culture of Singapore, gaining participant and/ or parental consent for youths to attend interventions with social contact may be challenging. In light of these challenges and the lack of evidence for longer term benefits of social contact, education-based interventions may be the most practical approach for local mental health awareness campaigns with video-based contact if needed. Emphasis should be placed on developing the educational Open Access aspects of awareness campaigns, both for longer term outcomes and to target the low mental health literacy in the local population. 10 Interventions should also be run regularly or even as part of school curriculum as opposed to isolated events to ensure that the effect is maintained. Although challenging, efforts should be made to measure the long term outcomes of these interventions. Dedicated experimental studies should be conducted to ensure that interventions are locally relevant and have the desired outcome. Future nationwide studies such as the Singapore Metal Health Study on prevalence and use of mental health services 7 and Mind Matters study on mental health literacy 9 may be replicated and include components that monitor stigma, mental health literacy and the treatment gap at a population level. To our knowledge, the ATSMI-AV 21 has not been used in the local sample; thus, factor analysis was conducted to explore its components. The items that loaded onto 'wishful thinking' were identical to those by Watson et al. 21 'Social construction/concern' was also the same except for one item 'I sometimes worry that I may have a mental illness' which did not load onto any factor in the analysis. Two separate factors were related to threat in the sample though the items fell under a single overarching factor of threat by Watson et al. 21 The remaining items in the out of control and categorical thinking factors in the study by Watson et al 21 were combined into one factor in this sample as 'categorical thinking'. The factor analysis of the Social Tolerance Scale also produced a slightly different factor structure from that suggested by Koller et al. 22 The authors suggested seven items for social distance and four for social responsibility. The two factors found in our analysis were similar except for one item which loaded onto the 'social responsibility' factor instead of 'social distance', unlike the findings by Koller et al. 22 This difference in factor structure may be indicative of cultural differences in social norms between Singaporean and Western youth populations and that social distancing may thus present differently. As the sample largely consists of youths from the main Asian ethnic groups in Singapore , further research in other Asian samples, including adults, is necessary to investigate if cultural differences are truly present. In terms of the sociodemographic correlates, ethnicity was correlated with one of the 'physical threat' stigma subscale factors and both subscales for social tolerance. While collectivist Asian values are linked with higher levels of stigma, 17 it appears that there are differences between the Asian ethnic groups. Those of Chinese ethnicity felt more 'physical threat', had more desire for 'social distance' and less 'social responsibility' than those of other ethnicities. Some of these findings are similar to those found in the adult population 10 where Indian and Malay participants scored lower on a social distance measure despite having higher personal stigma. Corrigan et al 27 found that individuals from minority ethnic groups are less likely to support prejudicial attitudes about mental illness. They postulated that this was because people from minority ethnic groups experience mental health stigma more harshly than those in the majority group 28 and appear less likely to endorse prejudice about mental illness. 29 Another possible explanation is the Chinese concept of 'face' which describes a person's moral standing in society. Having a mental illness may be a mark of 'losing face' which can greatly affect one's access to social capital and bring shame to oneself and one's family. 30 Although collectivist constructs of 'face' exist in many Asian ethnic groups, some researchers argue that the Singaporean concept of 'face' presents uniquely to the country and that Chinese Singaporeans place more emphasis on 'saving face' than the other ethnic groups despite all participants residing in the same country. 31 Those of Chinese ethnicity may thus feel more threatened by mental illness as a mark of shame, which in turn leads to greater feelings of physical threat, more desire for social distance and less sense of social responsibility towards the mentally ill. Gender differences were found on three ATSMI-AV subscales and for social responsibility. The gender differences for ATSMI-AV subscales in our sample were different to those found by Watson et al. 21 They found that boys scored higher on threat, a finding that was not replicated here. However, males in our sample did endorse more 'categorical thinking' and Watson et al 21 had similar findings with males scoring higher on categorical thinking, a factor which shared some common scale items. Males also scored lower for 'social responsibility' than females, suggesting they were less likely to endorse supportive action towards the mentally ill. Conversely, females showed more 'social concern' and 'wishful thinking' than males. Ng and Chan's 32 study on Hong Kong secondary school students revealed similar findings with females showing higher benevolence towards the mentally ill and males showing more stereotyping, restrictive, pessimistic and stigmatising attitudes towards mental illness. Female adults in Singapore also scored lower for stigma in previous nationwide studies. 10 Differences between youths with Singaporean citizenship and permanent residents were also found. Permanent residents endorsed more 'wishful thinking' and 'categorical thinking'. A possible explanation for this finding could be that permanent residents may constitute a mix of different foreign ethnicities including those of the main Singapore ethnic groups. Thus, permanent residents may be classified as a heterogeneous group but actually have diverse views. The findings of this study should be considered in light of its limitations. First, the voluntary nature of the study may create bias in the responses as students who refused to participate may hold more prejudicial attitudes towards mental illness. Second, despite the anonymous nature of the survey, the participants may have shown social desirability bias when responding to the questions, particularly if they felt that their schools had access to their data. --- Open Access Lastly, the schools were not sampled across all regions and were not representative of all schools in Singapore. Despite the limitations, the present study has notable strengths which include the large sample size and inclusion of the different ethnic groups which was absent from previous studies involving Singaporean youths. It is also one of the few, if not the only study thus far to investigate the constructs of mental illness stigma and social distance in youths residing in Singapore. Although not all regions of the country were covered, Singapore is a small city-state and students were recruited from schools across three of the five regions of Singapore. Furthermore, the sample was similar to the overall target population of school going youth in terms of age and gender distribution. These strengths help increase the generalisability of the findings. Overall, a significant proportion of youths in Singapore report having little education about mental health which could explain the stigma endorsed by the sample. Future research could use qualitative methods to understand the construct of stigma better in the Asian context, particularly in light of changing values in younger generations. This study could be replicated in the future to evaluate the effectiveness of public education campaigns when they are launched locally. contributors SP and JL are joint first authors. SP, JL, MM and BYC conducted the fieldwork. Analysis was done by JL. SP led the project. SP, MM, BYC, SS, SPL, JAV, EA, DSSF, SAC and MS helped to design and deliver the project. All authors revised and approved the final version of the manuscript.

► This study is the first to assess stigma towards the mentally ill in a multiethnic sample of youths residing in Singapore. ► Students from six schools in three different regions of the country were included in the study. ► The study highlights potential misconceptions Southeast Asian youths have about mental illness which should be addressed in mental health awareness programmes. ► Due to the voluntary nature of the study, the attitudes of non-responders towards mental illness may be different from that of participants and was not captured.

Introduction The impact of sociology on the level of preschoolers' motor skill developed is very important. Preschool children are the cornerstone of future social growth, and their physical and locomotor skill development has far-reaching implications for personal and social development [1][2]. The issue of the sports skill developed by children of pre-school age has gradually received widespread attention with the socio-economic development and the improvement of living standard [3][4]. But in fact, many children's motor skill development level is still insufficient, which is closely related to many factors such as family, school and social environment, it is need to study the effect of sociology on the motor skill development level of preschool children. The physical and mental development of preschool children, as the mainstay of our future society, is of even greater concern. E. Kipling Webster examines associations between basic motor skills, bodily mobility in preschool children. The main study reports baseline cross-sectional data from 126 children with intact basic motor skills and surface timed data [5]. Kara K. Palmer examined the impact of a 5 week long locomotor skills intervention on preschool children's motor skill abilities and their motor behavior while participating in a locomotor skills intervention or outdoor free play. All children showed significant improvements in locomotor skills from essentially baseline to the posttest, and children in the locomotor skills innovation demonstrated higher rates of variability and better scores on all locomotor skills at the posttest when measured against a larger set of skills [6]. The goal of the Thanda Aye was to assess and explore gendered disparities in the development of gross motor skills in 5-year-old Japanese children. Sixty healthy 5-year-olds were recruited, and the majority of subjects were at average levels of overall athletic skills. Boys had clearly greater skill in control of objects [7]. Ruri Famelia investigates the feasibility of collecting data on basic motor skill ability, perceptual motor skills and school sports program for preschool children in Indonesia. He emphasized the imperative of intervention in early locomotor skills and motor physical inactivity and stressed the significance of perceptual-motor competence [8]. The motor skill acquisition of preschoolers is of great importance to their physical and mental health, their learning ability and their social adjustment. The level of motor skill progression in preschoolers is influenced by a variety of factors, with sociological factors being an area of research interest. The motor skills of preschool children refer to the ability to coordinate, flexibly and nervously all parts of the body through movement and training. Luciana Ferreira aims to study how the home setting impacts the athletic growth of children of primary school age, and mediating and modulating analyses have shown that motor development increases with socioeconomic status [9]. Fabio Saraiva Flores provides a descriptive layout overview of the diverse offerings of children's motor development. Prior research on school and physical education environments has not focused specifically on those features of the milieu that facilitate or impede motor learned outcomes, implying that past studies should assess these associations by manipulating features of the milieu in these different microsystems [10]. The development of locomotor skills is essential to a child's quality of life, mental health and future development. However, the motor skill progression among preschoolers varies from person to person, and these differences may be influenced by a variety of sociological factors such as family environment, educational resources, and sociocultural factors. This article helps to improve the quality of family education. Family environment is the first social environment for children's growth. Family education has a significant impact on the way children develop locomotor skills. By analyzing the influence of the home environment on the social activities of children's learning, it can provide parents with scientific educational guidance and help them care more about and support the progress of their musical skills, so as to improve the quality of family education. --- Sociological Factors Affecting the Development of Preschool Children's Motor Skills --- Influence of Motor Skills of Preschool Children Preschool is one of the most critical stages in the human life cycle, a period of physical, intellectual and social development, which can lay the foundation for children's future development [11][12]. Motor skills are one of the most important skills that children need to master at this stage, including physical coordination, strength, balance and flexibility [13][14]. The development of motor skills has a profound effect on mental health, social adjustment and physical well-being of children. Developing motor skills is very important for children's physical health. Preschool children can strengthen their health through physical exercise and prevent chronic diseases such as obesity and cardiovascular disease. In addition, good motor skills can help children participate more in outdoor activities and improve immunity and endurance. The development of motor skills has a positive impact on children's psychological quality. Through physical exercise, children can learn to face challenges and overcome difficulties, thereby developing psychological qualities such as self-confidence, perseverance and self-discipline [15][16]. At the same time, physical activity can help alleviate the anxiety and stress of preschool children and promote emotional stability and mental health [17][18]. The development of motor skills helps to improve children's social adaptability. By participating in sports activities, children can learn to cooperate, compete and communicate with their peers, and cultivate team spirit and social ability [19][20]. In addition, the increased sense of physical achievement and self-esteem helps children to build good interpersonal relationships in society. --- Sociological Factors of Motor Skill Development Family environment is the first social environment for children to grow up, and family members' parental concepts, parenting styles and family economic conditions would affect the development of children's motor skills. Children with better family economic conditions often have a higher level of motor skills development, and differences in family rearing styles may also contribute to disparities in child locomotor skill developed. Therefore, an in-depth study of the effects of the home environment on the desirable behavior of preschool children's locomotor skill is helpful for parents to obtain scientific educational guidance to better focus on and support their children's locomotor skill development. Educational resources are another major player in the effect of movement skill devel-opment in preschoolers. The quantity, quality and equity of educational resources affect the motor skill profiles of children. In areas with abundant educational resources, children's motor skill development level is higher, while in areas with uneven distribution of educational resources, children of disadvantaged groups have limited motor skill development level. Therefore, the study of the influence of educational resources on the motor skill development level of preschool children can provide a strong basis for policy makers to promote the rational allocation of educational resources and ensure that all children have equal opportunities for motor skill development. Social and cultural factors are also an influential role in the evolution of motor skills in preschoolers. Different social and cultural backgrounds have different understanding and emphasis on sports, which affects the evolution of children's locomotor skills. Examining the effects of socio-cultural factors on preschoolers' locomotor skill progression levels can inform policy makers and the community, and promote the formation of a sociocultural atmosphere that attaches importance to physical exercise and pays attention to children's health. --- Sociological Influencing Factors --- Experimental Purpose This paper aims to explore the sociological factors influencing the development level of preschool children's motor skills, in order to provide useful reference for early childhood education and family education. --- Experimental Subjects In this study, 100 preschool children aged 3-6 were selected and divided into 2 groups on average. The first group received targeted motor skills training, and the second group maintained normal living and learning arrangements. The specific data of the experimental subjects are shown in Table 1. --- Motor Skill Test Before the experiment, all the participating children were tested on their motor skills, including basic motor skills and comprehensive motor skills . Basic motor skills and comprehensive motor skills were scored. The full score was 10. The basic motor skills and comprehensive motor skills levels of the children in the two groups were observed before the test and the average scores of the children in the two groups were recorded, as shown in Figure 1. As can be seen from Figure 1, the children's "throwing" motor skill score is high, the test result of children in group 1 is 8.9, the test result of children in group 2 is 8.8, and the children's "balance" motor skill score is low, the test result of children in group 1 is 8.2, and the test result of children in group 2 is 8.1. There is no obvious disparity in the test results for the two groups of kids, which is consistent with the results of the experimental test. Children Coordinate Improve children's overall coordination ability through skipping rope, kicking shuttlecock, and other forms After the training was completed, the basic motor skills and comprehensive motor skills of the two groups of children were scored again, and the scoring results were shown in Figure 2. As can be seen from Figure 2, children's "jumping" motor skills score is higher, children's "balance" motor skills score is lower, group 1 children's basic motor skills and comprehensive motor skills scores have been significantly improved, while group 2 children have no significant change. The average score of the children in group 1 was 9.2, and the average score of the children in group 2 was 8.55, and the children's motor skill scores improved by about 7.6 percent. --- Investigation of Sociological Influencing Factors In order to investigate the extent to which sociological factors affect the movement skill developed by children, the children tested were investigated and the results of the investigation are presented in Figure 3. As can be seen from Figure 3, the results show that family background, educational resources, socio-cultural environment and policy environment all have significant effects on children's motor skills development. Family background had the greatest influence on child locomotor skill to develop, followed by educational resources and social environment, while policy environment had the least effect on child locomotor skill to develop. Children from wealthier families have better access to sports equipment and track resources, which gives them greater access to sports activities and the development of better motor skills. Parents' education level and views on sports play an important role in children's participation in sports and skill development. High-quality physical education programs can give children more access to sports and professional teaching, so as to better develop sports skills. A positive socio-cultural environment contributes to children's increased participation in sports and the development of skills. Good policies can provide better sports environment and conditions for children and promote the all-round development of motor skills. --- Conclusion Motor skill development is critical to a child's emotional, social and physical well-being. The purpose of this paper is to examine the sociological elements that influence the level of motor skill development in preschool children. By analyzing the literature and collecting the data, this paper discusses the various sociological factors affecting the development of preschool children's motor skills. Family environment is a key factor affecting the development of preschool children's motor skills. Factors such as family economic status, parents' education level and exercise habits all have an impact on children's motor skills development. In addition, educational resources and social and cultural environment also have an impact on the motor skill to develop in children.

As people's living quality improves and their concern for health deepens, the spiritual and intellectual growth of children of pre-school age is receiving more attention. This paper analyses the sociological factors affecting the growth level of preschool children's motor skills, points out the influence of preschool children's motor skills, explores the sociological factors of motor skills development, and finds that the development level of preschool children's motor skills is affected by a variety of factors. Through targeted motor skills training for children, it is found that professional training can improve children's basic motor skills and comprehensive motor skills. The experiment shows those children's motor skills score has increased by about 7.6%. Family background has the greatest influence on children's motor skills development, followed by educational resources and social environment, and policy environment has the least influence on children's motor skills development. By focusing on these sociological elements in child sport skill on development, problems in sport skill development in children can be better understood and addressed, providing targeted recommendations for policy development and practice.

Let the expedition begin The DARPA Ground Truth project provided an abundance of novel and unexpected opportunities for participants to select and implement similarly novel methodological approaches. Its stated purpose was to test the viability of using simulated social systems to conduct productive social scientific research. To advance this goal, hundreds of researchers and key support personnel participated in the GT project in various capacities. The project was highly structured in how researchers were organized into groups, how these groups did communicate, what information was provided to each group, and what problems each group had to solve. Otherwise, participants were free to choose how to achieve their goals. See Fig. 1 for an overview. --- Four worlds The formal structure of the GT project, in terms of research teams and their roles, involved the creation of one simulated social system by each of four simulation teams , which our team affectionately titled "Urban World", "Power World", "Disaster World" and "Conflict World". These four "worlds" were then explored and studied by two different research teams The research teams who explored and studied these worlds had no knowledge about the simulator teams who created them nor, for the first year, each other. Our team-one of the research teams-employed a pluralistic approach that sought to engage many possible methods and models. The other research team focused on a mixed methods design, applying and extending advances in causal analysis, conducting sociologically informed modeling, and using agent-based modeling to replicate phenomena of interest. A seventh testing and evaluation team provided guidance to those building the simulations about the substantive features of each world, determined what data could be shared by simulation teams with research teams, and evaluated the performance of both simulators and researchers. Within this structure, T&E was understood to represent the research interests and intended goals of DARPA as the organization funding this research. Finally, an eighth team from the Pacific Northwest National Laboratory was assigned to replicate results submitted by the research teams during the final phase of the GT project. As can be read in other papers of this issue of CMOT, the four worlds were varied in substance and mechanism. "Urban world" was created by a simulation team of quantitative geographers from George Washington University on the street grid of a modern city and was powered by an agent-based model that balanced personal preferences and exigencies of locations, such as work, recreation sites, restaurants and home. Individuals had money, made friends, and could eventually contract disease, and research tasks surrounded the creation of more income equity, more social connections, and lower morbidity. "Power world" was created by an industry-led engineering team at Raytheon in collaboration with social scientists based on principles of social groups and complex collective behavior. Also powered by an agent-based model, Power World involved regional elections and policies, group competition, and group-acquired income. Individuals exhibited levels of happiness, and research tasks involved the creation of policies that increased happiness while achieving particular election outcomes and other group success criteria. "Disaster World" was created by machine learning and artificial intelligence researchers from USC's Institute for Creative Technologies who developed an agent-based model of human behavior in response to risks associated with a hurricane, which was driven probabilistically with partially observable Markov decision processes . Disaster World inhabitants maximized their "reward" by reacting to perceived risks and realized outcomes associated with different courses of action vis-a-vis hurricane impact. "Conflict World" was created by a team from Wright State research institute in 1 3 Does big data serve policy? Not without context. An experiment… partnership with experienced intelligence analysts to generate an agent-based model that reflected a state in crisis, with civil conflict, violent insurgency, food shortages and popular unrest. Their model represented the synthesis of multiple types of perspectives about the state of the world, possible courses of action, and the role of path dependency. Its defining feature was a simulation where actors' preferences and choices responded to a dynamic shared environment. The GT project timeline consisted of 30 months of activities organized around three project phases that each consisted of three tasks. These tasks were formally labeled Explain, Predict, and Prescribe, corresponding to the type of questions or problems research teams were expected to answer or solve. While these tasks were uniform in character across phases, it was understood that simulation teams would make their worlds more complex with each phase. Aside from the common use of agent-based models to generate data, the content and structure of the GT project's four worlds differed vastly and in fundamental ways. For example, agents in Power World sought to maximize their utility, simulating the familiar Homo economicus located in an unfamiliar landscape of opportunity . Conversely, agents in Conflict World engaged in stigmergic behavior , actively shaping and responding to their simulated environment in a manner characteristic of Homo sociologus , an agent driven by personally acquired or cultivated values in a landscape of dynamic, shifting possibilities. This range in simulated characteristics of social systems, from market-based utility maximization to culture-based co-constitution of agents and their environment, was not known to research teams during the GT project but reflects extreme variation in assumptions made about human behavior within social systems. This variety affected not only the properties of agents, such as their motivations, but other properties of the simulations, such as the kinds of social processes that existed in each world. Within this framework, research teams sought to validate existing quantitative tools and test novel methodologies to conduct social science on these simulations, DARPA's stated goal of the GT project. These efforts spanned a range of approaches encompassing traditional social science methods, bespoke and novel methods reliant on sophisticated computational strategies, and bleeding edge neural models from computer science, which we discuss in the final section of the manuscript. Our team also tested the efficacy of crowdsourcing for completing tasks in Phases 1 and 2. This work sought to realize the stated goals of the GT project through any means necessary. It also seemed plausible to our team the GT project might serve other valuable purposes for DARPA, such as the creation of better wargames or the evaluation of standard social science methodologies' ability to identify a novel process not previously found in the real world but built into the simulation. The ambiguity of potential uses for project outputs created a unique situation where decisions by T&E to restrict or allow the flow of certain kinds of information between simulation and research teams had definitive impacts on how and what research could be conducted and, due to these impacts, inspired discussions about the metagame that may be structuring these decisions. This feature of the GT project is notable as most large-scale social science research involves proposing a specific, highly detailed research plan with clear aims that align with the funding organization's goals. While we were provided specific tasks, our performance was evaluated using known metrics, and project goals were documented/discussed explicitly, we perceived ambiguity in project goals which led to real ambiguity about what we believed to constitute productive versus unproductive effort within strict project timelines. The combination of strict timelines, restricted information flows, and the assumed known unknown of the GT project's potential "true" goal presented a novel problem-solving environment that negatively affected our team yet inspired substantial reflection about the basic nature of social scientific research tasks such as survey data collection and ethnography. Along with strict timelines, teams participating in the GT project experienced other constraints which shaped lines of inquiry and thus generation of knowledge about each world. First, simulation teams created initial data for analysis by research teams. In Phase 1 and 2, these initial data were incomplete, intentionally excluding the vast majority of data generated for each world since data collection was a major part of research teams' tasks. To complete data collection tasks, research teams posed well-constructed questions and requests for data. Within this framework, questions were used as a way to learn about the existence of objects in a simulated world, while requests for data, termed "research requests," were used to simulate the implementation of social scientific research methods to collect data for analysis. Submitted questions and research requests had to adhere to strict guidelines regarding their structure, content, and framing. In Phase 3 this constraint was removed, with simulation teams providing all generated data to research teams, precipitating the inverse problem of data management . In both cases, the novelty of the problem-solving environment again inspired self-reflection about the idiosyncratic nature of social scientific inquiry as it has evolved in the real world, particularly its tendency to expect certain inputs for analysis. Specifically, we learned tacit disciplinary knowledge about human behavior led to the expectation that certain features existed in the simulation because they were so "fundamental" to understanding agent behavior or, conversely, certain features of human experience were so basic they could be ignored, both of which revealed how difficult it was for us to decontextualize simulated human behavior. A second constraint faced by simulation and research teams was the fact that the research request system was managed by the T&E team, a third-party mediating how teams interacted. Research requests generated by research teams were expected to reflect actual methods of research used within the social sciences, such as fielding a survey asking a series of questions about agents' age, current status, and preferences. In this context, each world had a temporal component similar to sampling frequency, which sometimes meant that well-formed RRs inadvertently led to the generation of massive amounts of data. For example, Urban World's agents were observed in 5-minute increments over a period of multiple simulated years. Not knowing the temporal scale of -mechanisms that governed these agents' behavior meant it was often safest to generate RRs with maximum sampling 1 3 Does big data serve policy? Not without context. An experiment… frequency. Related, when simulation teams received RRs they had to both interpret the intended meaning of the proposed data request and conform to limits placed on the kind and amount of data they could provide according to the T&E team. This system enabled "blinding" of simulation/research teams by making T&E the intermediary between both. This constraint introduced yet another novel feature of the problem-solving environment in the form of "blinded" data collection which required self-reflection about the biases we held about the data collection process. We found through trial and error that biases about which features to try to measure could, again, turn assets into liabilities, expertise into folly, if we failed to accept certain seemingly natural lines of inquiry were unproductive. Within these constraints, the GT project consisted of three distinct tasks during each phase. Each task focused on producing a particular kind of knowledge: The "Explain" task required research teams to develop as full an understanding of the causal relationships within each simulated world as possible. This task required identification of nodes consisting of features within the simulation, such as agent age and current location, and edges consisting of causal relationships between features, such as the tendency for older agents to visit particular sites. Fulfilling this task required discovering all features within each world and performing an exhaustive search for all causal relationships between these features. The "Predict" task required research teams to make an out of sample prediction of agent behavior-individually, in aggregate, or as an observable property of the simulated world-based on data provided by or solicited from simulation teams. This task always followed the Explain task and so, while identification of additional features and causal relationships sometimes continued, efforts often focused on converting a working knowledge of each simulated world into a model of each social system. The primary requirement of this model was fidelity to the ground truth of each simulated world. Fulfilling this task required discovering methods that could accurately reproduce observed outcomes, such as agent behavior, within each world. Nevertheless, some prediction tasks were counterfactual-if X happened tomorrow, what would happen next year?-demanding a rich understanding of causal relationships that were supposed to be uncovered during the "Explain" task. The "Prescribe" task required research teams to optimize a set of possible interventions for each world to maximize or minimize a specific property of the world . As in the Predict task, team performance was evaluated via out of sample observations: simulation teams implemented the interventions provided by research teams then observed their effects, recording relevant metrics for a period of time following the time series observable by research teams. Fulfilling this task required developing a counterfactual understanding for each simulated world, specifically submission of RRs to provide data on the holistic effect of interventions on various aspects of the social system and development of an optimization strategy. The latter logically focused on the features and causal relationships that most strongly impacted the evaluation metrics before considering secondary and tertiary aspects of the world that perturbed these metrics. Here we report how the structure of the GT project and the nature of its activities contribute to our understanding of knowledge production about social systems via our experiences studying the simulated worlds generated by simulation teams. This is from the perspective of our research team, initially branded as Social Machine Intelligence for Novel Discovery to reflect our commitment to explore the connection between emerging artificial intelligence techniques and familiar social science goals. Many of our insights are best characterized in terms of ontological, epistemological, and methodological issues raised and accompanying lessons learned. We believe these issues are relevant to knowledge production in the social and behavioral sciences broadly, especially within collective problem solving settings where rapid development of effective public policy is the primary goal, such as during the COVID-19 pandemic. --- Ontology of in silico worlds The type and extent of ontological issues associated with the GT project were unexpected. Some issues, such as discovery of features and causal mechanisms, stemmed directly from the structure and stated goal of the project. We briefly discuss these expected challenges but focus discussion on broader questions with substantial implications for collective problem solving. These questions arose from parallels between the GT research setting and real world problems, like political decision making about complex social issues or environmental challenges. We generally conclude that ontological questions about the nature of social systems can pose significant risks to the ability of disparate groups to arrive at a consensus about how to solve complex problems, such as racial disparities in public health and appropriate local responses to a global pandemic. We organize these questions according to the type of "game" being played when we study social systems, the ambiguity around what constitutes a social system, and the process of data collection given limited resources. --- Expeditions into the unknown: a game within a game? Limited initial knowledge about simulation teams' worlds presented a unique opportunity to test the efficacy of different problem-solving strategies. One novel aspect of this extraordinary ambiguity was the fact that we had no sense of the type of "game" being played by simulation teams nor the potential metagame being played by DARPA, though the assumption of an unknown metagame was unique to our team alone. Thus, there was little reason to believe any one disciplinary perspective or methodological skill set would be better suited to achieving GT project tasks than any other. To this end, we tested the efficacy of approaching each task as a lone explorer, as two explorers, and as a team or community expedition into the four worlds. Multiple worlds were studied as single explorer expeditions where we tested if the most effective way to complete a task was for one individual to direct the bulk of our activities. In our team, each world had either one or two primary "explorers" with additional individuals providing support. A single explorer approach meant the use of a novel methodological strategy whose execution was led by a single individual, sometimes supported by others acting at their direction. For example, for Urban World we used Granger causal networks to complete tasks in Phase 1 and Phase 2 . This novel method was content free in terms of being broadly applicable to the type of data associated with Urban World regardless of what these data represented. We found that this flexibility sometimes helped ensure certain tasks could be completed, but that it relied on a small handful of experts to produce results which were not always interpretable within our broader understanding of the world. Single explorers were good at finding a way to play the game using their own rules but the idiosyncratic nature of their efforts could limit the generalizability of their modeling strategy even among those with domain knowledge. Another approach to GT project tasks was to treat these tasks as a two explorer cooperative expedition. A two explorer approach meant that two individuals cooperatively explored and tested different methods for completing tasks. Cooperative exploration and problem solving allowed us to leverage both individuals' perspectives and skill sets to iteratively uncover effective solutions to the problems faced in each phase. We expected that pairing researchers with differing skill sets, differing levels of experiences, or both could be effective for completing GT tasks . Overall, work on Urban and Disaster Worlds supported this hypothesis about the nature of the GT project's challenges and the best way to handle inherent uncertainties about simulated worlds. In each case, insights from both explorers contributed significantly to completing tasks depending on the underlying ground truth of simulated relationships. For example, in Urban World one player was well-versed in GIS and spatial analysis, which proved more salient to performing certain tasks, while the second player was versed in social theory and network analysis, which helped provide initial directions for analysis. We elaborate below how this role-taking worked in practice. We observed that even when empirically grounded, theorybased analysis proved uninformative it still allowed us to rule out common theories of social behavior or common modes of social scientific analysis . Our experience with the GT project suggests that two heads are, indeed, better than one when presented with a complex research task whose parameters are not fully known-which reflects the vast majority of research tasks . The final approach we tested for effectively completing GT tasks was to assume each task represented a team or community expedition. In this context, a team expedition involved many individuals working to independently complete a task using their personal understanding of the task and the most suitable methods for completing the task. The "team" aspect of this approach is embodied by shared striving toward a common destination given identical information and constraints. Team expeditions were operationalized through use of crowdsourcing competitions to solicit solutions to the challenges associated with GT tasks. For Phase 2, we worked with TopCoder.com, which maintains a vast community of "solvers" who engage with data science, programming, and other scientific challenges to win prizes for the top solutions. TopCoder represents a large global network of technologists, boasting approximately 1.5 million community members. Up to 450,000 members of this community are data scientists with backgrounds in computer science, physics, mathematics, and engineering. In Phase 2 of the project, we involved TopCoder in a four tournament Grand Challenge through which we solicited community assistance to: suggest possible social entities, forces, and causal relations in each of the four worlds to stimulate our requests for data, as well as accomplish the explain, predict, and prescribe tasks alongside our teams. We scheduled completion of these challenges a week before each of our solutions were due, with our independent or cooperative explorers intensively engaging with their results to improve our final submissions. Across all three competitions we had 686 registered participants, from 15 countries; 64% from India, 7% from Russia, 5% from Indonesia, 4% from the United States and less than 2% were from Kenya, Iran, China, Philippines, South Africa, Italy, Canada, Romania, Brazil, Mexico and Egypt. For the first of our four challenges, we received a barrage of suggestions that broadened our scope of questions and research requests to simulation teams . These included using more of the research methods available to us, such as experimentation, social media analyses or use of government records, and the proposition of out-of-the box causal possibilities . For the remaining challenges, TopCoder solvers had a more difficult time, at least in part because they had to work with an evolving dataset in a compressed time frame without the benefit of direct feedback on the success of their strategies. We found this problem solving context made the creation of a leaderboard, and the accelerating competition it engenders at the end of a competition, infeasible. This finding illustrates how the features of such contexts shape how effectively crowdsourcing can be leveraged as a form of collective problem solving. Moreover, the tasks were complex and involved many continuously evolving forms of data to produce many different required predictions and prescriptions. We observed it was difficult for solvers to put together a complete solution despite efforts to share code that integrated and cleaned relevant data. In other cases, community solutions misunderstood the data or questions, or added nothing new to own analyses. Other times, community members proposed methods like Bayesian Networks for causal discovery or categorical boosted forests, implementing these methods via working python code that successfully uncovered world dynamics. We then extended these with new data and applied them to those same worlds and others. Consider Urban World's Phase 2 Prescribe task as an illustration of how these different problem solving strategies worked in practice to help triangulate the best possible submission. The goal of this task was to select a subset of 200 agents in the world who will, over a 30 day period and in isolation , collectively exhibit higher average daily friendship network degree over the final week of this period than any other subset of 200 agents subject to the same conditions. As part of this task, we were provided a sample evaluation metric in the initial data package representing the outcome of drawing a uniform random sample of 200 agents. We were then allowed to submit up to four mock test sets of 200 agents where the simulation team would then provide the corresponding value of the evaluation metric based on each test set of agents. Table 1 summarizes the results of these mock tests and relative change in the evaluation metric versus the sampling strategy and metric included in the initial data package. Each of the four mock tests in Urban World's Phase 2 Prescribe task represented a systematic approach to the selection of 200 agents. The first mock test drew a stratified random sample of agents designed to reflect the age, education, and income distributions of all agents, under the belief that these agentbased characteristics, including homophily , might systematically affect friendship networks. The second mock test used a matched sampling approach designed to maximize insight into how other features of the world affected friendship networks, such as the role of location. The third mock test focused on the effect of geographic propinquity on friendship networks . The fourth and final mock test represented an entirely new perspective on the selection of agents. The first and second test sets were drawn based on input from a team member trained in social theory and social network analysis, the third test set was developed between this team member and another team member trained in GIS methods, but the fourth set was generated based on intuitions about spatial processes as understood by the GIS-trained team member. The fourth set was derived by applying a clustering algorithm to the location of agents' homes weighted by the "popularity" of each home . The substantial improvement in evaluation metric over the other three mock tests was exciting and moderately unexpected. However, the overall utility of this clustering approach became apparent when we learned that one of the TopCoder submissions used clustering in a similar yet more intricate manner. This work inspired us to pursue essentially the same strategy used in fourth mock test for our final submission, tweaking our approach based on input from the TopCoder submission that also used clustering. Ultimately, this fine tuning led to a slightly lower final evaluation metric. However, without the TopCoder submission we would have felt more uncertain about the exclusive use of clustering and might have made more substantial changes to our strategy that would have resulted in a much lower evaluation metric. In this case, the two explorers experimented with strategies based on analytic perspectives cultivated in two distinct fields of study, but the value of one explorer's perspective became unambiguous via use of crowdsourcing to externally validate its efficacy. --- Simulated social systems still require real definitions Ontological questions about what constituted a social system presented both the most difficult challenges to accomplishing GT tasks as well as, arguably, the most insight into social science as a practice. The basic ontological questions that plagued both simulation and research teams was "What exists in the simulated social system?" and "How do objects in the simulated social system interact with each other?" when it was made clear to all involved that these simulations were not meant to be realistic representations of social behavior; the GT project recreated the basic features of "first contact" with an unknown civilization . The setting, then, was intended to be functionally similar to a real world social system where individuals' roles, motivations, and behaviors are not known to the observer. Determining what objects exist and how these objects interact was the most fundamental task in the GT project; without this information it is impossible to explain/predict behavior or prescribe interventions. Because GT worlds were simulated social systems whose features were artificially generated and intentionally not reflective of real social systems, and because research teams had to communicate with simulation teams to request additional data, who in turn had to interpret these requests in order to provide the requested data, the interpretation of these requests proved to be fundamental for making sense of each world. Basic ontological questions about what objects exist and how these objects interact were thus complicated by each team's use of words in naming data attributes. For example, the team behind Urban World explicitly renamed variables in Phase 2 to prevent observed cases in Phase 1 where research teams had inferred erroneous information based on variable names . Another ontological question raised by GT project tasks was the overall purpose of each world or, more properly, the win condition of each game embedded within these four worlds. For example, Conflict World presented unique challenges to our team because we assumed that all simulated agents followed a basic utility maximization principle. While more or less true in other worlds, the stigmergic basis of Conflict World meant that agents pursued shifting goals in reaction to their experiences as well as changing motivations in response to these experiences and available options . Put simply, we found that the existence of a stable human nature and stable social system was itself an assumption that we should not have made. Making this assumption led to significant 1 3 Does big data serve policy? Not without context. An experiment… misinterpretation of data and a poor understanding of the world in question, despite success in some tasks. This finding lends novel support to the idea that even certain forms of social science inquiry can be effectively blind to coherent, rule-based behavior if agents' understanding of their world differs from researchers' understanding. Such an idea has gained prominence in discussions about systemic inequality, especially racial inequality, where scholars dispute what constitutes agentic versus structural sources of inequality . Our experience with the GT project indicates it is possible for agents to be embedded in a social structure perfectly navigable to themselves but so alien to scholars that the latter cannot conceive its existence. Moreover, the social ontology of Conflict World was relatively intuitive once revealed, suggesting that access to agents' own understanding of their social system would have been key to understanding it correctly and thus making accurate inferences. --- Data collection versus data generating processes: the language of observation A final constraint that applied across simulated worlds was, again, the role of language, but this time in the context of data collection. Research teams requested information from simulation teams in order to test hypotheses about each simulation. These requests were a focal point of frustration due to the desire to test the full range of social science research methods for studying simulated social systems and the reliance of research teams on the resulting data to both construct a basic understanding of each simulation and complete GT tasks. The former is best illustrated by persistent attempts to employ qualitative methods, in particular ethnography. We suggest that such dissonance between stated GT project goals and the specific goals of our research team reflect distinctions between data collection as often practiced in the social sciences and data generating processes foundational to simulated social systems. In short, it was effectively impossible to conduct ethnography in these simulated worlds. This finding was not self-evident at the time, nor do we believe it intentional on the part of simulation teams or the T&E team. Rather, the natural value of ethnography is its ability to leverage human perception to identify gaps in our understanding of social phenomena , often through discovery of persistent, multi-dimensional configurations. For example, in Urban World if an agent enters a site and we can determine how that agent knows what time of day it is then we may better understand why agents spend less time at one type of site compared to another type of site-an important factor in human behavior exploited by casino designers. An ethnographic account of agents visiting each type of site could reveal that one type of site always has a clock prominently displayed on the wall while the other type of site never does. However, we ultimately learned that agents simply "know" what time it is, and that time spent at sites was an inherent property of site type. Similar instances of perfect information and just-so features of each world could be found in other simulations. For example, in Disaster World, hurricane dynamics were governed by fixed parameters with deterministic impact on the risk posed to regions/individuals. Yet our experience with this ontological question provided an important lesson learned that complemented the lesson learned from Conflict World's radically different ontology compared to the other simulations: Regardless of the type of social system being studied, the accessibility of one type of data does not negate the need for access to additional types. Because we knew the simulations were simulations, we understood that they lacked the complexity of real life, but uncertainty around what was observable compounded uncertainty around the objects and causal relationships being simulated. This doubly shifting landscape is a common feature of real social systems and represents the social scientific equivalent of the "state of nature" where one has no knowledge of what is or is not socially meaningful. Historically, empirical social science was more qualitative in nature and made extensive use of interviews, participant observation, and ethnography. Over time, quantitative social science developed in parallel, emerging as soon as suitable methods were developed . Our experience with the GT project suggests that the initial use of qualitative methods as a tool for uncovering ontological properties of social systems logically precedes quantification of those properties. The GT simulated worlds represented an anachronistic case where we began with quantification then attempted to understand how a social system works in order to develop effective interventions. However, we did not have the benefit of human perception to identify meaningful properties and/or their configurations in each world. Ultimately, this parallels the hyper-quantification of social behavior we observe today and suggests that without the ability to observe the social system in vivo we risk developing a working model of behavior that excludes key properties of the world. For example, in Urban World there existed an entire process for "eating" that we never uncovered because we thought it was self-evident from other behavioral data, but this process proved crucial for understanding simulated disease transmission in Phase 3. --- Ontological lessons learned In addition to the specific lessons learned noted throughout, we identified two broad lessons learned from the ontological questions raised by the GT project. First, successful quantitative social science requires well-posed questions that use welldefined terms. Data does not, in itself, induce understanding, and descriptions of data can impart significant bias, even when it is known that such bias exists and could substantively affect analysis. Second, collective problem solving without a shared ontology about the object of study is extraordinarily difficult. Both have real world implications, but we will focus briefly on the latter as the former may be an artifact of the GT project's design. The lack of a shared ontology concerning what objects exist in a social system and the causal relationships between these objects generated significant challenges for resource allocation. Specifically, cognitive responses to GT tasks required some level of certainty concerning this ontology, but without this certainty many team members could not intuit how to proceed. For example, consistent lack of progress in learning how agents behaved in the Conflict World simulation made it difficult to justify devoting additional resources to associated tasks. Conversely, the expansive nature of the Urban World simulation meant that it was difficult to gauge whether progress had been made at all since we never knew if we had uncovered all relevant properties of the world. Both situations led to ambiguity about the entities and relationships under study which often resulted in pursuing an exhaustive understanding of causal relationships . However, this ambiguity has implications for real world problem solving in the form of issue advocacy and public policy recommendations. A limited understanding of salient social entities limited our ability to conduct experiments within these worlds. Experimentation requires control in the form of sufficient knowledge about relevant forces shaping behavior and the variation of one or a few factors. Without this knowledge, experiments can become fishing expeditions where other research methods may produce more insight with less effort. It was not until we approached the end of the project that we came to understand enough about these worlds to cultivate expectations about which features would be most insightful to systematically vary for the direct identification of causes and useful policies. This delayed understanding, combined with the significant turnaround time of research requests, made experimentation a high-risk endeavor during most of the project. Future simulation studies of this kind might consider how to support a sufficient ontological framework for posing meaningful experiments. The GT project constituted a highly structured yet cooperative team-based assessment of agent based simulations as a platform for productively testing different methods for studying social behavior. Despite prolific uncertainty around the form and content of these simulated social systems, project participants committed to achieving the same goal under the belief that this goal was, generally speaking, beneficial to advancing social science and its applications. Most complex policy debates involve similar levels of uncertainty about the true nature of social problems and policymakers must typically make decisions based on an incomplete understanding of these problems . However, policymakers are also often subject to a barrage of information from issue advocates, popularly termed lobbyists . These advocates advance a particular understanding of a social problem so that policymakers can address this problem in the manner they believe most effective. Yet it is rarely clear which policies will be most effective because advocates must begin with their own assumptions about the social system they intend to influence . Competing assumptions will naturally produce disparate policy solutions. Thus, uncertainty regarding assumptions in real world policy debates, to some extent, mirrors the ontological uncertainty experienced by GT participants. In the case of the GT project, participants' commitment to rigorous scientific inquiry and shared striving toward a common goal were not enough to overcome the effects of this uncertainty in many cases. More broadly, we tentatively conclude that ontological differences in how issue advocates understand the same social problems will constrain the ability of policymakers to identify compromise solutions to these problems. Further, our experience with Conflict World suggests it is possible for good faith actors to be incapable of identifying key features of social systems if they rely on a limited array of evidentiary sources. It is unclear how this issue might be addressed in the short term except to acknowledge its effects. Our experience with the GT project, however, suggests that long term efforts to formalize ontological assumptions about social systems must be supported to avoid inefficient, ineffective, or counterproductive public policies enacted by elected politicians. Because there is much we do not understand about social systems, and representative political systems rely on popular understandings of social problems and their solutions, it is important that social scientists highlight this barrier to consensus building and begin adopting a means of communicating the social ontologies used within their work. --- Epistemology All work that overlaps neighboring fields, such as we occasionally undertake and which the sociologists must necessarily undertake again and again, is burdened with the resigned realization that at best one provides the specialist with useful questions upon which [they] would not so easily hit from [their] own specialized point of view. ~ Science as a Vocation, Weber The type and extent of epistemological issues associated with the GT project were often predictable and provided significant insight regarding fundamental tensions associated with doing social science research versus applying findings from social science research. Issues, such as choice of problem solving strategy for each task, stemmed directly from the structure and stated goal of the project. We do not discuss the specifics of these issues but, as with our examination of ontological issues, focus on broader questions with substantial implications for collective problem solving. These questions arose from the nature of the GT tasks performed during each phase of the project and speak to what Weber termed "science as a vocation" in reference to the external economic forces associated with and the lack of intrinsic value characteristic of scientific inquiry in practice. We thus organize our discussion of these issues and related lessons learned according to each type of GT task. A clear pattern emerged over the course of the GT project around the types of knowledge that were necessary to complete GT tasks. Each phase of the project was identical insofar as research teams had to complete the same three tasks for each world: explain, predict, and prescribe. While social systems differed in their structure and content according to the simulation teams' models, explaining each social system, predicting out-of-sample properties of each social system, and prescribing interventions for each social system represented loosely related yet distinct epistemological goals. The first epistemological goal during each phase of the GT project was to explain the simulated social system. Explanation was defined as the development of a directed acyclic graph constituted by all relevant features and causal relationships between features . Relevancy was defined through prompts-such as "explain how agents form friendships"-that served to anchor inquiries for additional information. This GT task required coping with the ontological uncertainties noted above to produce an acyclic graph effectively reproducing the agent-based rules and/or agent-based model parameter relationships used to generate the simulation data provided to research teams. A defining characteristic of this task was its holistic framing. While prompts provided a way to help research teams begin studying each phase's simulated world, research teams were evaluated according to their ability to uncover nodes and directed edges. This evaluation approach presents an egalitarian epistemology where all knowledge had approximately equal value regardless of the relative importance of individual pieces of knowledge for understanding the social system. This egalitarianism incentivized investigation of potentially second and third order effects that were, at best, tangentially related to properties of interest. Any knowledge was good knowledge. A secondary characteristic of this task was significant slippage between what research teams believed they were investigating and what simulation teams had identified as relevant features. Specifically, the ground truth under study was an assortment of data generated by simulation teams and analyzed by research teams. The terminology used by simulation/research teams was, as noted, illdefined and referred to simulated social processes. To "explain" these social systems research teams had to name causal nodes and provide a written explanation of how each node influences another if a directed edge existed. However, evaluating these explanations required simulation teams to interpret nodes and directed edges according to their internal understanding of the data generating processes they had developed. This situation meant it was possible that research teams could describe a simulated social process but, without further clarification, simulation teams infer research teams were referring to a different simulated social process than intended, a phenomena reminiscent of boundary objects that attract shared attention despite being understood and used in very different ways. Once the ground truth of each simulation was revealed, it was clear that there were many cases where this applied, such as agents' affinities for particular sites in Urban World. While our understanding of site visitation was based on the desire to form and maintain friendships with similar agents, the Phase 2 Urban World simulation replicated this agentbased form of homophily through a process where agents chose to visit preferred sites based on both agents' and sites' characteristics. However, our understanding of the ground truth causal diagram for Urban World's Phase 3 simulation suggests that agents of a similar type formed friendships by choosing to visit similar sites rather than explicitly choosing similar agents. This minor distinction had serious implications for our ability to understand the simulated social system in the predict and prescribe tasks and highlights the fact that simulation/research teams could believe they are referring to the same social processes verbally but are actually referring to different ground truths. Our experience with this epistemological issue suggests that clearly defined, shared referents are vital for constructing an accurate ground truth understanding of a social system, and we recommend future simulations testing the utility of different social science methodologies remove this ambiguity to ensure the robustness of results. Conversely, terminological ambiguity also meant that research teams could refer to features using different words or phrases than simulation teams yet still reference the same ground truth processes. For example, it is unclear how much our failure to identify the nodes and directed edges associated with the process of food consumption in Urban World affected the evaluation of our explanation of this world in Phase 3. Without additional information, this lack of clarity cannot be resolved since we assumed data referring to relevant agent behavior were sufficient to understand the process of quelling hunger and we could not know this explanation was deficient since we, as a research team, did not have access to the ground truth or other feedback indicating that our understanding was incomplete . Our experience suggests it is vital to identify such measurement issues before attempting to construct an exhaustive causal model of a social system. This finding supports multiple arguments made in the social sciences about the role of measurement theory and highlights their applicability to data-driven analysis where measurement is assumed to be error free . The second epistemological goal of each phase of the GT project was to predict out-of-sample characteristics of the simulated social system. For each world, simulation teams provided data for a discrete period of time. The primary goal of research teams was to then make predictions about what happens immediately following this time period. Evaluation metrics were typically based on predictions made over a discrete period of time . This GT task avoided many of the ontological issues noted above since the simulated data was itself the ground truth for each world and research teams did not need to explain the set of causal relationships used to generate predictions. A defining characteristic of this task was a data-centric approach to modeling each simulated social system. Because the goal was to predict outcomes based on the data provided it was not necessary to construct any broader understanding of the social system than was needed to accurately predict future behavior. This feature of the predict task provided both natural scoping conditions for analysis and clear priorities in terms of analytic effort. The most efficient strategy was to identify the primary causal relationships governing the outcome of interest, such as the average number of friendships in the world or agents' responses to a natural disaster. Other aspects of the simulation could be ignored to the extent that they did not affect this outcome. --- 3 Does big data serve policy? Not without context. An experiment… In some ways, the prediction task was the most familiar to our research team members as it most accurately reflects both current expectations around effective data science and historic efforts in the social sciences to identify the features most important for understanding social phenomena. For example, residential segregation is a well-known phenomenon in the United States which could allow simple and often correct guesses about your neighbors' race given your own . Even if predicting your neighbors' race based on your own race is not something social scientists typically do, the fundamental epistemological issue remains reproducibility via prediction. In some ways, data scientists are more familiar with this issue given that many applications of machine learning involve generating out-of-sample predictions. Our experience with the GT project suggests that, when interpretation of features is irrelevant, even basic data science approaches, such as clustering, can yield a huge payoff in terms of predictive accuracy. Notably, however, the explain task and predict task were only loosely related. Knowledge production in the former required a robust understanding of the social systems' ontology in order to exhaustively test possible causal relationships, but knowledge production in the latter only required a "good enough" understanding of each social system to accurately reproduce and then predict agent behavior. Emphasis on explicit directed causal relationships and predictive accuracy by those studying causal inference in the social sciences would suggest both tasks are important, but our experience with the GT project suggests the opposite: Fully understanding how a social system operates does not necessarily reveal which parts of this system are most important for outcomes of interest, and the ability to predict outcomes of interest does not indicate a full understanding of the social system. One can even imagine a GT task between explanation and prediction in which a causal weight is assigned to each edge in the graph. There are three potential implications from this observation. First, if the goal of generating new knowledge is not clear, then it is easy to fall into a trap where researchers examine either a small part of a social system or attempt to fully explain the social system when one or the other will suffice. Second, marginal advances in our understanding of a social system need to be put into context relative to an outcome of interest. If no outcome is specified, then the value of these advances cannot be judged. Third, when the social system is unintuitive, as in the case of Conflict World, then it may be unclear whether a full or partial understanding has been attained. This last implication is the most serious as it means we can have a working understanding of one part of a social system that produces good predictive accuracy but that does not capture key causal relationships operating essentially out of sight. Consider Captain Cook's fateful encounter with indigenous Hawaiians: he believed he understood enough about Hawaiian culture to play god, but was killed when he and his men failed to meet Hawaiian expectations about how gods behave. It was not that Cook could not predict the average reactions of the Hawaiians, but that he did not fully understand the social system generating those reactions . The third and final epistemological goal of each phase of the GT project was to optimize a set of policy prescriptions to maximize or minimize an outcome of interest. As with the predict task, research teams were evaluated based on out-ofsample prescriptions where interventions had been introduced into the simulation as prescribed. The primary goal of research teams was to construct counterfactual predictions of agent behavior based on their understanding of how each world operated. In this respect, the predict and prescribe tasks were tightly coupled, largely avoided ontological issues noted above, and reflected a pragmatic approach to knowledge production. The prescribe task differed from the predict task in that the effects of interventions had to be estimated and, once known, optimized to elicit the "best" possible outcome in each world. This task was akin to data-driven policymaking where the effectiveness of policies are empirically tested after the fact. Given the practical implications for policymakers, the purpose of this knowledge production was clear and the problem well-defined. As with the predict task, however, only a pragmatic understanding of each social system was required to generate high quality prescriptions. Our experience with the prescribe task during each phase of the GT project again led us to conclude that the types of knowledge generated during each phase were loosely coupled, at best. Identifying the optimal timing and implementation of prescriptions for each social system did not require a complete understanding of its causal structure. It also did not necessarily require high predictive accuracy, only high impact prescriptions leading to the best outcomes possible. The latter was illustrated in Urban World during Phase 3 when efforts to develop an accurate predictive model of the simulation fell short of our expectations, yet we had sufficient knowledge of the world to develop effective prescriptions based on little more than logic and basic estimates regarding the relative efficacy of each type of prescription available. Other epistemological issues came to the fore during the GT project, such as what constituted ground truth and how to evaluate it, but we have organized our experience by task since each task had unique requirements that led to fairly discrete forms of knowledge production. Performing well on each task required a slightly different type of knowledge. Due to the fast-paced nature of the GT project, these differences manifested in our problem solving strategies. As noted, the epistemological approaches used for each task were only loosely related: explaining the social system had marginal benefit for predicting behavior, and predicting behavior had marginal benefit for developing an optimal set of prescriptions. While fine-grained causal information mostly distracted from the largest factors impacting outcomes of interest, knowing the largest factors impacting outcomes of interest mostly distracted from the relative impact of each potential intervention and thus the task of formulating an optimal portfolio of interventions. Our experience with epistemological issues during the GT project suggests three possible lessons learned of use to an array of stakeholders ranging from non-profit organizations to academic researchers to public office holders. First, understanding a social system is not the same as learning about a social system. It is possible to learn more than enough about a social system to accurately predict behavior and generate effective policy prescriptions without a thorough understanding of the system as a whole. Yet this pragmatic focus, which creates a natural scope condition for data collection and analysis, can miss pivotal features of the system while still 1 3 Does big data serve policy? Not without context. An experiment… performing well on predictive/prescriptive tasks. Our tentative conclusion is that the need to balance holistic understanding with pragmatic inquiry emerges precisely because the knowledge produced by each strategy speaks to different epistemological goals. Alternatively, it is possible that real world social systems exhibit more tightly coupled relationships between these goals, as in cases of heterogeneous treatment effects that imply a more complex set of causal relationships than originally assumed when designing an intervention. A second lesson learned was that knowledge production in an academic setting, specifically quantitative research in the social sciences, often, but not always, focuses on marginal improvements in our understanding. This focus means an academic mode of inquiry, which involves the demonstration of new social objects, forces, or relationships, will tend to be less effective at generating policy relevant findings. It is not that scholars lack the tools to produce such findings, but that the epistemological goal of social science differs from that of policymaking. However, pragmatic and holistic approaches to policy development and implementation are not antithetical to the practice of academic research insofar as scholars can demonstrate that their marginal improvement in our understanding has a substantive effect on outcomes of interest. In fact, this approach is often adopted as a model for evidence-based policymaking and major grantmaking organizations often demand that proposals explicitly identify concrete implications for policy . A final lesson learned was that, even if knowledge produced by quantitative academic researchers in the social sciences has identified the primary causal relationships associated with an outcome of interest, it is very difficult to uncover this information for reuse. To find and apply this work requires not just methodological competency but also domain knowledge of the field/sub-field and its internal debates, willingness to identify and attempt to overcome disciplinary biases both in the field/sub-field and as individuals, and access to diverse sources of research, including not just a wide range of peer-reviewed journals but also respected, if informal, repositories for research, such as the National Bureau of Economic Research working papers, and high quality studies produced by non-academic institutions run by academic researchers. The amount of time and energy necessary to effectively search the literature is thus prohibitive even for the best trained, most well-read social scientist, suggesting that devolving that burden onto policymakers, which is often the tacit strategy of social science, is unrealistic. In fact, our experience with the GT project suggests that the inability for social scientists to move beyond their own disciplinary biases can lead to significant wasted time and effort when those biases strongly suggest a course of action inapplicable to the situation at hand. --- Methodology With the ontological and epistemological issues above in mind, we now turn our focus to methodology. It was expected that methodological issues would be both prolific and highly productive. We found this to be true but not always in ways we anticipated. The variety of approaches employed reflected a diversity of perspectives about which methods might be most applicable in each world during each phase and task. While not always followed, we found the strategy of working backward from the goal of each task helpful for constraining the set of methods to be tested. When we found a particular method was useful in one world/phase/task we often attempted to deduce its relevance in another world/phase/task. Before discussing an example of each strategy, it is useful to review both the variety of methods applied and the immediate methodological issues they often raised. During the GT project, members of our team applied a menagerie of methods ranging from information visualization and bespoke Google queries to standard statistical models to machine learning approaches for agent-based models of complex systems to emerging artificial intelligence techniques to simple searches of the Internet. These included geographic information visualization, correlation, linear regression, logistic regression, Shapley regression, Granger causality estimation, auto-regression and time series analysis, sparse regression, survival models, markov models, clustering, Bayesian graphical models, decision trees and random forests, ensemble methods , support vector machines, k-nearest neighbors, Hawkes process analysis and simulation of spatially interdependent point processes with probabilistic finite-state machines, agent-based models, probabilistic programming models , along with a wide array of deep learning approaches from standard, feed-forward artificial neural networks and recurrent neural networks , to auto-encoders , sequence-to-sequence neural networks , graph convolutional networks , theory of mind neural networks , and world models . We also used network analysis and methods for analysis of GIS data when applicable. Choice of method was, ideally, based on its suitability to a particular task and/or form of data. We first review general approaches to method selection and their efficacy before discussing three approaches that showed the most promise-Granger causal networks, probabilistic programming, and neural networks. Given the varied backgrounds of team members, bias towards particular habits in problem solving sometimes led to -reliance on a favored method. For example, regression analysis and clustering algorithms were commonly employed if only because they were familiar tools for exploring unfamiliar data. This meant that choice of method was not always ideal given the task/data, but the relative suitability of a method would often quickly become apparent. One clear case of this situation was when we attempted to study a social network in Urban World with more than 1000 nodes using exponential random graph models only to find one 1 3 Does big data serve policy? Not without context. An experiment… of the most robust packages for studying ERGMs struggled to estimate even simple properties of the network . Conversely, there were times when clustering algorithms outperformed more refined models based on our working understanding of the social system. In this regard, pragmatism was the rule and deliberate planning the exception. Related to bias in habits of problem solving, domain specific knowledge affected both the collection and interpretation of data. Each world tended to encompass an overarching set of social processes that could be described in succinct and coherent terms. For example, we came to understand the first simulation as Urban World because it involved agents living in a geographic space and moving between points within this space to achieve a variety of goals, in the process entering and exiting sites where they performed actions according to site type. This conception arose out of our team's intuition about the features within the simulation and their relationship to each other. In response, analysis of Urban World was primarily left to an expert in GIS methods and an urban sociologist. However, the Urban World simulation team was primarily composed of geographers. The result was that Urban World was aptly named, and this heuristic often helped us to intuit the existence of basic features, but that differences in domain expertise between geography and sociology led to focus on different social processes. This disparity was most evident in the fact that agents did not directly pursue relationships with similar agents but formed relationships with similar agents by seeking out sites of shared interest where they could then meet and, potentially, form a friendship. The former is foundational in sociological understandings of friendship formation , while the latter is foundational to geographic understandings of travel patterns . To grossly exaggerate this distinction, we might say those who built "Urban World" viewed it as a social system where people often focused on the process of choosing where to go next while our team viewed it as a social system where people often focused on the process of choosing friends. Misalignment in domain expertise, in this case, taught us an important lesson: Similar understandings of a social system can be equally grounded in empirical research but methodological strategies aligned with the ground truth may better catalyze knowledge production. This illustrates and underscores the No Free Lunch theorems in machine learning proved by David Wolhpert and William Macready , which "state that any two optimization algorithms are equivalent when their performance is averaged across all possible problems" -or all possible worlds. If a method works well in some social world, it will work poorly in another. Our understanding and methodological strategies worked well when they aligned, and poorly when they did not. An alternative to applying domain specific knowledge to study a social system is to adopt a Bayesian approach to knowledge production. Specifically, domain expertise generates strong priors about the form and function of a social system, but it is always possible to consider semi-informative priors to tentatively adopt then test. A significant amount of basic analysis involved this kind of work. This often meant iterative testing of relationships between features of each social system with basic constraints on what relationships were thought to be possible. We found that this could produce results where all features were related to all other features but, when the approach was properly specified, could also help advance our understanding of each world significantly. While making too many assumptions could prove problematic and systematically testing a set of assumptions could prove useful, a third methodological strategy was to begin from the tabula rasa of no assumptions about each social system. Methods that embody this strategy were highly data driven, which translated into a need for high levels of technical expertise unrelated to social science. Under the circumstances, these methods often showed the most promise for use in social science research precisely because they were not beholden to relationships we expected to find and focused on uncovering or modeling the relationships we did find in the data. This strategy was especially useful given the ontological issues we faced. For example, we explored whether we could construct causal predictors from observation sequences of variables alone. Designing an efficient causality test, that may be carried out in the absence of restrictive presuppositions on the underlying dynamical structure of the data at hand, is non-trivial. Nevertheless, ability to computationally infer statistical prima facie evidence of causal dependence may yield a far more discriminative tool for data analysis compared to the calculation of simple correlations. On this line of thought, we devised a non-parametric test of Granger causality for quantized data streams realized from the variations of the observed variables in the world simulations. In contrast to state-of-the-art binary tests, this approach computes the degree of causal dependence between data streams, without making any restrictive assumptions, linearity or otherwise. Additionally, without any a priori imposition of specific dynamical structure, we were able to infer explicit generative models of causal cross-dependence, which may then be used for prediction. These explicit models are represented as generalized probabilistic automata, referred to crossed automata, and are shown to be sufficient to capture a fairly general class of causal dependence . The proposed algorithms are computationally efficient in the probably approximately correct sense ; i.e., we find good models of cross-dependence with high probability, with polynomial run-times and sample complexities. The causality network inferred from this dataset revealed non-trivial relationships, and laid the groundwork for such deep data-driven interrogation of complex social phenomena in the future, particularly in situations where sequential observations on many interacting variables are available. Another data driven approach that enabled the explicit incorporation of social theoretical intuition was probabilistic programming . Probabilistic programming languages allow stochastic elements to be included in deterministic models by treating statistical distributions as objects on which we may perform basic logical/mathematical operations. This functionality further allows us to create a generative model of behavior within which we may embed prior information about the social system . For example, a Bernoulli random variable, such as a coin toss turning up heads, may determine whether a person 1 3 Does big data serve policy? Not without context. An experiment… in Urban World contracts a disease given exposure, and the severity of the disease for that person represents a second random variable, perhaps normally distributed such that it typically causes discomfort and the potential to infect others but not lasting disability. In extreme cases, however, the disease may cause death, or spread much more rapidly than normal, like COVID-19 at a super spreader event . Within a PPL framework, the coin toss determines whether the disease is passed then contraction itself has a chance of leading to transmission and loss of health or loss of life. If we wanted to understand the effect of contraction on severity and transmission, we would observe the distribution of these events, conditional on one another, then tune the probabilities of our program in order to generate the appropriate distribution of outcomes, which would later be available for us to determine whether or not a new disease had emerged from the same "world" as the last. Within the GT project, early stages of Power World used PPL as a means to answer questions about patterns in the data, trying to support or disprove potential hypotheses about the way the world works . A sketch of the overarching structure of the world was built using information provided in briefings and communiques. We knew the general outline of the program we were modeling but only general things about what happened at distinct points in time. We hard-coded the things we did know and set a parameterized distribution over the space of programs consistent with behavior expected at the unknown regions. PPLs are able to search this space of programs to find the one most consistent with observed data. Many of the hidden processes we wanted to model, such as conflict between two groups, had binary outcomes. To figure out which features contributed to a given outcome, we mapped each state of each feature to a value and then used linear combinations of these values to produce the weighting of a coin, the flip of which stood-in for the process we were modelling. This approach of searching through the space of weights for factors helped guide data analysis. For example, if the inferred program for determining outcomes for group conflicts weighed group sizes heavily, then we would know to try to look at the data to see if indeed a large group size was reported shortly before a conflict and whether it was correlated with victory. Once features had been narrowed down to those contributing most strongly to certain outcomes, we could hand-craft competing models that only considered those features. PPLs iterate over a "program trace" or snapshot of various states during the execution of a probabilistic program. However, our program was written such that it could also provide us with a likelihood of actually observing a particular trace. Given that we know that at a particular time the target system was in a particular state, we could force our program to make those same choices and change any other unconstrained choices in our program such that it maximized the likelihood of the observed data. One such choice might be the weight given to a particular feature. The challenge arises when trying to propose a new value for a particular choice, especially when choices are tightly coupled. If a change we proposed to a variable made the trace less likely we were cautious about accepting it. One simple but common problem occurred when variable A only takes on a particular state when the states of variables B and C agree . It is possible to detect such cases and handle them appropriately with PPLs, but it currently requires explicit knowledge of the underlying system as well as expert-level understanding of the MCMC algorithms employed. Furthermore, under certain conditions, probabilistic programs are guaranteed to converge, but they are not guaranteed to converge quickly. Reasonable convergence time comes down to well-designed model spaces informed by knowledge of the target system. In this regard, PPLs may be well-suited to doing social science in domains where experts are readily available, well-informed, and forthcoming about plausible and implausible mechanisms of behavior. A final promising methodological approach we found for studying agent behavior was deep learning in general, and graph convolutional networks in particular . GCNs are a subset of graph neural network models that characterize nodes by including features of neighboring nodes . We also tested the efficacy of attention-based approaches, weighting neighboring nodes according to their "importance" for ego nodes. To capture the time-varying nature of the networks involved we explored the use of long short-term memory propagation for constructing successive GCNs over time , one of many ways GNNs can be constructed to suit specific use cases . Finally, we tested an approach to dynamic graphs designed to reduce computational complexity by focusing on temporal dynamics over node representations and the use of hyperbolic GCNs that preserve scale-free or hierarchical graph structures . The most salient use case for these approaches was Urban World, where all three phases/simulations included multiple dynamic social networks in the form of friendships, work relationships, and site co-location. In essence, GNN methods assign latent states to graph nodes by embedding these nodes in a geometric space. GCNs assign latent states based on the states of a node's neighbours and can be propagated using an LSTM-like mechanism. Indeed the aforementioned methods seem to adequately capture the structure of social science problems in question: Agents can be represented by nodes, their associations by edges, and time evolution corresponds to the evolution of agents' states. These approaches seemed a natural fit during, for example, Phase 3 of Urban World where tasks focused on understanding, predicting, and intervening in disease transmission networks. For instance, we expected that the embedding of agent features using latent states would prevent human bias in feature selection while still retaining maximal information. However, these methods only partially solved the problem of disease evolution because almost all focus either on node state prediction or link prediction separately. Only the EvolveGCN framework purported the ability to do both simultaneously, but this feature was novel and its implementation brittle. In the case of these synthetic worlds, many types/levels of associations existed: node states had discrete properties, and we had to predict the evolution of the whole system, not just the state of nodes, for example. All of these are complications we believe had yet to be addressed by methods at the time but represent ready targets for the future. Perhaps the most notable aspect of applying GCNs in the GT project was their persistent inability to reproduce macrosocial properties of social systems based on 1 3 Does big data serve policy? Not without context. An experiment… microdata. That is, disease transmission is a fundamentally network-based process and so we expected GCNs to perform well when modeling the variety of factors influencing this process. However, we were never able to reproduce the accuracy obtained from applying simple compartmental models of disease evolution, such as the susceptible, infected, and recovered model. These models consist of basic differential equations whose parameters determine the overall distribution of disease states within the population at an aggregate level. We expected that GCNs would at least be able to reproduce compartmental models but, despite significant effort, we found that they could not. Given the structural similarity between the GCN architecture and network-based mechanisms of disease transmission we are forced to conclude that either our implementation of GCN was conceptually flawed, or the use of GCN fails to capture a fundamental property of network evolution. We discovered later that disease spread in Urban World was modeled with SIR-like models and so our models may have been too precise for the coarse-grained spread of disease in data. In either case, GNNs appear to be a promising new method for network analysis in the social sciences but may require further development before scholars can realize their full potential in complex social settings. The methodological issues we faced during the GT project suggested three basic lessons learned. First, the level of analysis and type of social process involved are critical for selecting the appropriate method. This lesson is almost remedial in nature given that it amounts to a reminder to select the right tool for the right job. However, the second lesson was methods that may seem intuitively applicable can fail spectacularly , but that openness to alternative approaches can allow for a process of self-correction. Sometimes the latest and greatest method seems like it should work but does not, and that failure to perform as expected can be useful for thinking about less complex but similarly applicable methods with a proven track record and which still embed unarticulated understandings about the world in question. Finally, we learned that imperfect knowledge about a social system can be good enough to find effective methods. Beginning from a tabula rasa typically does not imply beginning from a state of total ignorance. Rather, acknowledging some level of ignorance can help guide the use of methods that have few if any assumptions and may thus "enlighten" our thinking about a problem. --- Discussion The ambitious DARPA Ground Truth project led to the simulation of four social worlds in which social science could be evaluated in silico. Because these worlds were based on simulations, simulation teams knew the causal ground truth-they had designed the programs themselves-but the research teams did not. Our experience attempting to crack puzzles of these worlds reinforced what AI pioneer Allen Newell stated about research: "You can't play 20 questions with nature and win" . And we couldn't play 20 questions about in silico social worlds and win consistently. Stochastic elements of the simulations resulted in a Bayes error rate or irreducible error far greater than 0, and natural limitations on certain forms of data gathering like ethnography and other qualitative methods in the in silico setting were awkward and limited the context research teams were able to achieve. Nevertheless, we and the other research teams were able to do better than random change on most tasks in most phases, and we improved over time and with additional data. Moreover, by confronting tasks with distinct ontological, epistemic and methodological requirements, we gained deep insight into the limits of quantitative social science, especially with respect to informing social policy. Faced with unfamiliar simulated worlds, we struggled to identify their underlying ontology. This highlighted the crucial role of grounded, qualitative insight from insider views of any social system, which cannot not be substituted with quantitative censuses or digital trace data. Why? Because data labels did not provide enough context. They became boundary objects, passed from simulators to researchers through T&E without a shared certainty of reference. This was not a flawed property of the GT program but reflects the limits of ungrounded quantitative social science-data science-where variable names disseminate with interpretations that shift with context. Without a tighter sense of not only the ontology of GT worlds, but what was salient, we struggled to construct experiments despite their availability as a sanctioned data gathering approach because, until the end, we did not know which critical factors to vary, holding others constant. This underscored the challenges of problem solving under conditions of extreme existential uncertainty that contribute to many complex societal challenges. The policy relevance of quantitative social science is also conspired against by the current epistemic standard for publication. Demonstration of novel entities and causes is expected in science, but this narrow exhibition can work against the ability to make meaningful interventions on problems and propose robust policies-from above or below. Finally, we attempted to use a vast menagerie of methods. Some of the most promising emerging methods included detailed bespoke descriptive data analysis, probabilistic programming, deep neural networks of many flavors, and systems of predictive probabilistic finite state machines, which we developed alongside robust statistical and machine learning approaches to supervised and unsupervised learning. Through this exploration, we learned that imperfect knowledge about the most important factors can be sufficient to generate robust predictions and policies. Moreover, applying competing approaches via distinct subteams, including at one point the vast TopCoder.com global community of program solvers, enabled us to discover relevant structure underlying worlds that singular investigators and methods could not. Collectively, these lessons suggest how different a policy-oriented quantitative social science would be than the quantitative social science and data science most commonly practiced to date. Data science and quantitative social science that serves policy will need to endure more failure, sustain more diversity, tolerate more uncertainty, and allow for more complexity than current institutions are well-positioned to support. formal languages, dynamical systems, and machine learning; aiming to formulate modeling approaches that work in the absence of subject matter experts, hopefully answering questions that we have not yet thought to ask. Ziwen Chen is a MA student in the Computational Social Science program at the University of Chicago. Her research focuses on using large-scale data and advanced computational methods to study complex human behaviors, including urban mobility, digital culture, and business innovation. Hongbo Fang is a PhD student in the Institute for Software Research at Carnegie Mellon University. His research focuses on the empirical study of online collaboration and social capital in the context of online interaction, and aims to understand the logic of internet-based voluntary participation and collaborative behaviors. Before joining CMU, he obtained his bachelor degree in computer science at Zhejiang University in 2019. Anuraag Girdhar is an MA student in the Computational Social Science Program at the University of Chicago. His research uses discussion-based games to explore the emergent properties of social network structure, including political polarization, social theory of mind, and the wisdom of the crowd. Xiaoshuang Jia is a Ph.D. student in the School of Sociology and Anthropology at Sun Yat-sen University. She was a visiting scholar at Knowledge Lab at the University of Chicago during December 2017-December 2018. In her research, she uses computational methods such as machine learning and social network analysis to understand social stratification and social structure. She also explores advanced techniques to do causal inference while taking the heterogeneity of population into account. Chen Liang is a MA student in the Computational Social Science Program at the University of Chicago. She graduated from the University of Michigan with a major in public policy. Her research focuses on using social network analysis and natural language processing techniques to understand the political polarization among policy experts in the United States. John Muchovej is a Research Assistant in the Computation, Cognition, and Development Lab at Harvard, where he focuses on how people develop their common-sense understanding of the world. Previously, he was an undergraduate at the University of Central Florida and an RA at MIT. Past and current research focuses on using computational tools to better understand how we summarize extract intent from language and action observation and how people drastically narrow solution spaces in question-answering and creative thinking. Alejandro Vientós is a PhD student at Rutgers-Newark working in the CoDaS Lab. Past and current work --- 3 Does big data serve policy? Not without context. An experiment… centers around social inference, trust, and cooperation in multi-agent, incomplete information settings. Research interests include the design of games or mechanisms where cooperation is both a stable and optimal strategy and systems for incrementally converging to such strategies in said domains. Meg Young is a postdoctoral fellow at Cornell Tech as part of the Digital Life Initiative in New York City. She uses ethnographic methods to study government technology. Much of her work explores artificial intelligence and public policy, with a focus on public-private partnerships, procurement, and accountability. Her other work explores artificial intelligence research and development through an organizational lens. She completed her PhD from University of Washington Information School. James Evans is Max Palevsky Professor of Sociology, Director of Knowledge Lab, and Faculty Director of Computational Social Science at the University of Chicago and External Faculty at the Santa Fe Institute. He is Editor of the new Journal of Social Computing . His research uses large-scale data, machine learning and generative models to understand how collectives think and what they know. This involves inquiry into the emergence of ideas, shared patterns of reasoning, and processes of attention, communication, agreement, and certainty. Thinking and knowing collectives like science, Wikipedia or the Web involve complex networks of diverse human and machine intelligences, collaborating and competing to achieve overlapping aims. Much of Evans' work has investigated modern science and technology to identify collective biases, generate new leads taking these into account, and imagine alternative discovery regimes. Evans also explores thinking and knowing in other domains ranging from political ideology to popular culture. --- Authors and Affiliations --- Chris

The DARPA Ground Truth project sought to evaluate social science by constructing four varied simulated social worlds with hidden causality and unleashed teams of scientists to collect data, discover their causal structure, predict their future, and prescribe policies to create desired outcomes. This large-scale, long-term experiment of in silico social science, about which the ground truth of simulated worlds was known, but not by us, reveals the limits of contemporary quantitative social science methodology. First, problem solving without a shared ontology-in which many world characteristics remain existentially uncertain-poses strong limits to quantitative analysis even when scientists share a common task, and suggests how they could become insurmountable without it. Second, data labels biased the associations our analysts made and assumptions they employed, often away from the simulated causal processes those labels signified, suggesting limits on the degree to which analytic concepts developed in one domain may port to others. Third, the current standard for computational social science publication is a demonstration of novel causes, but this limits the relevance of models to solve problems and propose policies that benefit from the simpler and less surprising answers associated with most important causes, or the combination of all causes. Fourth, most singular quantitative methods applied on their own did not help to solve most analytical challenges, and we explored a range of established and emerging methods, including probabilistic programming, deep neural networks, systems of predictive probabilistic finite state machines, and more to achieve plausible solutions. However, despite these limitations common to the current practice of computational social science, we find on the positive side that even imperfect knowledge can be sufficient to identify robust prediction if a more pluralistic approach is applied. Applying competing approaches by distinct subteams, including at one point the vast TopCoder.com global community of problem solvers, enabled discovery of many aspects of the relevant structure underlying worlds that singular methods could not. Together, these lessons suggest how different a policy-oriented computational social science would be than the 1 3 Does big data serve policy? Not without context. An experiment… computational social science we have inherited. Computational social science that serves policy would need to endure more failure, sustain more diversity, maintain more uncertainty, and allow for more complexity than current institutions support.

Introduction In France or the United States , users under 50 years old tend to prefer using social network services on their smartphone instead of their computer. These mobile devices have become the keystone of contemporary network societies . This social phenomenon invites us to analyze how mobile social media and social messaging applications lead users to manage their mediated sociability in a specific way. It can be difficult to identify, on mobile devices, what can be considered as a social media application and what cannot, as several smartphone applications integrate social media functions without being real social network services . Therefore, research on the uses of mobile social network services distinguishes two types of mobile social network services . The first type includes the applications developed by major social media companies , microblogging sites , and online dating services . These applications are mobile versions of services originally designed to be used with a Web browser. In addition to these applications, "native mSNS" were specifically designed to manage digital sociability using smartphones' specific technological resources. For instance, these applications make use of the camera and GPS to share location-based snapshots or meet close-by potential partners . These mobile applications are not really usable from a computer, unlike the first ones that were designed as an extension of previously static position usage. Even though these two different types of mSNS led to a significant number of studies in computer science and human-computer interaction, social sciences did not give the same importance to both of them . As a matter of fact, social sciences have mainly studied native mSNS, showing, for instance, how geolocalization technology promotes social encounters . This is due to the fact that those mobile technology functions are truly integrated in these applications and generate new types of social networking practices . Over the past years, research analyzing the specificity of SNS uses on smartphones has mostly highlighted the way these applications can increase psychological addiction to mobile phones . Nowadays, exploring these uses of mobile social network platforms seems to be a real challenge. These platforms tend to complicate mobile phone previous practices and require, to be fully understood, developing new methods adding devices' mobility-related concerns to ego-centered networks approaches. Kuru, Bayer, Pasek, and Campbell tried to measure the differences in Facebook uses between users of the app and those only using the desktop version of the service. It appeared that usage motivations did not differ from one another. However, they showed how mobile uses of Facebook are positively correlated with a more automatic and immersive type of usage. As a matter of fact, Bayer, Campbell, and Ling explained that mobile media uses are based on more automatic cognitive triggers than other media as a result of the wider range of contexts involved. This contribution positions itself in that stream of findings by describing the appropriation and the usage patterns of mSNS in different contexts during daily commutes. It aims to highlight how these smartphone usage patterns are linked to day-to-day urban experiences and the way mobile media uses, with their automatisms, affect the organization of mediatized sociability. To achieve this goal, we developed a research protocol to show how the design of mSNS notifications has to be taken into account in order to understand how these technologies have significant effects on mediatized communications and interpersonal relationships. By focusing on these specific functions, our research questioned the possible increase-in number and frequency-of relational demands due to mobile devices' capabilities and the way they promote digital sociability and social connectedness. --- Theoretical framework --- SNS apps extend mobile phone communicational resources and amplify what Christian Licoppe calls the "connected mode" in the management of distant relationships . The concept of connected presence refers to how mobile technology leads users to interact more often with close others than they did when social encounters were based on face-to-face interactions. In this mode of communication, composed of short and frequent contacts between two persons, what is being said is less important than the fact of keeping in touch. Ling and Yttri identified the instrumental dimension of these modes of permanent connection by showing how this "microcoordination" tends to reinforce social cohesion with a small circle of peers. This trend seems to have effects on social networks' structure, as users tend to strengthen their links with a circle of close friends instead of extending their personal network to new relationships . Smartphones and social messaging apps, such as WhatsApp or Facebook Messenger, create new configurations of permanent connection by allowing multisided interactions that facilitate task-based chat groups. They lead users to develop collective forms of "microcoordination 2.0" with a large circle of relations . The authors suggest that, with instant messaging apps, microcoordination that developed through SMS or mobile phone calls is no longer a specific trend of close relations as it used to be. Microcoordination 2.0 broadens to develop within a variety of groups. Therefore, these apps create new forms of microcoordination by increasing social connectedness, especially because users check incoming messages frequently to avoid being frustrated by missing a piece of information in these multisided interactions. These "checking habits" and the way they can be amplified by social media apps is an important issue to highlight the configuration of smartphone usage patterns because they "can act as a 'gateway' to other applications, leading to other actions being taken with the device" . This research proposes to better understand the configuration of these "checking habits. " Böhmer, Hecht, Schöning, Krüger, and Bauer showed that these habits are central because the average duration of smartphone usage did not exceed 5 seconds in 50% of the sessions in their study. In that research, what participants checked most often were notifications of incoming messages displayed by communicational functions . These uses have been shown to be the most important in smartphone usage patterns . We develop a videorecording protocol in order to analyze in detail the spatio-temporal organization of smartphone usage patterns. We focused on how users manage their smartphones during daily commutes in public transport to go to work, home, or move around in urban environments. This method allowed us to record the small amount of extended sessions during which at least three applications were opened. Therefore, the objective here is not to quantify the frequency of "checking habits." We will analyze usage patterns by showing how this permanent attention to the notifications of incoming messages operates at the beginning of most sessions, even during the longest ones, and how this habit acts as a gateway by projecting specific "next actions" that are, themselves, rather polarized around the management of social demands. These habits are obviously completely linked to technologies themselves, and especially the way users mobilize their various functions to manage their availability. As shown by Rainie and Zickuhr , users always turn on their phones to maintain a constant connectivity during daily commutes . As it is important to users to have access to notifications in these contexts, they adjust the ringer mode of their smartphone so they are able to "noticing important notifications" while "avoiding disrupting the environment" . In other words, they keep their availability constant by checking the visual notifications displayed on the screen. We know how the design and "metrics," pushed for example by Facebook, preconfigure sociality and encourage users to "desire for more" likes, comments, or friends . Therefore, the appropriation of mSNS and social messaging apps needs to be analyzed as the product of both users' specific ways of managing their relationships and mSNS design that shapes mediatized sociabilities. The notion of "connection cues" gives important insights into the effects of mobile phone design and, more generally, the role of social connectedness in daily life. To illuminate nonconscious triggers to check mobile devices, this notion identifies and separates three types of media triggers: Technical cues refer to the explicit notifications and signals that come directly from a mobile device . Spatial cues refer to triggers that occur in the surrounding environment of the individual . Mental cues refer to triggers that arise from the individual's internal cognition . This approach helps to understand how appropriation of mSNS design-such as notification push-and the way smartphones' interfaces display social demands are managed by users according to societal expectations , users' automatic behaviors, and technology itself. These behaviors are deeply connected with "spatial cues," that is, concrete situations in which people are engaged and the way these situations frame their mobile phone uses. This notion of connection cues is based on a representational view of contexts that takes "content" and "organization" of smartphone uses as a whole to describe the configuration of contexts and the way they lead people to perform their activities . Therefore, we contribute to analyze a specific set of connection cues by focusing on mSNS uses during daily commutes. Based on a video-ethnography of smartphone uses in situations of mobility, this contribution will describe how users tend to manage a specific set of technical cues: smartphones' notification functions and their display of social demands. This will lead to highlight how these technical cues can frame and amplify social connectedness. --- Method and dataset While many sociological investigations have studied social network sites' uses, they have rarely investigated how mobile social media and social messaging applications lead users to manage their mediated sociability in specific ways. In contrast, mobility is what we focus on. As we wanted to study the whole spectrum of social media uses, all mSNS have been taken into account as far as they were used by the study participants. We also aimed to understand how mSNS complement uses of other mobile phone functions by positioning those uses within a more general analysis of smartphone usage pattern. This was done by conducting interviews during which participants detailed their uses, the way they manage online sociability, and how smartphones renew their communication practices. However, even though this declarative approach is valuable to probe users' reflexivity, it does not give access to the usage patterns themselves, nor does it help in understanding the way mSNS design concretely shapes digital sociability. These patterns may be partly identified by means of sensors installed on participants' phones . This type of data describes the temporal organization of distinct mobile phone functions . For instance, it has been shown that people often start by using Internet before they start sending SMS, etcetera . Yet, besides these investigations' general focus on usage patterns, we consider it relevant to complement their findings by collecting fine-grained data in order to highlight the way mSNS uses are concretely performed according to urban settings. We achieved this goal by collecting data via first-person video recordings instead of using sensors. This video-ethnography "on the move" was produced by asking users to wear camera glasses and record their uses for a week, especially during their daily commutes. While such portable set-up may provide rich and detailed data about the "natural" uses of smartphones, as in the case of mobile map-enhanced walks in urban public places or in museums , it also gives access to all the information displayed on the screens. To be able to fully analyze smartphone usage patterns, we had to synchronize contextual recordings made with camera glasses with video recordings of screen activities collected with a specific screen-recording application . Thus, video-ethnographic data presented in what follows is based on the combination of context-oriented recordings and smartphone screen captures . Hence, what these double recordings reveal is both on-screen content as well as gaze switches, that is, when users look towards or away from their smartphones. Such gaze switches are interpretable as switches between attention to the smartphone and attention to other meaningful aspects of the surrounding environment. Such data, therefore, provide a rich source of information on the way smartphone users manage multiple engagements in public settings. By using this empirical data on the occurrence and direction of gaze switches, the video recordings allow us to account for the temporal organization of smartphone usage patterns in situations of mobility and how these patterns are organized according to spatial cues related to urban settings. Our research protocol, therefore, is an effective practical way to collect contextual data on the uses of smartphones and mSNS apps in order to analyze the organization of connection cues and, more precisely, the articulation between technical cues and spatial cues. Through this video-ethnography we collected 110 video recordings, representing 42 hours of smartphone usage in public settings. We asked participants to wear camera glasses and record their mobile phone activities during their daily commute for a period ranging from a week up to 10 days. Each participant was asked to record 10 sessions. The quality of an important part of the sample was not good enough to exploit and analyze, thus resulting in a total of 110 video recordings. The average duration per recording is 20 to 30 minutes, which represents the average time spent on public transport in Paris or Toulouse. To respect the users' privacy, they could decide when to record their uses and they could also stop or delete them as they wished. These options, however, constitute a strong limitation since they prevented us from having complete control over the constitution of the sample. After retrieving the recordings, we conducted interviews during which users were asked to comment on significant sequences, sharing their interpretations about their uses and especially mSNS. The method, therefore, is a double mixture: a mixture of speech and pictures, with pictures being themselves a combination of a contextual and general view on the one hand, and an in-depth and precise view, on the other hand. This video protocol was used with 20 users, 10 women and 10 men between 18 and 35 years old, living in Paris or Toulouse, who frequently use mSNS-Facebook in particular-in mobility settings. The 15 other participants of our total sample only agreed to answer to the interview part of the protocol; thus, they could not be included in our video sample because they refused to wear camera glasses in public spaces. However, these interviews allowed us to better understand the uses and "checking habits" related to mSNS. From these interviews and the overarching analysis of the video recordings, this first section of our findings will highlight the organization of smartphone usage patterns and how mSNS are used during daily commutes. We will report several social media apps' triggers before focusing on a specific technical cue: notification functions and the way users tend to manage the social demands they relay. This moment-by-moment video recording analysis will allow us to show how usage patterns are often organized-especially during session openings-according to social demands displayed by mSNS and the way this trend can amplify social connectedness. The second section of the findings shows how these patterns are also organized according to users' social interaction management within public transportation settings. Participants coordinated their gaze switches and distributed their attention between the screen of their mobile phone and the surrounding environment. This video recording analysis reveals how temporal organization of smartphone uses in urban environments is articulated with the "temporal parameters of interpersonal observation" . The section "Spatial Cues: How Usage Patterns Are Formed According to Public Settings?" will extend the section "How Social Demands Shape Smartphone Usage Patterns" by focusing on spatial cues related to urban settings in order to highlight how technical cues-such as notifications or loading times-change the way users manage visual forms of social coordination in public settings . --- Findings --- How social demands shape smartphone usage patterns This video-ethnography aimed at collecting an audiovisual sample to analyze social media app uses during daily commutes. As outlined in the previous section, fine-grained data from this study will complement previous work based on large-scale smartphone usage analysis using sensors . It seems relevant to start by comparing that quantitative data with the qualitative data produced here based on a smaller sample. mSNS notification functions in smartphone usage patterns. In order to build the aforementioned comparison, audiovisual sequences were represented in histogram charts. Each bar stands for a session; each distinct color symbolizes a type of application or activity. Total session duration is indicated at the bottom of the bar. For instance, Figure 2 displays Emilie's eight recorded sequences. This 30-year-old woman lives in the suburbs of Toulouse and works in a notarial office located downtown. She takes a 40-minute bus ride to get to her workplace and uses various smartphone applications as pastime during her commuting. Since she uses outdoor public transportation, mobile networks are perfectly available unlike in underground transportation where users must configure their smartphone usage patterns to bypass connectivity issues . This example is interesting because Emilie can freely organize her phone uses. First, it is interesting to point out that the average session duration in this context is 18.14 minutes. If we compare this result with that of Böhmer et al. , in which the average duration did not exceed 5 seconds in 50% of the sessions, the sessions in our sample were clearly longer. Compared to the overall duration of smartphone use during a day analyzed by Böhmer et al. , our sample focus on the small amount of extended sessions during which at least three applications were opened. This was accomplished by requiring participants to start recording before engaging in this specific type of session. Therefore, this research will specifically document these longest chains of app usage. However, it is relevant to hypothesize that these longer chains only prolong the usual opening patterns observed more broadly during shorter chains. It actually appears that most participants reproduce the same usage pattern during the first phase of their sessions, whether they are brief or long. Our results extend studies showing how users generally start by one of the smartphone's communicational functionalities . This trend is exemplified in Figure 2, which shows that Emilie starts all her sessions by opening the Facebook app and by reading the notifications it displays . These notifications are usually checked at the beginning of sessions, as users start by checking them in 53.4% of the sequences. They are mainly notifications displayed by the Facebook app and Snapchat . Notifications of other SNS or microblogging apps are secondary in this sample . The way users deal with notifications and the role they play in the organization of relational practices clearly appeared during interviews, as seen in the following dialogue with Sarah, a 29-year-old executive assistant: Researcher: When you use the Facebook app, what section do you visit? Sarah: Thus . . . notifications and, afterwards, posts. After that, I quickly read news published by my friends or the group pages I subscribed to. Sometimes, if I'm interested, I can comment on my friends' posts but it's often brief, less than a minute. It is interesting to note that she starts using Facebook by checking notifications and immediately explains how this is usually very brief. Mark, a 24-year-old student, does the same and checks Facebook notifications constantly, opening short sessions in various contexts to follow new posts published by his friends: "When I go on Facebook [using his smartphone], it's very short. I check, I leave. I check, I write something, I leave. So in an hour, I can go 4-5 times but it's never longer than 3 minutes." Users appropriate notification functions to continuously monitor relational events published on social network sites. These "checking habits" reveal an intensification of "connected presence" first configurations, characteristic of basic call features of older mobile phones . The three-step opening sequence of smartphone usage patterns. To understand this phenomenon, let's take the example of Caroline. This 21-year-old student takes the tram several times a week to go to the university. During the interview, she described a video recording and explained how she started using her smartphone by opening the Facebook app, among the variety of available applications, because three notifications were displayed on the Facebook logo as well as an incoming message displayed on the Android top bar, on the upper part of the screen . She pointed out: "I went on Facebook because there were these two things there: there were notifications on the 'F' of Facebook and, here, I got a message. I wanted to see what it was." The app starts displaying comments of a previous post . Various notifications are displayed at the top of the screen: there are three new posts in the news feed, three incoming messages, and three recent notifications. Among these types of notifications, Caroline starts by checking the messages and opens one of them . Figure 6 shows a message containing a request sent to Caroline by a friend. They are working on a presentation together and Caroline is on her way to the library where they are going to work together. Her friend asks: "Can you find a way to shorten your intro a bit? I think it is too long." Caroline answers: "Are you sure about the 4 to 5 pages length? It isn't much but, yes, I'll have a look." She responds quickly to her friend even if they are about to meet. Her answer highlights her availability and commitment to their common activity. Thus, the description of this sequence shows how Caroline starts using her smartphone by reading private messages. In other words, she favors notifications that display the strongest and most important social demand and call for a short-term response. Our data generally describe the way smartphone usage tends to be organized according to a specific technical cue: notifications. This usage pattern is composed of three sequential steps: users prioritize applications that display notifications; they especially favor those involving social demands; they choose certain notifications over others according to the social status and type of relationship and their representation regarding a legitimate response time that may be inferred according to specific relational contexts. The first section of these findings highlights how notifications and social demands made visible by mSNS app design renew mediatized interactions by promoting social connectedness. The next part of this article will continue to explore this social phenomenon by describing how the pervasiveness of notification functions also changes spatial cues, and more precisely, visual forms of social coordination in public space . --- Spatial cues: How usage patterns are formed according to public settings? By providing evidence of the way smartphone usage patterns are strongly dependent on notifications, we extended previous research showing how planned or passive alerts change usage sequences and encourage users to open apps, especially those that convey social demands . These previous studies also analyzed smartphone usage patterns and the way users manage notifications but by analyzing data collected using sensors. These data do not provide accurate contextual information about the way participants adjust their smartphone uses to contextual settings . Neither do they allow understanding how different contexts-such as home or workplace, urban mobility or waiting time-can influence the way people deal with notifications. However, our video recording protocol and the interviews we conducted give us the opportunity to discuss former findings by describing practices in more detail. The section "Main Connection Cues for Social Media Apps During Daily Commutes" will therefore analyze the main connection cues for social media apps in a mobility context. We emphasize the importance of a specific spatial cue in the organization of usage patterns-boredom during waiting times. The section "Visual Forms of Social Coordination Performance According to Technical Cues" will describe moment-bymoment how smartphone usage patterns are organized according to contextual issues specific to urban mobility settings. This aims at reporting spatial cues related to these uses and mediatized forms of social connectedness. This analysis allows us to highlight how mSNS and social demands displayed by notifications renew social interactions in the context of public transportation. Main connection cues for social media apps during daily commutes. Our video recording protocol allows us to understand smartphones and social media triggers in public settings, as Alexandra, a 23-year-old student explains: Researcher: In which situations do you use the Facebook app? Alexandra: In which situations . . . in the subway most of the time or in the street when I have nothing to do. When I walk, it's less easy but, yes, when I'm in a waiting room, when I'm waiting for something, when I don't know what to do. When I'm just waiting without anything special to do on the Internet and I don't necessarily want to send messages or don't have anything to say to my friends, I like to go on Facebook, have a look at my account, it allows me to make use of the time. I like to check the latest publications. The Facebook app evidently gives this user the opportunity to make the most of waiting times and moments when she does not know what to do, as it is often the case when she is on the subway. If boredom seems to be the main reason why she checks her smartphone, this does not explain how she concretely appropriates and uses it. She chooses to consult Facebook when she does not know what else to do on the Internet or does not want to send messages. As she says, she would do differently if she wanted to communicate via SMS or browse the Internet to look up for something specific. Her usage is defined according to this temporary desire. This finding shows how problematic identifying a dominant usage pattern can be when user's main motivation is to stop boredom. Most large-scale analyses of smartphone uses, however, make the mistake of forgetting that users' first motivation is relieving boredom. Thus, we will focus on identifying regularities in usage patterns while recognizing their randomness. The example of Emilie is useful to highlight this issue. She usually prioritizes notifications when she uses the Facebook app . She sometimes extends her use to reading the news feed or closes the Facebook app to open another one . She can also start by reading news posts before checking notifications , especially because the app opens directly into the news feed, so she sometimes takes a quick look at it before reading notifications. Even though variations appear within Facebook usage patterns, it is still possible to identify regularities as seen above and confirmed with Jennifer's-a 26-year-old saleswoman-practices: I do not often use all Facebook functionalities, but I check it, yes. I check the news feed almost every day. But I prefer to read other people's posts than posting things myself. I like to read people's profiles, look at their photos, well, that is something I enjoy, especially with people I like. This tendency to appropriate Facebook on a consultation mode is particularly linked to mobile applications' usage constraints, as exemplified by Betembi, a 36-year-old nurse: Well, it's true that, on my phone, my uses [of Facebook] are very, well . . . very brief. I don't know how to say, browsing is not very smooth and it gives me less freedom than on the computer. With the computer, I really have a global vision, so it's faster, I can easily read one thing or the other whereas, on my phone, I just read the latest comments, profile updates, possibly profile pictures that have been changed. This is a recurrent trend among users. They appropriate their smartphone to check the latest SNS news and prefer to participate when they are at home, using their computer. They participate via mSNS when they have a request or are sent a comment or message, to coordinate in real time, or during particular events. As Francis, a 40-year-old associate professor puts it: "[On the Facebook app] I just have a look at what has happened, what people have posted and I react in the evening. There really needs to be a special event for me to respond quickly." This example suggests a distinction between answers to direct forms of interactionvia private messages or comments-and more public participation initiated by users when posting something or sharing content. It also exemplifies the time dedicated to consultation or participation in mSNS, as shown in Figure 2. Emilie published four comments over a week: she wished a happy birthday to a friend, then published three comments related to pictures: a friend's selfie, a friend's trip, and a consumer product. Even though publishing that kind of content is part of her usage, it still is much less frequent than the time she spends reading news or writing replies. This is why users report being less active when they use SNS on smartphones. However, users put this trend in perspective by emphasizing their appropriation of one-click participation options, such as the "Like" or "Share" buttons: "I really enjoy reading posts but I prefer commenting by saying 'I like'" [by clicking the "Like" button], I like this thing. In the end, we implicitly take part without necessarily saying something or writing a sentence, I find it easy, natural." These functions appear to be a form of participatory trigger that is suitable for the way smartphones can be mobilized to maintain mediatized relationships, especially during transportation times in public settings. We have detailed several connection cues for social media apps in a context of mobility such as boredom or one-click participation options. In what follows, we extend this analysis by showing how mSNS design frames the way users distribute their attention between their smartphone's interface and public environments. Visual forms of social coordination performance according to technical cues. Smartphone usage patterns are organized according to contextual factors such as urban mobility settings. Parisian users, for instance, have an opportunistic consumption of mobile TV and other Internet services in public transportation, taking advantage of public Internet connection when it's available, even though its quality is not always good . Beyond these technical constraints, social constraints such as overcrowded public transportation also play an important part in usage organization. As this 22-year-old student puts it: Researcher: Your trip [to go to university] is about 1 hour long? Mickaël: Yes. I only use the subway so I've got connections to make. The fact is when one is standing, well, it depends if you can sit or not, but when one is standing, when it's packed, it is more difficult to use the telephone. However, if I can sit down, I use it and it [the mobile phone network] generally works fine. Smartphone uses are indeed closely related to micromobility: the walk between two subway lines, other passengers' movements, the possibility to sit down, etcetera. From this point of view, our video-ethnographic protocol provides important insights to complement studies analyzing mobile phone uses in situations of mobility , especially in public transportation . One of the things it reveals is the way users organize their visual engagements-their gaze switches between their smartphone screen and the urban environment . It appears that this specific form of multiactivity-the joint commitment to mobility and screen-based activities-is organized according to the flow imposed by the devices. For instance, Clement, a 27-year-old student, opens the Facebook app when he takes the tram to go home. He starts by browsing through the news feed for 2 minutes. When he reaches the posts he has already seen in the morning-published "6 hours ago" -he quickly scrolls the screen up and, once at the top, he operates a top-down index motion, swiping his finger through the screen to "refresh" the application . A progress icon appears at the top of the feed indicating that the application is loading new content. We can note that Clement seizes this loading time as an opportunity to stop watching the screen and look up around him to scan the environment from right to left . Due to its important recurrence in our sample, this simple gaze switch highlights the way users manage urban settings during loading times and inactive phases of smartphone apps. It is not possible to tell what users are effectively looking at when they look up around them; they can make these gaze switches in response to a variety of causes . However, transition phases in smartphone usage patterns are essential to understanding usage complex cognitive mechanisms. The gaze switches example helps to demonstrate how visual forms of social coordination are reorganized nowadays according to mobile ICTs and their specific flow. This finding renews the relevance of the analysis of the "civil inattention" phenomenon described by Goffman to account for the organization of ordinary interactions consisting of glances exchanged between strangers. This visual form of coordination cannot be reduced to a simple internalized social norm consisting of a set of visual exchanges between two interactants who use and negotiate visual courtesies. The visual grammar of social interactions is now modified by smartphone screen issues and the way they frame the way we operate gaze switches in public settings. From this perspective, another usage pattern emerges from video recordings; users maintain their visual commitment towards the screen at the beginning of their use sessions. Then, over time, they look around them more often, visually monitoring the environment. Figure 10 illustrates this phenomenon. As shown in the section "How Social Demands Shape Smartphone Usage Patterns," users prioritize notifications when they activate their smartphone. They try to fulfill relational demands and obligations to respond quickly to incoming messages or comments. When they write answers, they tend to stay focused on this activity, which is quite absorbing on a cognitive level. Then, they stop for a while to look around. Once they have responded to incoming messages, they may put down their phone to commit to another activity or simply to look around them waiting to arrive to their destination. If they go on using mSNS, they enter what we called a "reading mode," during which they pay less attention to the screen. This becomes more obvious over time, as usage goes on. Their attention is highly focused when they start to read news and decreases as they reach older or previously read posts. This gradual loss of interest in mSNS use is coupled with an increased frequency of gaze switches and attention to the surrounding context. This phenomenon also appears during the longest app chain sessions, when users switch to other applications such as video games, most of which are simple and quick games that can be rapidly paused, resumed, or stopped and that do not require important or constant attention. For instance, it is possible to observe, when users play the widespread game "Candy Crush," how their gaze switches occur mostly during the transition phases of the game, during loading times between levels, or when candy clusters explode. In other words, when users appropriate their smartphone to manage their microcoordination of incoming messages-as during the opening of mSNSthey tend to disengage from physical contexts to focus on their device. Social demands mediatized by these media come first at that time and lead users to defer gaze switches and their participation in visual forms of coordination. This trend shows how relational norms at stake in mSNS use regulation can impact the economy of visual forms of coordination and social interactions between strangers. --- Conclusion and discussion Our study has provided an overview of connection cues for mobile social network sites and social messaging applications as a way of explaining contemporary forms of social connectedness. We aimed at illuminating nonconscious triggers in "checking habits" by focusing on a specific technical cue and spatial cues characteristic of daily commutes. We described how this specific set of cues frames mobile devices' opening patterns and usage in interaction with public transportation environments. We argued that the description of these connection cues helps to understand the way social media apps promote permanent connection and "ambient-mediated sociation" through notification of social demands. To achieve this goal, we developed a video recording protocol in order to collect fine-grained data. This method has strong limitations, however. First, this protocol only allows to record and document the longest chains of app usage, when smartphones are used for more than 10 minutes. This is an issue as 68% of use sessions only involve a single application and the average length of 50% of them does not exceed 5 seconds . Then, we tried to describe the temporal organization of smartphone usage patterns; identifying a usage pattern is a difficult task for several reasons. There are too many applications available today to identify a dominant pattern and their uses vary from one country to another . To identify patterns of use, we focused on mSNS-given that they are the most popular applications -in order to correlate them with sociodemographic variables and the contexts of use . We also focused on the beginning of apps usage chains by describing how notifications and applications that display social demands-such as incoming messages-tend to be used first. This video-ethnography, therefore, extend previous studies by providing a detailed analysis of opening sequences of usage patterns. It shows how they tend to be structured according to three phases: users often start by using applications displaying notifications; they favor those that display social demands; they hierarchize and prioritize these relational solicitations in accordance with social status or types of relationships , specific circumstances, and representations about legitimate or adequate response times. According to the way this technical cue mediatize social demands, this opening sequence can be considered as a vector of microcoordination and social connectedness enhancement. The design of smartphones effectively encourages users to be reactive and respond quickly to mediatized social solicitations, even during daily commutes; as we have noted, they do, thus showing how mSNS notifications strongly shape smartphone usage patterns as they promote ubiquitous forms of participation in social network services. This video-ethnography also shows how technical cues change visual forms of social coordination in public settings. When they start using mSNS, participants first manage notifications and answer solicitations, and they tend to stay focused on this absorbing screen-based activity. During this microcoordination phase, they managed their mobility by looking around them during loading times and technically designed activity breaks. Over time, when they went on to read news feeds, their gaze switches were more frequent. This phenomenon reveals how visual forms of social coordination are not only organized according to internalized social norms, in a Goffmanian perspective , but also depend on flows imposed by the device and ways in which mediatized forms of social demands are pervasively displayed on smartphones screens. Finally, this video-ethnography shows how mSNS design and their notification functions promote permanent connectedness. The boring and routine nature of urban mobility encourages users to look for new sources of entertainment, especially by experiencing new digital sociability, multiplying and diversifying their mediatized exchanges with close friends. This pervasive screen-based activity brings as a result an anonymous form of copresence that could translate into an uncommitted form of participation in public space, everyone being "alone together" . However, it is too radical to consider that this contemporary form of "media immersion" is becoming the dominant form of visual coordination in public settings. On the contrary, our investigation shows how this state of "being alone together" is rather a temporary and dynamic form of immersion in media uses, characteristic of a specific phase of usage patterns, the microcoordination phase. After that, once they have checked social solicitations mediatized by their smartphone, users pay more attention to the surrounding environment. It is also problematic to interpret this media immersion as an antisocial behavior, as we can see that when they are focused on their screens, users are most of the time engaged in social activity with their mediatized relationships. --- Author biographies Julien Figeac is a sociologist and researcher at French Center for Scientific Research , LISST , EHESS-Université de Toulouse Jean Jaurés. His work involves observational methods for studying the uses of mobile Internet and social media apps. He combines a videoethnographic approach with quantitative methods. Johann Chaulet is a sociologist and researcher at the French Center for Scientific Research . His investigations, mainly qualitative, focus on sociability, trust, and ICT relational practices.

This paper aims to analyze the uses of mobile social network services (mSNS) during daily commutes on the basis of a video ethnography conducted with 35 users of the Facebook app. This method is based on the combination of context-oriented recordings made with user-worn camera glasses and mobile screen video capture. These data reveal the way smartphone usage patterns tend to be organized according to notification functions (mSNS, SMS), a specific set of technical cues that mediatize social demand and promote social connectedness. Users manage these cues through a recurrent trend composed of a three-step sequence: they often start by using applications displaying notifications; they favor those that display social demands; and, among them, they prioritize these relational solicitations in accordance with social status or types of relationships. By examining the distribution of users' attention between urban environments and smartphone applications, this video-ethnography also highlights how these "checking habits" are organized according to a set of spatial cues and some daily commute characteristics, such as visual coordination with passengers in public transport. These technical cues mediatize a growing number of social demands that encourage users to keep their eyes focused on their smartphone's screen in public spaces. We argue that these technical cues create a temporary bubble effect and social isolation at a proximal scale, which mostly operate at the beginning of smartphone usage patterns.

Introduction ''La ciencia experimental ha progresado en buena parte merced al trabajo de hombres fabulosamente mediocres, y aun menos que mediocres'' Ortega y Gasset In contrast to Newton's [1,2] well-known aphorism that he had been able ''to see further only by standing on the shoulders of giants,'' one attributes to the Spanish philosopher Ortega y Gasset the hypothesis saying that top-level research cannot be successful without a mass of medium researchers on which the top rests comparable to an iceberg [3,4]. A third possibility offered by Turner and Chubin [5] is the so-called Ecclesiastes hypothesis: these authors argue that scientific advancements can be considered as the result of chance processes or fortune using an evolutionary model of science. The issue, discussed by many eminent scientists and philosophers, is highly relevant for today's research funding policies. From this perspective, one can discuss whether research funding should be focused on elite scientists or rather aim at generating scientific capacities in the broad range of scientists. In this study, we address this question from a bibliometric perspective using capabilities in literature databases that became recently available [6]: Unlike the Science Citation Index of Thomson Reuters, the Scopus database of Elsevier-launched in 2004-enables us to determine whether highly-cited papers themselves cite highly-cited papers to a significant extent. This provides some insights into whether giants in research like to build on the research of other giants. We gained these insights into four major fields of science: physical sciences, life sciences, health sciences, and social sciences. Both the citing and the cited papers were identified within the field-specific journal sets covered by the Scopus database. From a sociological perspective, our bibliometric approach may have only limited value because citations are an imperfect proxy for the actual usage of research results. Citations are just one parameter of scientific quality. However, the strength of this approach lies in the large number of observations that can be evaluated. Statistical analyses of bibliometric data may allow us to cast new light on the validity of the three hypotheses and give insights into the expected effects of different research funding models. Recently, there is a trend away from a model to allocate research funds on the basis of block grants to institutions towards resource allocation based on the principle of merit of individual researchers [7]. Institutional allocation which follows a principle of equality can perhaps be legitimated in terms of the Ortega hypothesis more than in terms of the elite structure proposed by the Newton hypothesis [8]. In the latter model funding were to be concentrated to the top scientists in order to create a critical mass of elite scientists [9]. This focus on top quality can perhaps be justified by the wish to obtain increased accountability of academic research [10]. Although we witness an increased focus on excellence in science funding [11], it is yet unclear which of the three competing hypotheses can be supported by the data. Is top-level research systematically connected to top-level research in the past or does top-level research also presume research at the medium level? The few studies which study the empirical merit of the three hypotheses were mostly published several years ago and based on small samples within single disciplines [12]. --- Methods Using citation analysis for the operationalisation, two basic assumptions are made. From the citing perspective, one assumes that papers cited by scientists represent a roughly valid indicator of influence on their work [13]. A cited reference can perhaps be considered as a reward for the usefulness of the cited paper [14]. The aggregate of cited references in a paper can be considered as indicating the theoretical and empirical resources for building an author's argument [15]. However, individual papers may accumulate citations for trivial reasons [16]. In the case of large numbers, these deviances may be averaged out. Thus, with sufficiently large numbers citation frequency can be assumed as a proxy for impact [17]. Our study is based on the Scopus database that offers the possibility of direct coupling between the cited references within a paper and their respective numbers of citations. In the Science Citation Index, a two-steps coupling procedure is then needed and the procedure is error-prone because the information in the cited references is often incomplete. In Scopus, cited references are uniquely identified as previously published papers. Although the Scopus database indexes more journals than Thomson Reuters' citation indexes [18], it may also contain more peripheral journals, that is, journals publishing papers with low visibility or publishing papers without applying the peer review process [19]. In order to control for this effect we use the intersection of 6,578 journals between the Scopus set and the Science Citation Index set in the publication year 2003. This group of journals is acknowledged by the teams at both Thomson Reuters and Elsevier as of sufficient visibility to warrant inclusion into its set. In other words, to use only journals in our study with a ''higher'' visibility , we restricted the Scopus journal set to those journals that are also included in the Science Citation Index. Like in the citation indexes of Thomson Reuters, scientific fields are defined in Scopus in terms of journal sets. There are 305 ''specific subject areas'' organized into 26 ''subject areas'' , plus a ''general subject area'' containing multidisciplinary journals such as Nature or Science [20]. The subject areas are grouped into four main fields: all articles and proceedings papers which were published in 2003 in the life sciences , health sciences , physical sciences , and social sciences , and all articles and proceeding papers which were cited within these publications. These cited references amount to: life sciences , health sciences , physical sciences , and social sciences . We only included references to papers published within the Scopus journal set since no citation data is available for papers outside this set. Since researchers grouped in the social sciences category frequently publish in books and non-English journals, the numbers in this area are smaller than in the life sciences, health sciences, and physical sciences [6,20]. As Scopus provides reliable citation coverage only from 1996 onwards [21], we included only cited references published since that date. We studied the citation impact of the papers which are cited in all the papers with publication year 2003. As normalizations, first, the citation windows are set to five years after the year of publication. In other words, we gathered the citations of a paper published in 1999 for the period 2000 to 2004. Secondly, all articles and proceedings papers-both the cited and the citingwere categorized in six percentile rank classes . This normalization accords with that of the National Science Board of the U.S. National Science Foundation [22]: percentile rank classes are suited for identifying lowly-, medium-and highly-cited papers in a field. Both the National Science Board [23] and the Essential Science Indicators of Thomson Reuters classify papers as highly-cited if they belong to the top 1% of papers worldwide . The Ortega hypothesis predicts that highly-cited papers and medium-cited papers would equally make references to papers with a medium impact . The Newton hypothesis would be supported if the toplevel research is more frequently based on previously highly-cited work than that medium-level research cites highly-cited work. If scientific advancement is a result of chance processes , no systematic association between the impact of cited and citing papers is expected. --- Results Figure 1 shows the percentile rank classes of the citing papers published in 2003 against the percentile rank classes of the cited references for each field. Both the ordinate and abscissa are used to describe the impact of the cited papers : The abscissa provides the percentile rank classes; the ordinate provides the percentage of the papers that belongs to this percentile rank class. The different impacts of the citing papers are shown by differently coloured lines. The share of cited references in the papers belonging to the 99 th percentile is represented by a black line, 95 th by a purple line, 90 th by a green line, 75 th by a blue line, 50 th by an orange line, and ,50 th by a brown line. The red line refers to the cited references in all citing papers published in 2003. Because citations to papers generally follow the well-known right-skewed distribution [24], the cited references in the figure are characterized by a citation impact that is at least at a medium level . In other words, the lower halfs of the distributions do not contribute to the citation patterns. As all cited references that are included in our study are cited at least once, the lowest impact class of cited references in all fields and of all percentile rank classes of citing papers is close to zero. The graphs in Figure 1 show both similarities and differences among the four fields. In all fields, the high-impact papers cite high-impact papers to a larger extent than the papers in the other impact classes . Conversely, medium-impact papers cite medium-impact papers to a larger extent than high-impact papers . This means for all four fields that the high-impact research is connected to previously high-impact research more strongly than low-or medium-impact research, and the lower the impact of a paper published in the four fields, the higher the share of cited references with medium impact. Both these findings support the Newton hypothesis for all four fields. In addition to these similarities, there are also differences among the fields. First, in the life sciences and health sciences the differences among high-, medium-, and low-impact research in using preceding top-level research are large; in the physical sciences and especially in the social sciences these differences are much smaller. This could mean that the Newton hypothesis is valid to a different extent: our results suggest that this hypothesis is more corroborated in the life sciences and health sciences than in the physical sciences and social sciences. Second, the red lines in the four graphs of the figure which show the aggregate of papers in each field refer differently to highly-, medium-, and lowly-cited papers. Whereas in the life sciences and health sciences the share of cited references within the top-level impact class is larger than 20%, it is less than 20% in the physical sciences and less than 10% in the social sciences. Correspondingly, there is a high share of cited references in the case of the social sciences and physical sciences in the medium impact class; this share is significantly smaller in the life sciences and health sciences. What are possible explanations for these differences among fields? The explanations could be technical or sociological in nature. A technical explanation might be that the differences in coverage of cited references by the Scopus database affect the results. Whereas in the life sciences 43% of the cited references were to journals indexed by Scopus, in the physical sciences, health sciences, and social sciences these percentages were 31%, 31%, and 3%, respectively. Perhaps, top-level research in the social sciences is predominantly published in media other than scholarly journals. A sociological explanation could be that the results reflect differences in the paradigmatic fragmentation among the fields. Paradigms can be considered as clusters of theories and practices that determine the direction of research [25]. Publications in the life sciences are more codified across the board and one can also focus on commonly shared goals across disciplines more than in the social sciences [26]. The research traditions in several subfields grouped under the denominator of ''social sciences'' are organized in many schools of thought which are not strongly interconnected [27]. One would like to be able to control for whether the pattern in the data of Figure 1 could find its origin in ''lazy authors citing the most obvious papers'' rather than ''giants citing other giants.'' Perhaps, highly-cited papers were so often cited because of the ''success-breeds-success'' phenomenon [28,29] in citation behaviour-rather than because of containing the crucial papers on which one builds-and thus overshadowing some innovative publications which would have deserved to be cited. In order to control for the validity of our results, therefore, we included in a second test only references from 2002 cited in papers published in 2003. One can assume that the authors of these papers could not know the subsequent citation impact of the papers published in 2002, since papers published in 2003 were written with only a few exceptions in 2002 or earlier. This second analysis included all articles and proceedings papers which were published in 2003 in the life sciences, health sciences, physical sciences, and social sciences and all articles and proceeding papers which were cited within these publications: life sciences , health sciences , physical sciences , and social sciences . As in the first analysis, there are large differences in the numbers of cited references between the social sciences and the other three fields. The results of the second analysis are presented in the right column of Figure 1. The differences to the graphs displayed in the left column are small. The similarities of the results between the graphs in the left and right columns reveal that the authors apparently disregard the citation impact of the cited papers in their decisions to cite these papers. --- Discussion In summary, highly-cited work in all scientific fields is more strongly based on previously highly-cited papers than on mediumcited work. Thus, we are able to demonstrate that papers contributing to the scientific progress in a field lean to a larger extent on previously important contributions than papers contributing little. In other words, the higher a paper's citation impact the stronger it is connected to preceding high-impact research . These findings support the Newton hypothesis and call into question the Ortega and Ecclesiastes hypotheses . Our results also suggest that medium-impact research plays a different role in the four fields: whereas in the social sciences and physical sciences scholars cite this underlying research, in the life sciences and health sciences the subtop is less important. Given that research funding is commonly scarce, it is the responsibility of the scientific community to most effectively utilize the resources available. [30]. Our findings raise the issue of whether limited resources might best be concentrated in support of those scholars who have already contributed to the literature by publishing high-impact papers . A concentration of resources on these elite structures seems to be practical especially for the life sciences and health sciences. Indeed, current courses of action in research funding follow the concentration of scarce resources on outstanding researchers. The Wellcome Trust will allocate 20% of its total budget to an Investigator Awards program [31]. This program will fund only the very best scientists to investigate challenging and long-term research questions. The German Max Planck Society follows the so-called Harnack Principle. One formulates at the website as follows: ''Max Planck Institutes are established only where the world's leading researchers are found. The researchers themselves determine their research topics and they are given optimum working conditions and the freedom to choose who they want to work with them'' . The U.S. National Institutes of Health supports researchers with similar programs. Against the backdrop of our findings, these courses of action seem to be sensible especially in the life sciences and health sciences. In these fields, one can probably follow the argument of Cole and Cole [13] that the progress of science would be little impeded if only scientific excellence were supported. It should be tested in follow-up studies in the next years whether this statement can be hold. Several limitations may have affected our results that should be considered in future studies: It is not yet clear whether citation impact is a good approximation of actual research impact and of the role of research in scientific advancements [32,33]. Furthermore, bibliometric research is limited to the analysis of scientific publications. The Ortega hypothesis relates explicitly to the experimental sciences thus including not only the previous literature as the basis of scientific progress but many different kinds of assistance . Newton [1,2] refers to what he had done as a theoretician ''on the shoulders of giants,'' which is more likely to be covered by literature. Our data indicate that a notable percentage of the papers cited in top-level papers is itself classified as medium-impact work. One does not know whether important contributions at the research front could have been made if only the top-level work had been available for referencing. Although de Solla Price [34] has postulated an ''immediacy factor'' in which most scientific publications usually cite recent work, and papers tend to become obsolete within five to ten years [35], it might be that limiting our data to papers published after 1996 affects the results [36]. If possible, this study should be replicated using the Science Citation Index which contains the historical backlog. The databases are restricted to mainly international journals and papers published in English. This restriction cannot be avoided by using current literature databases, but may affect especially the results for the social sciences. It could be interesting to repeat the same analyses excluding self-citations. Although Boyack and Klavans [37] showed that self-citations cannot explain the strong association between citation impacts of the cited and citing papers at the aggregate level, the strong connection between current and previous top-level research in the life sciences and health sciences might partly be the result of large research programs that cite to a large extent internally. However, the systematic correction for self-citations is nearly impossible at the author level because of the strong homonymies among author names. For example, the Scopus database covered 8,173 documents of authors with the name ''Singh'' in 2009. We proceed on the assumption that these limitations do not affect our results to such an extent that they are not valid. ---

In contrast to Newton's well-known aphorism that he had been able ''to see further only by standing on the shoulders of giants,'' one attributes to the Spanish philosopher Ortega y Gasset the hypothesis saying that top-level research cannot be successful without a mass of medium researchers on which the top rests comparable to an iceberg. Methodology/Principal Findings: The Ortega hypothesis predicts that highly-cited papers and medium-cited (or lowlycited) papers would equally refer to papers with a medium impact. The Newton hypothesis would be supported if the toplevel research more frequently cites previously highly-cited work than that medium-level research cites highly-cited work. Our analysis is based on (i) all articles and proceedings papers which were published in 2003 in the life sciences, health sciences, physical sciences, and social sciences, and (ii) all articles and proceeding papers which were cited within these publications. The results show that highly-cited work in all scientific fields more frequently cites previously highly-cited papers than that medium-cited work cites highly-cited work. Conclusions/Significance: We demonstrate that papers contributing to the scientific progress in a field lean to a larger extent on previously important contributions than papers contributing little. These findings support the Newton hypothesis and call into question the Ortega hypothesis (given our usage of citation counts as a proxy for impact).

Introduction The US has experienced alarming growth in obesity in recent decades . Although most socio-demographic groups have been affected, race/ethnic disparities are pronounced . Obesity prevalence among Hispanics, particularly of Mexican origin, is considerably higher than among whites . There is evidence that the disparity in prevalence between Mexican-Americans and whites masks important heterogeneity. Foreign-born Mexican-Americans are characterized by a lower body mass index compared to the US-born despite having lower socioeconomic status . There is also important heterogeneity within foreign-born groups; individuals residing in the US for a longer period of time tend to be more obese, possibly reflecting the impact of behavioral change and increased exposure to obesogenic environments However, previous studies examining weight differentials in nativity and length of residence have relied on single cross-sections. Moreover, none has examined whether these relationships have changed over time. A more complete understanding of this within-group heterogeneity and of changes in within-group differences over time is critical to identify vulnerable subpopulations among Mexican-Americans, the largest immigrant group and Hispanic subgroup in the US . There are several reasons why the relationship between nativity/length of US residence and weight may be changing over the past two decades. First, obesity rates in Mexico are rapidly increasing; among more recent immigrants, the obesity epidemic in their home communities has become comparable in scale to that of the US communities to which they are migrating According to a highly publicized report published by the Organization for Economic Co-operation and Development in 2010, Mexico's obesity prevalence is now only marginally lower and second in the world to that of the US. By another measure, Mexico's overweight prevalence ranks as the highest in the world . The rapid rise in obesity in Mexico is also evident when compared against patterns among Mexican-Americans. In 1988, obesity prevalence among women in Mexico was 9 %-considerably lower than the 35 % prevalence among Mexican-American women ; but by 2006, national estimates among Mexican women reached 37 % which was only slightly lower than the 42 % prevalence among Mexican-American women . These patterns suggest that more contemporary immigrants may be entering the US with higher BMI levels than immigrants from earlier cohorts. If new immigrant BMI estimates are also rising at a faster pace than those among US residents, this could result smaller weight differentials by nativity and length of residence in more recent times. If foreign-birth no longer exerts the health 'protection' implied by past patterns, this will have implications for the targeting of public health interventions. Another factor that may underlie secular variation of weight differentials by nativity/length of US residence is the changing nature of Mexican migration to the US. Research shows that since the late 1990s, migration has shifted away from traditional origin regions in the Western and Central parts of Mexico to poorer areas in the south . Although the southern states of Mexico are characterized by lower obesity than other parts of the country, it is these poorer areas that have seen the greatest increases in obesity over the past decade . Over this timeframe, Mexican migration has also shifted to new destination areas in the US like the South and away from traditional receiving states like California . Migration to new areas previously unaccustomed to immigrants may result in increased marginalization with implications for the acculturation process and its association with weight over time. Using nationally representative data on Mexican-American adults, we examined whether associations between nativity/length of US residence and BMI and waist circumference varied over a period of 20 years. We also investigated whether secular trends in these associations varied by education. Recognition of these patterns and the processes that underlie them is fundamental to understanding the causes of the obesity epidemic in Mexican-Americans and to the development of more effective strategies for prevention. --- Methods Data came from successive waves of the third National Health and Nutrition Examination Survey III ), and the continuous NHANES . NHANES is a series of cross-sectional nationally representative health examination surveys beginning in 1960. In each survey, a nationally representative sample of the US civilian noninstitutionalized population was selected using a complex, stratified multi-stage probability cluster sampling design . Oversampling of Mexican-Americans did not begin until NHANES III precluding use of earlier surveys. NHANES III was conducted between 1988 and 1994, and was designed so that the entire 6 year was a national probability sample. In 1999, NHANES became a continuous survey, in which ~5,000 individuals of all ages completed the health examination component of the survey each year. The continuous surveys, a series of repeated crosssections, are available in 2-year blocks . There were two phases of data collection: in the first phase, researchers collected information from household interviews on demographics, socioeconomic indicators, past medical history, and health behaviors. In the second phase, participants were administered a physical examination in a mobile examination center. To achieve sufficient sample sizes, we pooled NHANES data to represent three time points : 1988-1994, 1999-2004, and 2005-2008 . We restricted the sample to adult, non-pregnant women and men aged 20-64 who self-identified as Mexican/Mexican-American. The sample was limited to adults < 64 years to avoid selection problems that may arise from morbidity/mortality associated with older age, and to allow for a more interpretable examination of BMI differences that are less likely to be influenced by age-related loss of body mass . Of the 4,614 men and 4,199 women meeting our inclusion criteria, we further excluded individuals with missing information on BMI and other key covariates of interest , yielding a final sample of 4,235 men and 3,914 women for analyses examining BMI. Waist circumference models included a sub-sample of 4,129 men and 3,808 women among whom this information was available. Height , weight , and waist circumference were measured via physical examination. Body mass index and waist circumference were examined as separate outcomes and as continuous variables. For descriptive purposes, participants were also classified based on the World Health Organization's criteria for abdominal obesity . Obesity prevalence was also modeled. Nativity , years lived in the US among the foreign-born, age, sex, and education were self-reported during the household interview. We created a single three-level variable to examine nativity and length of US residence together: US-born , <10 years in the US, ≥10 years in the US. --- Statistical analysis Results were stratified by sex. Appropriate sampling weights were incorporated to produce national population estimates for Mexican-Americans for each calendar period. Sampling weights accounted for unequal probabilities of selection, non-response, and non-coverage. All analyses were conducted using SAS version 9.2 and SUDAAN version 10.01 with Taylor series linearization methods to adjust for the complex survey design. Age-adjusted means and prevalence were first calculated and plotted for each level of the nativity/length of US residence variable both within and across survey periods. For reference, estimates for whites were also plotted. To facilitate comparisons, estimates were age-adjusted by the direct method to the 2000 US standard population . Differences across nativity categories, length of US residence and survey years were evaluated using the t statistic, and a p value of<0.05 was considered statistically significant. Multivariable linear regression was used to separately model associations between nativity/length of US residence and continuous measures of BMI and WC among Mexican-Americans, controlling for age and NHANES survey period [1988-1994 , 1999-2004, 2005-2008]. An age-squared term was retained in models in which age had a non-linear relationship with BMI and WC . An age-by-period interaction was also included since the association between age and all anthropometric measures was not constant over time. Subsequent models controlled for education. To investigate whether the relationship between nativity/length of US residence and BMI/WC varied with time, models included an interaction between the nativity/length of residence variable and NHANES period. We also considered whether secular trends in the association between nativity/length of residence and BMI/WC differed by education. Multivariable logistic regression was similarly used to estimate odds ratios for obesity; however, since results were similar to those obtained with BMI, results are not shown. --- Results Across all time periods, foreign-born Mexican men and women were more likely than USborn Mexicans to have less than a high school education . However, the educational distribution among all Mexican-Americans shifted toward higher educational attainment over time. Among men, the foreign-born consistently had a lower mean BMI and WC, and a lower prevalence of obesity and high-risk WC than the US-born. In contrast, among women, except for mean WC in 2005-2008 which was lower in foreign than in the US-born, there were no statistically significant differences by nativity for other years or other anthropometric measures. Among men, BMI, WC, obesity, and high-risk WC were higher in 2005-2008 than in 1988-1994, regardless of nativity, though the increase in high-risk WC among the foreign-born was not statistically significant. Among foreign-born women, estimates for all anthropometric measures were higher in 2005-2008 than in 1988-1994, but only the difference in WC was statistically significant. Among US-born women, all anthropometric measures, except BMI, were significantly higher over time. Figure 1a-f shows age-adjusted mean BMI, mean WC, and obesity prevalence across NHANES survey periods by nativity and length of US residence among Mexican-Americans and whites for men and women. Among men, US-born Mexican-Americans had the highest BMI, whereas Mexican immigrants in the US <10 years had the lowest BMI . Immigrants in the US ≥10 years and whites had intermediate and similar BMI levels. These patterns generally held across time and were also present for WC and obesity prevalence . In contrast to the patterns among men, all Mexican-American women, regardless of nativity and length of residence, had higher BMI than white women . In general, there was little difference in BMI among the three Mexican-American groups. Patterns were similar for WC and obesity prevalence . Table 2 presents adjusted mean differences in BMI and WC by nativity/length of US residence by period , and mean differences in BMI and WC across time periods by nativity/length of US residence among Mexican-Americans. Estimates were derived from a model including nativity/length of US residence, age, age 2 , NHANES survey period, and interactions between age and period, and between nativity/length of US residence and period . Age was mean-centered to the gender-specific sample mean . Among men, Model 1 confirmed findings illustrated in Fig. 1a,b. All men experienced increases in BMI and WC over time, although changes were of a smaller magnitude and not statistically significant among immigrants in the US < 10 years . However, there was no evidence of secular variation in the association between nativity/length of US residence and BMI or WC [p-interaction = 0.3 ; p = 0.5 ]. Adjusting for education did not appreciably alter estimates . Among women, age-adjusted models confirmed patterns from Fig. 1d,e. BMI and WC increased over time in all groups though BMI increases were small and less likely to be statistically significant than WC increases . Mean BMI and WC did not differ by nativity or length of residence and this pattern did not vary over time [Model 1; p-interaction = 0.9 ; p = 0.9 ]. However, after adjusting for education, foreign-born women, regardless of length of residence, had a lower mean BMI and WC than US-born women, but there was still no evidence of any time variation in this pattern . In women, there was modification of these secular trends by education. Since trajectories were similar among individuals who completed high school and those with more than high school, we collapsed this category into 'high school or more' to ease interpretation and improve stability of estimates. Among Mexican-American women with less than high school education, BMI differences by nativity and length of residence became magnified over time: in 1988-1994 there was little difference in BMI across the three groups, but by 2005-2008, the US-born had a considerably higher BMI than either of the foreign-born groups . This gap was attributable to large increases in BMI among US-born women with less than high school education, and little increase in BMI over time among similarly educated foreign-born women. Among women with high school or more education, the USborn had a higher BMI than both foreign-born groups in 1988-1994, but by 2005-2008, nativity differences were minimized due to little change in BMI among the US-born paired with marked increases in BMI over time among both foreign-born groups . Similar findings were observed for WC [3-way interaction between nativity/length of US residence × education × survey period: p = 0.0114; p = 0.05]. --- Discussion Using data from nationally representative samples of Mexican-Americans, we found that among men, foreign birth, regardless of length of residence, was associated with lower BMI and WC. In addition, immigrants living in the US longer had higher BMI/WC than more recent arrivals. Foreign birth was also associated with lower BMI and WC in women but this finding was only evident after adjusting for education. Anthropometry gradients by length of residence were also less apparent in women than men. There was no evidence that the associations between nativity/length of US residence with BMI/WC varied over the 20 years spanning 1988-2008 regardless of sex. The nativity and length of residence gradients we report, particularly in men, are consistent with patterns described in the literature . BMI and WC were highest among the US-born and lowest among the most recent immigrants. Explanations for these patterns have focused on selective migration and protective cultural characteristics among newer immigrants. Migrants are thought to be healthier relative to their native populations, and are thus selected for their ability to cope with the rigors of migration . To explain the later decline associated with longer US residence, hypotheses have focused on acculturation, a process whereby immigrants progressively adopt the detrimental behaviors and norms of their new culture, such as poorer diet and sedentary lifestyles . Some studies have also reported a stronger association between longer length of residence and weight among immigrant women relative to patterns among men . In contrast, we found these associations to be smaller in magnitude in women than in men. Reasons for this discrepancy are unclear however results from previous studies applied more to the broader Hispanic population in the US without distinction by country of origin , or were specific to a more localized community of Mexican-Americans . We found no evidence of secular variation in the association between nativity/length of US residence and BMI or WC, regardless of sex. In the context of Mexico's emerging obesity epidemic, we anticipated some narrowing of the nativity/length of US residence gradient over time. However, we observed no such pattern. Disparities in BMI or WC between foreign-born and US-born persons may be related to differential exposure to obesogenic environments and associated behavioral consequences. Alternatively they could reflect selection factors by which immigrants are a healthier subset of the population from which they migrate. The extent to which Mexican immigrants are positively selected on health is unclear, but one study that compared the weight of Mexican migrants to non-migrants demonstrated a lower prevalence of overweight among migrants . If selection factors are responsible for the stable nativity differences we observed, they must be operating similarly over time despite a background of rising obesity in Mexico. Our results suggested complicated interactions over time among nativity, sex, and education. Education was more strongly associated with BMI and WC in women than in men, and functioned as a negative confounder of the association of foreign birth with anthropometrics. In addition, there was evidence that education modified the association between nativity and anthropometrics over time. Among women, we observed a widening of the nativity gradient in BMI among the less educated, driven by alarming BMI increases among the US-born. In contrast, among women with high school or more education, the nativity gradient narrowed over time as a function of BMI increases among the foreign-born and little BMI change in the US-born. These results highlight differences in the socioeconomic patterning of obesity that appear to differ by nativity: among less-educated women, BMI increased more rapidly among in the US-born than the foreign-born, while the opposite was true in the more educated. Although SES gradients in health among US Hispanics are reportedly weak , these findings point to the emergence of an SES gradient, especially among US-born Mexican-American women. They also underscore the importance of simultaneously accounting for the joint influence of nativity, SES, and time to better understand health patterns among Mexican-Americans. Our study had several strengths. First, we used large, nationally representative datasets that over-sampled Mexican-Americans over a large time span. Second, use of clinically measured anthropometric indicators mitigates problems with validity and reliability inherent in self-reported measures. Third, unlike previous studies that explored anthropometric trends among Mexican-Americans as a single group , or that restricted examination of nativity differentials to a single time point , we were able to investigate how weightrelated patterns by nativity and length of residence may have changed over time within the largest US Hispanic subgroup. This work also had some limitations. Although NHANES data on Mexican-Americans is intended to be nationally representative, the extent to which undocumented individuals were captured is unclear. The undocumented are estimated to constitute more than half of the Mexican immigrant population in the US . In general, data among newer immigrants may not be adequately representative of all new Mexican immigrants to the US Other data sources may be necessary to better quantify health patterns for newer arrivals. Reliance on repeated, cross-sectional data is also a limitation, although the NHANES are arguably one of the most important data sources for reporting US obesity trends. Moreover, national, longitudinal samples of Hispanics are limited. We documented trends in anthropometry, but these repeated cross-sectional samples may be composed of individuals different on several unmeasured characteristics. Circular migration is not uncommon, particularly among Mexican men , which complicates findings based on length of residence. Without the ability to follow the same individuals over time, we cannot ascertain, for example, the extent to which the higher weight among long-term immigrants is a reflection of greater return migration of healthier individuals. Studies documenting migration patterns of Mexicans to the US also reveal a greater likelihood of return migration in the wake of enactment of immigration policies aimed at legalization of long-term immigrants. After the passage of the Immigration and Control of 1986, for example, return migration rates increased dramatically , possibly biasing findings among long-term immigrants that remained in the US. If immigrants more physically capable of return travel are not represented in estimates of long-term immigrants, we may falsely attribute a decline in health among immigrants that remain in the US to their greater exposure to US society. Analogously, rather than operating as a risk factor for weight gain, longer length of residence may instead be a reflection of who remains in the US over the long-term. Additional research on migratory patterns and future studies that integrate other measures of exposure to US society may help shed light on the dynamics underlying the patterns we report. --- Conclusions We contribute to previous research on weight patterning in Mexican-Americans by examining whether the association between nativity/length of US residence and BMI/WC has changed over time. Our findings illustrate the ways in which the impact of migration on health cannot be understood in isolation from the roles of gender, SES, and time. Although overall nativity/length of residence gradients in weight appeared stable over time, we noted an alarming rise in BMI among less-educated, US-born Mexican-American women-a finding that is all the more notable in light of the weak SES gradients in health that are often reported in studies of US Hispanics. Changes to US immigration policy and other economic and social factors that motivate migration will likely continue to play an important role in shaping the health profile of immigrants in the future. Continued monitoring of these trends in Mexican-Americans will be essential to the development of more effective strategies for prevention for this fastgrowing segment of the US population. Age-adjusted body mass index , waist circumference , and obesity prevalence for Mexican-American foreign-born by length of US residence and US-born men and women by National Health and Nutrition Examination Survey years. Estimates for whites presented for reference. Ageadjusted by the direct method to the year 2000 US Census population using age groups 20-34, 35-44, and 45-64

Objectives-We investigated whether associations between nativity/length of US residence and body mass index (BMI) and waist circumference (WC) varied over the past two decades. Methods-Mexican-Americans aged 20-64 years from the National Health and Nutrition Survey (NHANES) III (1988-1994), and NHANES (1999-2008). Sex-stratified multivariable linear regression models further adjusted for age, education, and NHANES period. Results-We found no evidence of secular variation in the nativity/length of US residence gradient for men or women. Foreign-born Mexican-Americans, irrespective of residence length, had lower mean BMI and WC than their US-born counterparts. However among women, education modified secular trends in nativity differentials: notably, in less-educated women, nativity gradients widened over time due to alarming increases in BMI among the US-born and little increase in the foreign-born. Conclusions-Associations between nativity/length of US residence and BMI/WC did not vary over this 20-year period, but we noted important modifications by education in women. Understanding these trends is important for identifying vulnerable subpopulations among Mexican-Americans and for the development of effective health promotion strategies in this fastgrowing segment of the population.

to ascertain demographic characteristics, date of symptom onset, university status, class attendance type , place of residence, and involvement in community-and school-related activities during the 14 days preceding illness onset. Illness onset was defined as the earliest date of reported symptoms for symptomatic cases* or as date of specimen collection for asymptomatic cases . Initial case reviews by ADH identified patients who reported recent participation in sorority or fraternity activities; subsequent infections were observed among cohabitants of these persons. A network analysis was performed to assess the relationship between participation in university fraternity or sorority activities and the spread of COVID-19 among residential communities at university A. MicrobeTrace, a network analysis and visualization tool developed by CDC, † was used to visualize and describe the full network of persons with COVID-19 and identify potential transmission-related gatherings , defined as one or more gatherings in which multiple cases are identified and epidemiologically linked. COVID-19 patients were included in the network if they lived on campus, participated in fraternity or sorority activities, or lived in the same off-campus dwelling as a person who participated in these activities. Case-to-residence and case-to-event networks were constructed to map residential university A transmissionrelated gatherings. These networks were used to infer gatherings by algorithmically partitioning the network into the most densely linked cases . In this network analysis method, the algorithm randomly assigns nodes a "community" identifier, and community connectivity is measured. For each person in the network, community identifiers are swapped with their neighbors and then remeasured. Only changes in community affiliation that improve connectivity are preserved, reversing any swaps that do not improve connectivity. This activity was reviewed by CDC and was conducted consistent with applicable federal law and CDC policy. During August 22-September 5, a total of 965 universityassociated COVID-19 cases with illness onset on or after August 20 were identified, including 699 confirmed cases and 266 probable cases . A 3-day, ADH-sponsored testing event, held during September 1-3 in response to an increase in universityassociated cases detected by ADH, resulted in a 22% test positivity rate and identified 324 cases; overall, 34% of cases were identified through this testing event. Among the 965 confirmed and probable cases with illness onset dates during August 20-September 5, 673 occurred in women and 936 were in persons aged 18-24 years . Five cases in persons identified as faculty or staff members were reported. The number of cases with reported illness onset on a given day increased and peaked on August 31 . Forty-eight persons with COVID-19 had received in-class instruction, 292 participated in fraternity or sorority activities, and 149 lived in a fraternity or sorority house. The network analysis linked 565 cases to 56 residences and to their case-to-event associations . The full network consisted of one large, linked cluster of 471 cases and eight smaller, unlinked gatherings of 94 total cases . Fifty-four gatherings were detected, including 27 with at least five cases. Among persons in 44 gatherings, at least one member regularly attended in-person classes, and at least one member of each of the 49 gatherings reported participation in fraternity or sorority activities or events. Gatherings included an average of 20.3 cases . Among 58 links between gatherings, 42 were associated with fraternity or sorority activities, 11 with on-campus dormitories, and five with off-campus apartments and houses. --- Discussion Within 2 weeks of the start of the 2020-21 academic year, COVID-19 cases rapidly increased among persons associated with university A. Transmission was likely facilitated by onand off-campus congregate living settings and activities, with a majority of the gatherings and links between them associated with fraternities or sororities. Most patients reported virtual instruction only, which indicates transmission likely occurred primarily outside the classroom; this finding is supported by the very small proportion of cases among faculty and staff members . Women constitute 54% of university A's 2020 student body but accounted for 70% of university A's COVID-19 cases. Among linked gatherings, women accounted for 86% of cases, a finding that could reflect involvement in gender-specific activities, including sorority rush week, which held an in-person outdoor bid day event and occurred before fraternity rush week, which was both held later and virtually. Understanding networks can provide insights into COVID-19 transmission dynamics and inform effective mitigation strategies. In the absence of detailed person-to-person transmission data from contact tracing, network analysis using US Department of Health and Human Services/Centers for Disease Control and Prevention --- Summary What is already known about this topic? Preventing COVID-19 in colleges and universities requires mitigation strategies addressing on-and off-campus congregate living settings, extracurricular activities, and social gatherings. What is added by this report? At the start of the 2020-21 academic year, COVID-19 cases increased rapidly at an Arkansas university. Network analysis indicated that 91% of gatherings were associated with fraternity or sorority activities. Recruitment events held virtually were associated with fewer cases than those held in-person. What are the implications for public health practice? Given the potential for rapid SARS-CoV-2 transmission in on-and off-campus settings and activities, colleges and universities should work with local health departments and student organizations to ensure compliance with mitigation guidelines. available data on place of residence and involvement in onand off-campus activities was used to describe university A's transmission network, potential gatherings where transmission might have taken place, and links between nodes. The network visualization tool depicted algorithm-detected gatherings to identify links indicating likely recent contact. Visualized in real-time, information from such links and networks could support implementation of targeted mitigation activities, such as isolation of cases and quarantining of contacts. The findings in this report are subject to at least four limitations. First, incomplete case investigations resulted in missing or unknown data and exclusion from the network analysis. Second, housing and event attendance might not approximate transmission histories. Third, many cases were identified during mass testing events, and event advertisement and location could have resulted in a biased sample. Finally, data were not collected from uninfected persons or on adherence to mitigation strategies, such as social distancing, mask use, and hand hygiene. Because of the potential for rapid transmission of SARS-CoV-2 in on-and off-campus university settings, student organizations could help ensure compliance with CDC-recommended COVID-19 mitigation measures, such as limiting the size of social gatherings, adhering to social distancing recommendations, requiring mask use, improving hand hygiene, and increasing testing. Encouraging more virtual activities, including those related to fraternity and sorority rush week, might help minimize the risk for transmission on university and college campuses. To ensure compliance with mitigation measures, health departments should work together with student organizations and university leaders. Corresponding author: Kristyn E. Vang, [email protected].

Preventing transmission of SARS-CoV-2, the virus that causes coronavirus disease 2019 , in colleges and universities requires mitigation strategies that address on-and off-campus congregate living settings as well as extracurricular activities and other social gatherings (1-4). At the start of the academic year, sorority and fraternity organizations host a series of recruitment activities known as rush week; rush week culminates with bid day, when selections are announced. At university A in Arkansas, sorority rush week (for women) was held during August 17-22, 2020, and consisted of on-and offcampus social gatherings, including an outdoor bid day event on August 22. Fraternity rush week (for men) occurred during August 27-31, with bid day scheduled for September 5. During August 22-September 5, university A-associated COVID-19 cases were reported to the Arkansas Department of Health (ADH). A total of 965 confirmed and probable COVID-19 cases associated with university A were identified, with symptom onset occurring during August 20-September 5, 2020; 31% of the patients with these cases reported involvement in any fraternity or sorority activity. Network analysis identified 54 gatherings among all linkages of cases to places of residence and cases to events, 49 (91%) were linked by participation in fraternity and sorority activities accounting for 42 (72%) links among gatherings. On September 4, university A banned gatherings of ≥10 persons, and fraternity bid day was held virtually. The rapid increase in COVID-19 cases was likely facilitated by on-and off-campus congregate living settings and activities, and health departments should work together with student organizations and university leadership to ensure compliance with mitigation measures. University A began the academic year on August 24, offering in-class and virtual instruction for approximately 20,000 students; a majority of students used virtual instruction. Before and during the start of the academic year, students might have participated in on-and off-campus fraternity or sorority activities. Cases were identified by the university during an ADH-sponsored testing event (September 1-3) or linked to the university by ADH. Case data were reported to ADH by clinic staff members at university A and stored in an electronic database (REDCap, version 8.8.0; Vanderbilt University). Using a standardized questionnaire, ADH nurses interviewed persons with university-associated COVID-19

Introduction Achieving health equity, or fair opportunities for the best possible health for all members of society regardless of their race, ethnicity, and other social determinants, has been a priority for policy and research [1,2]. Given the persisting trends of health inequities in the USA [3,4], different health and social science research fields, from basic sciences to epidemiology and public policy, have raised the call for increased attention to and awareness of health equity research through an antiracist lens [5][6][7]. There are still substantial gaps in our knowledge regarding the complexity of interrelated and multifaceted social, health, and systemic determinants of health equity, and effective actions to achieve it [8][9][10][11]. Health equity research is multidisciplinary and translational, as it recognizes the multilevel and embedded nature of context and unique needs and characteristics of populations [12]. To effectively tackle this complexity, translational, and transdisciplinary research is needed to meaningfully engage and activate various partners [13][14][15][16][17]. Researchers from various disciplines, administrators, and implementers of interventions and policies, and community members facing health inequities should be engaged to make sure the questions, data, methods, and findings are relevant and useful to real-world settings [13]. As a result, the majority of the Clinical and Translational Science Awards consortium leaders identified health equity research as a priority and stated their commitment to facilitate more health equity research supported by the CTSA programs [18]. Expertise and interest in health equity, similar to other translational concepts, is distributed across various basic sciences, clinical, and population health disciplines, crossing formal organizational and departmental barriers and forming invisible communities [19]. Invisible communities [20] are comprised of researchers who collaborate and cooperate based on their mutual interests [19] and may go beyond traditional institutional boundaries. These communities play an essential role in the production of knowledge and generation of innovations [21]. Network analysis offers a novel methodology for understanding the structure and configuration of these invisible communities [22,23]. By identifying the invisible communities and understanding their compositions and dynamics, academic institutions can develop tailored interventions to empower and bridge communities of expertise and facilitate knowledge sharing and dialog, towards building capacity for translational and crossdisciplinary innovations. We aimed to understand the distribution of racial and ethnic health equity research, education, and social/administrative activities in an academic institution. We mapped the network of health research scholars active in racial and ethnic health equity research and practice at the University of Rochester Medical Center , based on a peer nomination survey. --- Materials and Methods This analysis underwent review by the University of Rochester Research Subjects Review Board and was deemed exempt. Context: The confluence of our institution's commitment to the development of a Learning Health System and to our Equity and Anti-Racism Action Plan , resulted in a high degree of coordination and collaboration supported by our Clinical and Translational Science Institute . In February 2021, UR CTSI established an Office of Health Equity Research to serve as a central hub for health equity research excellence, advancing health equity, promoting new research partnerships, providing pilot funding, and developing training/technical resources, as a key part of the EARAP. We identified UR faculty with experience and/ or interest in racial and ethnic health equity, to develop an inventory of potential users and contributers to this office. Snowball Approach: To map the distribution of expertise in racial and ethnic equity research and practice, we conducted a snowball survey of faculty members at the School of Medicine and Dentistry and School of Nursing of University of Rochester with experience and/or interest in promoting racial and ethnic health equity. Each respondent could nominate other investigators who also had experience and/or conducted research/ educational/capacity-building activities related to racial and ethnic health equity. Furthermore, the respondents were asked to identify their areas of expertise in research, education, or social/administrative activities to improve racial and ethnic equity in health, based on the following classification: • Research: involving observational research, that focuses on quantitative or qualitative study of the extent, mechanism, and impact of racial and ethnic disparities on the health of individuals or populations , and interventional research, focusing on the effect of interventions to improve racial and ethnic health equity. • Education: involvement in formal and informal training to students, staff, faculty members, and also in the communities. • Social/administrative activities: activities to improve racial and ethnic equity in health . Each respondent could choose multiple expertise categories, if relevant. The respondents could also describe their research and practice in the free text boxes provided for each expertise category . We calculated the frequency of various health equity expertise categories identified by respondents and used Cohen's Kappa to indicate the involvement of respondents in multiple categories, beyond chance. The analysis was carried on in STATA 15.1 program [24]. Nomination network analysis: We transformed the identified name lists into a nomination network, in which actors were connected to each other by nomination ties . We limited the analysis to part-time or fulltime faculty members at the School of Medicine and Dentistry and School of Nursing of the University of Rochester, that are parts of the University of Rochester Medical Center . We calculated structural indicators of the nomination network, including density , and reciprocity , and centrality of network actors . Given the inherent dependence of social relations in a network, traditional general linear model techniques are not suitable, and there is a need to adjust the variance of outcomes based on the dependence in the network. We used permutation techniques that involve random shuffling of the rows and columns of the matrices several times to develop a distribution of all possible network compositions under null hypothesis [25]. We used linear Table 1. The areas of expertise of participants related to racial/ethnic equity "quantitative or qualitative research about the extent, mechanism, and impact of racial and ethnic disparities on the health of individuals or populations ." • extent, mechanisms, and impact of racial and ethnic disparities • extent, mechanisms, and impact of different types of racism 78 • 62 regression to predict actors' centrality by their expertise categories. We applied quadratic assignment procedure logistic regression [26] to predict nominations between pairs of actors based on the similarity of expertise categories, as well as being affiliated with the same departments. The quantitative nomination network analysis was conducted using version 6.7 of Ucinet program for Windows [27]. We thematically analyzed the answers to the open-ended questions regarding respondents' involvement in various expertise categories, using a descriptive qualitative approach [28]. We used the classification of expertise fields mentioned above as the initial coding framework. Within each category, we thematically analyzed the subject matter topics using an inductive process. The qualitative network analysis was conducted using Dedoose software [29]. --- Results --- Characteristics of participants: In the first round of the survey, invitations were sent to 15 experts who were known by the research team as investigators involved in health equity research and practice. We received 10 responses, nominating 46 other investigators. In the second round, we sent the survey to 51 individuals, including the recently nominated investigators, as well as five of the initial contacts who did not respond to the first invitation. We received 35 additional responses, nominating 33 new investigators. Subsequently, in the third, fourth, and fifth rounds, we contacted 52, 56, and 38 individuals, respectively, who nominated 33, 30, and 14 new investigators. In the sixth and final round of the survey, we contacted 18 individuals and received 9 responses, nominating 3 new investigators. We stopped the survey after the sixth round due to information redundancy, with a total of 121 respondents. The respondents were affiliated with 20 different departments across three schools of Medicine, Dentistry, and Nursing. The most popular among the respondents were Pediatrics , Public Health Sciences , department of Medicine , Psychiatry , Obstetrics and Gynecology , and the School of Nursing . Areas of expertise: Of the respondents, 64% were involved in research related to the extent, mechanism, and impact of racial/ ethnic disparities, 48% were involved in interventional research, 55% were involved in education, and 50% were involved in social/ administrative activities . Among respondents who were involved in research, a larger percentage focused on racial and ethnic disparities than studying various types of racism . Social/administrative activities were also more likely to happen at university and local community levels. The overlap between expertise categories that were identified by respondents was small . The statistically significant and positive overlap between pairs of expertise categories was observed between education with social/administrative activities , as well as between research on interventions and education . It means that there was a significant and positive overlap between choosing education and social/administrative activities, as well as between education and interventional research. Although not statistically significant, the other observed positive Kappa agreement was between doing research on extent and outcomes and research on interventions . Structural indicators of the nomination network: Visual inspection of the nomination network map of health equity investigators showed a well-connected network with a few actors who were visually more central and had no noticeable clusters. The most central actors were involved in multiple expertise categories, as represented by the pink color in Fig. 1a-d. The nomination network had a density of 2% and a reciprocity of 15%. Predictors of nomination: Findings from the QAP regression analysis to predict nomination ties based on shared expertise categories and shared departments among pairs of actors are provided in Table 3. Respondents were more likely to nominate someone if both were involved in research and education and both were affiliated with the same departments . The regression analysis to predict actors' centrality by their expertise categories showed that being involved in health equity research was the only significant predictor of actors' centrality, with a coefficient of 2.1 . Thematic content analysis: We classified and aggregated the free text answers regarding the respondents' areas of expertise and presented the classification in Fig. 2 andAppendix 2. The main themes related to the research on the extent, mechanisms, and impact of disparities included studying access , disparities in health outcomes , systemic disparities , and environmental disparities . Themes related to the extent, mechanisms, and impact of racism included the experience of racism based on gender identity and sexual orientation, the experience of Sign Language users, and racism in the context of opioid prescription. Some themes of interventions to address disparities included strategies to improve access , training and education, resilience building, use of Themes related to education included course development , small group discussions, workshops, lectures, podcasts, and mentoring of medical trainees, healthcare providers, and health researchers. We classified social/administrative activities based on the scope. At the university level, themes included participating in advocacy groups and diversity committees and improving inequities in students' admission process. In the local community level, themes included community outreach and partnership. In the regional/ state level, activities included participation in regional committees and taskforces, giving lectures at different agencies, collaboration in state-wide health promotion activities, and restructuring electronic medical records to capture social determinants of health. In the national level, themes included participating and organizing national forums, committees, and conferences, conducting antiracism training, social media activities, and scientific publications. --- Discussion This study mapped the network of scholars active in racial and ethnic health equity research and practice at an academic institution. We found that central actors in the nomination network were involved in multiple expertise categories. Doing research was the only significant predictor of popularity . Scholars were more likely to identify others based on homophily . Respondents were more involved in research about racial and ethnic health equity and antiracism, than participating in local, regional, or national social/administrative activities; and being involved in social/administrative activities did not significantly associate with popularity. We also provided a thematic analysis of the research, education, and social/administrative activities, which could be used as a preliminary frame for the classification of health equity research and practice [30,31]. To advance racial equity and antiracism in academia, research, education, and social/administrative activities should go hand in hand. Education is a critical pillar in advancing equity. Academics lack an optimal recognition and understanding of racism and have an under-developed vocabulary to communicate about it [31,32]. In addition, academics usually under-recognize [33] the importance of race and racism and address them as external to their institution [5]. In our study, two groups of respondents who were involved in interventional research and in social/administrative activities had little overlap, while individuals who were involved in education had significant overlap with both. This implies that education can play a bridging role in connecting these lessconnected specialty clusters. Our thematic analysis showed that educational activities are diverse, ranging from academically oriented education to communityfocused , and could serve different audience groups. Further involvement of scholars involved in research and social/administrative activities in equity education will provide opportunities for collaboration across disciplines and will improve the relevance and suitability of educational activities for different audience groups. Racial equity and antiracism in academia require direct engagement of researchers in activities to raise awareness, build capacities, motivate and involve the leadership, and build trust and dialog with communities. However, academics tend to reflect more than they act [34]. Social and administrative activities are inseparable from research and education, as the engagement of community partners and minoritized researchers are critical to guide the direction of research and educational activities [31]. In our study, we found that even though individuals involved in social/administrative activities significantly overlapped with the ones involved in education, there was little overlap between social/ administrative activities and research. In addition, despite other categories, co-involvement in social/administrative activities was not associated with peer nomination. This implies that scholars who are involved in social/administrative activities are dispersed and disconnected, compared to the ones who are involved in research. This could be because of the diverse nature of such activities, including membership in professional boards, involvement in systemic reforms, and collaboration with communitybased organizations and outreach activities, as reflected in the thematic analysis. On the other hand, we found that involvement in social/administrative activities did not contribute to centrality, which implies that these activities are not usually well recognized and celebrated by the institution. Social/administrative activities to promote racial equity and antiracism are less likely to result in publication and are often led by racially minoritized faculty members, which deepens the equity gaps in faculty retention and promotion [35,36]. Recent attention to this equity gap has resulted in movements by various academic institutions to recognize such activities in the tenure and promotion mechanisms [37,38], including at the University of Rochester [39]. In addition to these efforts, it seems that there is room for better translational synergy between the research and social/administrative activities, bridging different domains of translational spectrum [12]. We found a few investigators who were nominated the most by their peers. These central actors were involved in multiple areas, including research, education, and social/administrative activities. By involvement in multiple expertise categories, these central actors bridged expertise clusters. Their popularity and bridging role are two important characteristics of effective organizational change champions [40]. Network analysis was a useful tool to identify potential champions who could be engaged and motivated [41] to lead the institutional movement towards antiracism and racial equity. There were a few limitations to this study. We limited the analysis to faculty members who self-identified racial equity as their fields of expertise and practice and nominated others with similar expertise. This method might have missed individuals who were not connected to the respondents, perhaps due to belonging to distinct professional clusters. In addition to nomination networks, future studies may also focus on collaboration networks among researchers using grant submission and coauthorship data. The indicators of centrality in our analysis are based on nomination by other experts, through a snowball process. Other central actors might have been identified if we surveyed all faculty members regardless of their expertise and involvement in racial equity research and practice. We limited the study population, for practical reasons, to the health-related faculty members, hence under-represented basic scientists, and nonfaculty investigators and educators who are also important part of the community at URMC. Finally, this study presents the patterns of nomination in one institution, which may have little generalizability. The findings of our study could inform institutional activities to promote collaboration among faculty members about racial and ethnic health equity research and education efforts. The findings highlighted the diversity of education and social/administrative activities and their current disconnect from research in this academic institution. Our thematic analysis of fields of expertise can also serve as a preliminary conceptual framework for racial equity research and practice at academic institutions. We suggest the following institutional activities as potential interventions informed by these findings: health investigators involved in observational and interventional research on health equity should be further invited and engaged in educational programs. This could happen through guest lecturing to present the research findings to the broader audience and also development of research excerpts to be reached to and used by broader communities. -The findings of this nomination analysis have already informed the development of a repository of investigators involved in health equity research. These repositories will facilitate network building and peer recognition, and future collaboration. -Involvement of researchers in education and social/administrative activities should be further recognized and celebrated by the academic institutions and be reflected in faculty promotion and evaluation frameworks. development of a taxonomy of research and practice in racial/ ethnic equity is an important step towards the development of sub-fields in an emerging research area. Recognition of professional clusters will also inform future funding mechanisms and priority setting for health equity research and practice. Our findings can contribute to the development of a classification system. Supplementary material. The supplementary material for this article can be found at https://doi.org/10.1017/cts.2023.555. Acknowledgments. The project described in this publication was supported by the University of Rochester CTSA award number UL1 TR002001 from the National Center for Advancing Translational Sciences of the National Institutes --- Competing interests. The authors have no conflicts of interest to declare.

Introduction: Health equity research spans various disciplines, crossing formal organizational and departmental barriers and forming invisible communities. This study aimed to map the nomination network of scholars at the University of Rochester Medical Center who were active in racial and ethnic health equity research, education, and social/administrative activities, to identify the predictors of peer recognition. Methods: We conducted a snowball survey of faculty members with experience and/or interest in racial and ethnic health equity, nominating peers with relevant expertise. Results: Data from a total of 121 individuals (64% doing research on extent and outcomes of racial/ethnic disparities and racism, 48% research on interventions, 55% education, and 50% social/administrative activities) were gathered in six rounds of survey. The overlap between expertise categories was small with coincidence observed between education and social/administrative activities (kappa: 0.27; p < 0.001). Respondents were more likely to nominate someone if both were involved in research (OR: 3.1), if both were involved in education (OR: 1.7), and if both were affiliated with the same department (OR: 3.7). Being involved in health equity research significantly predicted the centrality of an individual in the nomination network, and the most central actors were involved in multiple expertise categories. Conclusions: Compared with equity researchers, those involved in racial equity social/ administrative activities were less likely to be recognized by peers as equity experts.

Background Sheltered housing is associated with quality-of-life improvements for individuals with serious mental illness . However, there are equivocal findings around safety outcomes related to this type of living condition. --- Aims We aimed to investigate raw differences in prevalence and incidence of crime victimisation in sheltered housing compared with living alone or with family; and to identify groups at high risk for victimisation, using demographic and clinical factors. We do so by reporting estimated victimisation incidents for each risk group. --- Method A large, community-based, cross-sectional survey of 956 people with SMI completed the Dutch Crime and Victimisation Survey. Data was collected on victimisation prevalence and number of incidents in the past year. --- Results Victimisation prevalence was highest among residents in sheltered housing compared with persons living alone or with family . We found that sheltered housing was associated with increased raw victimisation incidence increasingly live within the community. Broadly, two housing strategies have been developed, in the hopes of facilitating recovery by integrating SMI individuals in the community. 1 The options consist of living independently by making use of out-patient support services or living in a sheltered housing arrangement, sharing some spaces with others and having staff that are available during working hours or overnight. 2 Although different terms are used in the literature , we discuss sheltered housing as defined by persons with SMI who live in shared community or halfway housing. The living conditions and therapeutic environment of sheltered housing in The Netherlands, the country of this study, are very similar to those in other Western countries. 3,4 This housing arrangement exists to provide people with SMI with an opportunity to reside within the general community and retain the benefits of a semi-monitored environment. 5 Sheltered housing has received much attention from studies investigating its merits, but less from quantitative investigations. 6 Equivocal outcomes linked to sheltered housing Research has found produced equivocal results on the quality and experience of sheltered housing. There appears to be a positive trend in quality-of-life outcomes, including better social functioning and living conditions, compared with alternative high-support accommodations, including the hospital. [6][7][8] More-focused qualitative research suggests that these trends are attributable to residents feeling safe and distancing themselves from past stress and trauma. 9,10 At the same time, recommendations have been made for widespread implementation of sheltered housing in European and international psychiatric rehabilitation programmes. 11,12 However, literature has also reported on high crime victimisation prevalence among English people with SMI who reside in sheltered housing. 13 Longitudinal follow-up on the safety of sheltered housing found that the enthusiasm of newfound security faded throughout the first year of residence, as over time, substance misuse relapses and conflicts with neighbours reintroduced stress and mistrust among residents with SMI. 14 This potentially reflects the general tendency for vulnerable populations to be housed in less-secure urban areas, satirically self-described in some contexts as 'mental illness ghettos' by tenants. 14,15 Victimisation is likely to lead to worsened symptoms and adverse mental health, thus disrupting the recoveryfocused goals upheld by the sheltered housing initiatives. 16 Gaps in the quantitative sheltered housing literature These reports have stimulated commentary, inviting more quantitative research into the quality indicators and outcomes associated with sheltered housing. 17 Although the literature so far describes in rich qualitative detail the experiences of individuals with SMI, very little research is available to quantify the threats associated, especially with regard to identifying demographic or clinical groups at increased risk of adversity. Gender is a prominent example of a factor often identified in forensic research as particularly relevant to victimisation and its complex implication on the consequences that follow. 18 Unfortunately, little such information is available on whether residents of sheltered housing report different degrees of victimisation according to gender, or across other basic demographic factors commonly studied in victimisation literature . As patients with SMI are at a much higher risk of becoming victims of crimes, there is potential for these risk factors to produce particularly salient differences in victimisation outcomes. 20 Considering that inhabitants describe safety as the most important factor for their well-being at home, it is important to identify risk groups and establish the incidence and prevalence of crime victimisation for persons with SMI who are living in sheltered housing. 14,21 --- Aims The current large-scale, cross-sectional study aims first to investigate differences in prevalence and incidence of victimisation in the past year in patients with SMI across three different types of living accommodation: living with family, living alone and living in sheltered housing. Furthermore, we also investigated which clinical and demographic variables are potential risk factors for victimisation, particularly emphasising potential gender differences. --- Method Design The current study is embedded in the Victimization in Psychiatric Patients study, a cross-sectional epidemiological survey of a large, random community sample of 956 patients with SMI in The Netherlands. 20 This study was approved by the Medical Ethics Committee of the Erasmus Medical Center, Rotterdam . Written informed consent was obtained from all participants. --- --- Procedures As previously depicted by Kamperman et al, 20 data on crime victimisation and clinical and demographic variables were obtained in a structured, computer-assisted, face-to-face interview. Respondents received a €20 cash incentive at the end of the interview. Interviews took 75 min on average , and were carried out at the respondents' discretion in their home or at the mental healthcare institution. Interviewers were master's level social scientists with training in conducting the interviews with patients with SMI, as supervised by an experienced coordinator. --- Instruments --- Main determinant and risk factors Participants self-indicated whether they lived independently or in a sheltered housing arrangement, with the latter category including any recovery-focused halfway or community housing for outpatients of SMI services. Those who lived independently were subdivided into living alone or living with family, which included a partner, child or member of extended family. Those who indicated living in sheltered housing provided an address, which was crosschecked to validate a sheltered housing establishment in fact operates on that address. Such establishments can vary in the number of people sharing space and degree of monitoring by staff. No additional data was available on these specifics for those who indicated living in sheltered housing. Clinical characteristics were operationalised with standardised instruments. Symptoms of post-traumatic stress disorder were assessed with the Self-Rating Inventory for Posttraumatic Stress Disorder. 22 Perpetration of physical violence over the past year was assessed by the physical assault subscale of the Conflict Tactics Scale short form. 23 The Dimensions of Anger Reactions Scale 24 was used to assess trait anger. Substance misuse was assessed with the Dutch version of the 12-month drug and alcohol use questionnaire of the European Monitoring Centre for Drugs and Drugs Addiction. 25 For this study, we operationalised alcohol misuse as drinking more than six consumptions per day at least one time over the past 6 months. Drug misuse was operationalised as using one or more types of drugs, or using medication without a doctor's prescription. All instruments exhibited good reliability . See the Supplementary Material available at https://doi.org/10.1192/bjo.2021.57 for detailed information on dichotomisation, reliability and construct validity for each clinical instrument. Sociodemographic characteristics were collected on gender, age, ethnicity, marital and employment status and educational level. Following the definition of the Dutch government, 26 ethnicity was classified on the basis of country of birth and parents' country of birth. --- Outcome Prevalence of crime victimisation and the number of incidents in the past year were assessed with the crime victimisation scale of the Dutch Crime and Victimisation Survey . 27 The IVM crime victimisation scale strongly resembles the International Crime Victimization Survey. 28 The IVM crime victimisation scale consists of 14 screening questions on various types of property crime, personal crime and vandalism. These include burglary, attempted burglary, bicycle theft, pick-pocketing, robbery, theft , vandalism , sexual harassment or assault, threatened with violence, physical assault and crime . For each reported incident in the preceding year, detailed information on the time and number of incidents, setting and perpetrator is assessed. To minimalize the effect of telescoping, the respondents are asked to recall incidents over the past 5 years, before recalling incidents over the past year. There are no traditional reliability and validity scores of the IVM crime victimisation scale. 27 --- Statistical analysis The main determinant of the current analysis was the living condition factor and its predictive value on victimisation. As such, we report on univariable, multivariable and interaction coefficient terms related to the living conditions predictor. A multivariable Poisson regression was conducted to investigate the role of living conditions on estimating victimisation. We utilised a logit link to model the count outcome, using the 'glm' package developed by the R core team . In reporting on the multivariable Poisson regression consisting of multiple interaction terms, we present effect plots of estimated number of victimisation incidences to facilitate intuitive interpretation of risk groups identified in the model, instead of focusing exclusively on coefficient values, which are difficult to interpret, if at all meaningful. 29 The reported results were stratified by gender, to meaningfully inspect differences between genders. All reporting on significant differences in estimated incidents are based on 95% confidence intervals. A model-building approach was taken in understanding the effect of living conditions on victimisation incidence. We used a logistic regression for unadjusted prevalence. Next, we estimated a univariable Poisson regression model to investigate the unadjusted association between living conditions and victimisation incidence. This univariable model contained dummy-coded living conditions variables corresponding to a sheltered housing versus family accommodation comparison and living alone versus family accommodation comparison. In a second step, a full multivariable logistic model was fitted, to include gender, education, comorbid drug use in the past year, comorbid alcohol misuse in the past 6 months, comorbid PTSD, perpetration of assault in the past year and dispositional anger. These variables were chosen on a theoretical basis, as the literature has previously identified them as relevant to victimisation. 19 They were also modelled as interactions with the living conditions dummy variables, to investigate their effect specifically within each living condition. Additionally, age, marital status and employment were included only as covariates. In the last step of variable selection, we applied a backward-model selection based on the Akaike information criterion , to find a more parsimonious set of variables explaining the data. We chose the AIC over other information criteria, like the Bayesian information criterion , as it is relatively liberal in allowing model complexity. 30 In total, 5% of data were missing; 52 cases had missing data. Five variables accounted for missing data: alcohol misuse , comorbid PTSD , perpetration of violence , housing and dispositional anger . Missing data patterns were explored to assess whether the missing-at-random assumption was met for the victimisation outcome, before dropping the missing data from further analysis. Descriptive plots of missing data patterns are presented in Supplementary Figure 1. Multicollinearity between predictor variables was assessed with variance inflation factors , as implemented in the 'car' package. 31 Six cases were identified as outliers, reporting extremely high numbers of victimisation incidents , and were additionally removed from analysis. --- Results --- Sample The full sample consisted of 956 patients with SMI: 608 men and 348 women . Mean age was 44.7 years . The respondents' demographic and clinical characteristics were consistent with nationwide figures for patients with SMI in The Netherlands. 32 Further demographic characteristics of the sample used for the main analysis are presented in Table 1. --- Prevalence and incidence, univariable and multivariable effects In the first step of the analysis, investigating the unadjusted prevalence and incidence revealed that persons using sheltered housing services have the highest victimisation prevalence , followed by persons living alone and persons living with family . Compared with persons living with family, persons living in sheltered housing or living alone reported significantly more victimisation incidents in the past year, at a higher incidence rate of 2.80 and 1.87 times, respectively , as shown in Table 2. Following the strong differences in prevalence and incidence between the living condition categories, we used a multivariable Poisson regression to identify high-risk groups. We included gender, education, comorbid drug use in the past year, comorbid alcohol misuse in the past 6 months, comorbid PTSD, perpetration of physical assault in the past year and dispositional anger as both predictors and interaction terms with the living condition term. Additionally, age, marital status and employment status were included as confounder variables. The backward-model selection approach revealed that removing the employment variable and dispositional anger interaction term produced the model with the relatively lowest AIC. This model was assumed as the best fit to the data, and results hereafter are reported from it. When comparing shelter with family accommodation, we observed a significant interaction effect between shelter and family accommodation and gender , education , comorbid alcohol misuse in the past 6 months , assault perpetration in the past year and comorbid PTSD . Comparing living alone with living with family produced similar significant interactions, but the living conditions and comorbid drug misuse interaction was also significant . The implications of these interactions is probed next, using estimated incidents for each category. A full coefficient table and plot for the final model are presented in Supplementary Table 1 and Supplementary Figure 2. Next, to examine the interactions with the living condition determinant, we report on the estimated incident counts obtained from the multivariable incidence model, visualised in Figure 1. It was estimated that women living alone reported the most incidents of victimisation in the past year , significantly more than those in sheltered housing . Victimisation incidents of those living with family were between the other two groups . Women with a comorbidity of drug misuse were estimated to be considerably more victimised when living alone . When reporting alcohol misuse, women became more likely to be victimised when living with family than when living in sheltered accommodation . Next, women who selfreported perpetrating assault had more estimated victimisation incidents in sheltered housing than non-perpetrators , but having slightly lower victimisation estimates than those perpetrators living with family or alone . Comorbid PTSD exhibited similar equivalence across all living conditions, thus all victimisation estimates were around 1 incident in the past year . Finally, clear trends in victimisation were found across education categories, such that more-educated participants were at more risk of becoming victims in sheltered accommodation, and were at less risk when living with family. For instance, highly educated women in sheltered housing were estimated to report 2.70 incidents , whereas the estimate when living with family was lower, at 0.34 . When examining the estimates for male patients, we found that in the reference category, there were no big differences between living with family , living alone and living in a sheltered home . As with women, men were especially at risk of victimisation in the past year if they reported drug use in the past 6 months and lived alone . However, victimisation in sheltered living became much likelier for men if they had comorbid PTSD , were perpetrators of assault and especially if they were highly educated . Again, there was a trend of less victimisation for educated men living with their family. Highly educated men were estimated to have 0.21 incidents in the past year when living with family, the lowest estimate to emerge from the analysis. --- Discussion The current study investigated the association between living in sheltered housing and becoming a recent victim of crime in a large-scale community sample of patients with SMI. Several distinct patterns of results were revealed by the analysis. First, the highest overall victimisation prevalence was reported by those living in sheltered housing, where half of the patients reported becoming a crime victim at least once in the past year, as compared with 38% if living with family or 43% if living alone. Regarding incidence rates, a striking trend was that the more educated patients were, the more likely they were to be victimised in the sheltered housing category, and especially so for men. Victimisation among sheltered housing residents was also particularly high for men with PTSD or if they were a perpetrator of assault themselves. Women were less victimised when living in sheltered housing compared with men, often reporting less incidents when living there than when living alone or with family. Finally, both men and women living alone had a higher risk of being victimised if reporting alcohol or drug problems in the past year. Previous research has reported mainly on positive outcomes linked to patients with SMI living in sheltered housing. 8 A direct comparison of the current results is possible with a previous report on victimisation of patients with SMI in The Netherlands, which reportedin contrast to our studythat those living in sheltered housing were comparatively less victimised than alternative forms of out-patient accommodations, and indeed less than the prevalences reported here . 33 There are two potential reasons for this discrepancy with the current sample. First, the patients in that sample reported considerable substance use, almost half in the sheltered housing group . This highlights the present finding that sheltered housing appears to be associated with less victimisation for those with drug and alcohol problems when compared with more deinstitutionalised settings, like living alone or with family. Additionally, however, the sample size of that study was much lower than that of the current study. Considering the current study only had a relatively small proportion of highly educated individuals, an even smaller study would not be able to pick up on the relatively rare, yet large effect of being highly educated in sheltered housing. 34,35 Two hypotheses on the effect have been offered: educated people make more attractive targets because of perceived higher status, 36 or alternatively, they are more ready to define an event as a crime. 37 Because of small numbers in the current study and the lack of previous verification of the abovementioned hypothesis, this relationship needs further study. Within SMI samples, much forensic research has been dedicated to investigating the association between the vulnerability of becoming a victim of crime and clinical factors like PTSD. 38 No study so far had examined the implications for those living in sheltered housing. The current study links that line of research with the current research question by pinpointing male patients with SMI with comorbid PTSD as a particularly vulnerable subgroup. This finding is particularly troubling, as established theoretical models point to victimisation as an important mediator in how PTSD positively relates to worsened psychiatric symptoms. 16 Thus, sensitivity to monitoring PTSD in sheltered housing is vital, as these individuals, and specifically men, are at a high risk of becoming victims of crime. Finally, demographic factors were also found to be associated with how often patients were victimised when living in sheltered housing. A striking result of the current study was that women living in sheltered housing tended to be less vulnerable to crime victimisation in the past year compared with men. In general, women with SMI are more likely to become victims of family violence compared with men, who are more likely to be victims of crime perpetrated by non-familial offenders. 39 The unfortunate reality of women being abused more by their own family is one explanation of the gender differences in the current study. The relatively depressed economic conditions around sheltered housing, a wellknown risk factor for male victimisation, might further explain gender differences. 39 These differences were especially evident for women reporting substance misuse, where victimisation prevalence was considerably lower in sheltered housing compared with living alone or with family. To a lesser extent, this was also true for men who reported substance misuse. Substance users tend to live in neighbourhoods with poor economic conditions, chronic disease and social disorganisation, 40 which predisposes them to crime and trauma and worsens their mental health. 41 This study provides clear evidence that sheltered housing is associated with less crime victimisation for individuals with SMI who have drug problems. Of note is that our model was adjusted for perpetration, therefore these victims were not also self-reported perpetrators. Future studies could thus establish sheltered housing as a potentially beneficial intervention for providing safety compared with the alternative, especially for women with drug or alcohol problems. The foremost strength of the current study was its sample size: a representative random sample of many individuals was obtained, so that we could include rare characteristics of patients with SMI in our analysis . Despite the many interaction terms included in our model, we could still obtain precise estimates for differences between each group. Robust as our results are, an important limitation is that they are confined to the sheltered housing system of The Netherlands, which can be heterogeneous itself in terms of monitoring and support. It is for future research to determine whether the findings are applicable in culturally and administratively different parts of the world, as well as within specific subcategories of sheltered housing. Additionally, the conclusions from a cross-sectional sample such as this one are correlational in nature, and thus prone to confounding. We do not make claims about the causal structure of the effects, nor can we establish in the current sample whether victimisation took place inside or outside of the housing premises. Nevertheless, this is the largest and most detailed study into the topic to date, allowing for robust prediction, if not causal understanding, of victimisation. There is potential for many further research questions to be formulated based on the current findings. In conclusion, the current study highlights the need for higher awareness and better surveillance of victimisation among residents of sheltered housing, to better facilitate a recovery-enabling environment for patients with SMI. Individuals with higher education, and especially men with a PTSD diagnosis, are particularly vulnerable to becoming victims when living in sheltered housing. At the same time, individuals with comorbid substance misuse report less crime victimisation when living in sheltered housing, particularly so for women who have used drugs in the past year. Further intervention research is needed in this area, to inform on the specifics of how a monitoring and prevention intervention might work in practice. Although the current results provide insight into which people might need to receive increased attention, our study does not address differences across sheltered housing establishments. One area of research could focus on neighbourhood-level characteristics and composition of sheltered housing establishments, and how they contribute to victimisation of patients with SMI. In light of the findings we have available, we conclude that the reality of sheltered housing and its implications for the crime victimisation of its inhabitants are more complex than can be captured by a purely positive or negative blanket statement. It can indeed be expected to provide a secure space suited for the recovery process of a given individual with SMI, but an important caveat to note is that some individuals inhabiting such accommodations are at a concerning risk of victimisation. --- Data availability Data are stored at the institutional database of the Erasmus Medical Centre in Rotterdam, The Netherlands. The data-sets on which the analyses are based are available on request to the Local Ethics Committee of the Erasmus Medical Centre in Rotterdam . The code used to analyse the current data-set is available in the Open Science Framework repository . --- Supplementary material Supplementary material is available online at http://doi.org/10.1192/bjo.2021.57 --- --- Declaration of interest None.

family; 1.87, 95% CI 1.59-2.20 compared with living alone). Incidence was especially high for some high-risk groups, including men, people with comorbid post-traumatic stress disorder and those with high levels of education. However, women reported less victimisation in sheltered housing than living alone or with family, if they also reported drug or alcohol use.The high prevalence and incidence of victimisation among residents in sheltered housing highlights the need for more awareness and surveillance of victimisation in this population group, to better facilitate a recovery-enabling environment for residents with SMI.

Introduction Groups openly advocating anti-immigrant sentiments have made a remarkable comeback in recent years in OECD countries. This trend, which began in the 1980s, can even be witnessed in countries where multiculturalism was once celebrated as a core value defining the national identity. Consider the Netherlands, where the PVV led by Geert Wilders increased its number of seats in parliament from 9 seats in 2006 to 24 in the 2010 elections. Another case in point is Sweden, where, for the first time in the country's history, the extreme right secured a seat in the national parliament in the 2010 elections. Attempts to understand the rise of support for anti-immigrant parties traditionally focus on macro-economic downturn and poor people or working class people. This is because realistic conflict and competition over scarce resources is experienced most strongly in such an economic climate [1], and among those with a lower economic position in society [2][3][4][5][6][7]. While we do not dispute that economic downturn or poverty can contribute to anti-immigrant sentiments, there is a growing body of work that suggests that this relationship is not as straightforward as often believed. In particular, while there is considerable evidence that economic downturn can be associated with harsh attitudes towards immigrants, there is also evidence for the opposite: that economic prosperity can produce harsh attitudes towards minorities-so called relative gratification [8]. Similarly, while negative attitudes towards minorities are at times more pronounced among the poor and working classes, such attitudes are under some conditions particularly prevalent among the wealthier groups in society, or those who expect prosperity in the future [9][10][11]. Indeed, while support for anti-immigrant parties flourished after the Global Financial Crisis , many commentators have noted that the rise of support for extreme right in Western Europe started long before the GFC-at a time when the economy in most Eurozone countries was booming and, relatively speaking, people saw their income and affluence increase [12][13][14]. To illuminate these opposing processes, we take a two-pronged approach. We start by demonstrating empirically that both relative deprivation and relative gratification processes relate to opposition to immigration. Second, in an attempt to better understand relative gratification processes, we examine whether the v-curve is affected by socio-structural changes relating to societal inequality, the forms of relative gratification that are related to opposition to immigration, and mediators underlying the relationship between relative gratification and opposition to immigration. In line with Guimond and Dambrun [9], we define relative deprivation as a group's perception that it is deprived in comparison to a particular standard or compared to a relevant outgroup. In contrast, relative gratification is the experience that the group is better off compared to a particular standard or a relevant other group. We begin by reviewing past research on the way economic performance affects prejudice and opposition to immigration. --- Hard times = harsh attitudes? The idea that economic hardship fuels negative attitudes towards minorities forms the basis for theorizing in social psychology, political science and sociology. Both realistic conflict and relative deprivation theorising [19] leads to the predication that, when confronted with economic hardship, people 'lash out' at minorities-in particular those minorities who appear to be competing for resources or are perceived to be the cause of relative deprivation . Consistent with this, political economy researchers have argued that attitudes towards immigrants are conditioned by the extent to which immigrants are perceived to threaten material self-interest [21,22]. At a societal level, there is considerable evidence for this relationship, both in Europe [23] and in North America [24,25]. Economic downturns have also been associated with an increased propensity to vote for parties voicing anti-immigration sentiments and an increase in anti-immigrant attitudes more generally [4,[26][27][28][29][30][31][32][33][34]. There is now also a considerable body of literature showing that poor uneducated manual workers are one category of voters to be drawn to extreme right wing parties with anti-immigrant agendas [27,31,35] and to hold more negative attitudes towards immigrants than their wealthier counterparts [6]. In sum, it appears that it is during harsh economic times, and among income groups experiencing harsh conditions that we encounter harsh attitudes towards minorities. However, even though such findings appear to explain some instances of prejudice, it is also clear that there are many cases where realistic conflict, self-interest and relative deprivation theorising fail to explain opposition to immigration. For one, some claims are poorly supported on closer inspection [36][37][38]. Indeed, on the basis of their review of the literature in the field, Hainmueller and Hopkins [37] conclude that "the significant majority of prior work finds that labor market competition does not shape attitudes [towards immigrants] of the mass public" . This state of affairs led to Ceobanu and Escandell [39] to recommend that it is time to move beyond theorising that explains anti-immigrant attitudes solely by pointing to individual and group-level competition and threat. It is also worth considering classic work by Hovland and Sears [40] purportedly showing that in the period between 1882 and 1930, low cotton prices were associated with a higher number of lynchings of Blacks in the Southern states in the US. A re-analysis of this phenomenon, taking account of a larger time period, failed to replicate this relationship between economic performance and hostility towards minorities [41]. There is also work showing no relationship between economic downturn or wealth of groups in society and support for political parties with an anti-immigrant agenda. For example, Lubbers and Scheepers' [31] analysis of opinion poll data collected in different German regions between 1989 and 1998 shows that there is no straightforward correlation between a region's economic performance and support for the anti-immigrant party, the Republikaner. Consistent with this, Hannah Arendt, in her classic book 'The Origins of Totalitarianism' [42] points out that totalitarian movements developed in Germany and Italy, but not in the United Kingdom, despite the fact that both countries suffered to the same extent from the consequences of the Great Depression . More interesting for our present purposes, research examining the electoral support for Hitler's National Socialists in different regions of the Weimar Republic found that support for the party was indeed strongest in depressed working-class regions but also in relatively affluent protestant rural areas such as Schleswig-Holstein, Mecklenburg, Pomerania and East Prussia [44]. Likewise, a study examining exceptionally high levels of support for the NSDAP in the city of Braunschweig in the early 1930s , revealed that support for the NSDAP was highest in the seven upper middle-class precincts of the city, with levels of support ranging from 61% to 65.5% [45]. Similarly, Vanneman and Pettigrew [8] found that in Chicago, resistance to black mayor candidates was highest among those who were fraternally deprived and among more affluent workers. The notion that sometimes it is groups at the top of the hierarchy that show most negative intergroup attitudes is also consistent with findings of a meta-analysis by Bettencourt, Dorr, Charlton, and Hume [46]. Their research, conducted in both real-world and laboratory settings, showed that high-status groups generally display more ingroup bias than do low-status groups. It has also been found that higher status groups are more likely than lower status groups to endorse group-based hierarchies and status-inequalities [47,48], and that those with greater wealth donate less generously, score lower in empathic accuracy, and are less responsive to the needs of others than their poorer counterparts [49][50][51][52]. Similar effects have been witnessed at the societal level . In a study in which the state of the national economy was experimentally manipulated, stronger anti-immigrant sentiments were found when the national economy was presented as prospering rather than contracting. These findings are also consistent with research testing Labor Market Competition Theory predictions [54]. Researchers working in this research tradition have suggested that opposition to immigration can be expected to be higher among the more skilled and the wealthier segments in society. This is because these groups are most concerned about increased immigration being a burden on the public purse, and factor increasing pressure to raise taxes. Other work suggests that these effects are most pronounced when high status group members perceive their high status to be unstable or insecure. High status group members whose status is threatened in this way might engage in status protection, or even oppression to avoid losing their higher status. For example, Haslam [55] argues that under such conditions, highstatus group behavior is "likely to take a more sinister form and be reflected in ideologies that attempt to justify and rationalize the ingroup's superiority and the outgroup's inferiority" . Consistent with this, Scheepers, Ellemers, and Sintemaartensdijk [56] showed experimentally that only high status groups who feared to lose their status in the future displayed a stress response in the form of higher systolic blood pressure and pulse pressure. Similarly, Rios Morrison, Fast and Ybarra [57] found that members of high status groups who were more highly identified with their group were only more likely to endorse inequality between groups when they experienced high intergroup threat. More generally, sociologists have argued that high-status groups perceive threats to their status position when they feel that they might lose the concrete privileges to lower-status groups . Indeed, it has been argued that it is the fear that one compares negatively to other relevant groups, concerns that others are climbing the ladder faster than oneself, or anxiety related to the belief that one's earlier economic gains might be lost that feeds discontent and hostility towards minorities . According to Blumer [58], prejudice arises when a privileged group develops fear that a minority group is 'getting out of place' and 'encroaches' on benefits that are typically seen as benefits of their group [59,62]. In line with this notion, a study by Martinovic and Verkuyten [63] found that endorsement of the belief that primo occupants are more entitled than those arriving later was related to prejudice towards immigrants only for Dutch natives who feared that they were losing their entitlements to newcomers. In sum, realistic conflict theorizing may be popular and intuitively appealing, the empirical evidence for the theory is rather mixed and inconclusive: negative attitudes towards minorities have both been observed in times of economic hardship and in times of economic prosperity; among members of low status groups as well as among those who belong to high status groups. While realistic conflict literature accounts well for the finding that economic hardship enhances harsh attitudes, theoretically, it cannot account for the finding that at times, we appear to be particularly harsh towards minorities such as immigrants when our own economic, financial and social status is relatively high, or when a society's economy is booming. --- How economic prosperity hardens attitudes Researchers examining the so-called v-curve developed the notion that both relative deprivation and relative gratification may underlie prejudice and negative attitudes towards minorities. Specifically, negative attitudes towards minorities such as immigrants are predicted to be highest among those who experience relative deprivation and among those who experience relative gratification (either because their status is high or because they experience economic prosperity more generally, [9], see also [64]. There is now a large body of work that has provided empirical evidence in line with the vcurve, both in real world contexts [65] as well as in more controlled experimental studies [9,66,11]. Previous research has offered a number of explanations for the finding that relative gratification is associated with opposition to immigration and more intergroup discrimination. Dambrun and colleagues [65] reasoned that relative gratification enhances the attractiveness of the ingroup, and this, so they argued, is associated with stronger intergroup dynamics, which translate into more intergroup discrimination. Consistent with this, they found that group identification partially mediated the relationship between relative gratification and negative intergroup attitudes in South-Africa. Likewise, Postmes and Smith [11] found evidence for the mediating role of group identification. Importantly however, in these studies it cannot be argued that it is identification with the income group that is mediating the effect. Dambrun and colleagues [65] measured identification in relation to participants' ethnicity while relative gratification was assessed in relation to personal economic conditions. Postmes and Smith [11] manipulated relative gratification as favourable job prospects for university students while identification referred to identification with Britain. In a further exploration of processes underlying the v-curve, Guimond and Dambrun manipulated relative gratification by providing feedback that job prospects of the ingroup were more favourable in the future than that of another comparison group. They found the predicted v-cure but failed to find evidence that social dominance or mood mediated the effect of relative gratification on prejudice and support for anti-immigrant policies. More recently, Moscatelli and colleagues [66] identified two mediators underlying the relationship between relative gratification and intergroup discrimination. First, relative gratification was found to lead to a fear of losing ingroup advantage, and this in turn was associated with greater intergroup discrimination. Second, relatively gratified groups were found to feel guilty about the advantage they had over relatively deprived groups. The authors reasoned that such feelings justified a "strike first" attitude, whereby intergroup discrimination is shown because the outgroup in question is expected to be biased against the relatively gratified ingroup responsible for their deprived state. While these lines of work provide some important insights into the mechanisms underlying the effect of relative gratification on opposition to immigration, a few considerations lead us to suspect that the mediation question has not yet been resolved. First, not all studies that examine the underlying processes are concerned with opposition towards immigration . For instance, Moscatelli et al. [66] manipulated relative gratification of two groups in relation to one another. In their study, participants allocated resources to their own privileged group and to the disadvantaged group. Even though this setting enabled the researchers to gain a deeper understanding of competition between two rival groups, it is not a proper test of the effects of gratification on hostility towards immigrants. After all, immigrants are not necessarily in direct competition with privileged members of society. Thus, this work cannot help us to understand how relative gratification shapes attitudes towards a third party . In order to make progress in this line of research, it is important to note an important difference. Whereas Moscatelli and colleagues found that guilt was an important mediator of the relationship between relative gratification and intergroup discrimination, these effects are probably restricted to contexts where relative gratification was achieved at the expense of the outgroup [66]. In our research, we therefore measure and manipulate relative gratification independently from the group to be evaluated . Second, rather than examining consequences of present gratification, Guimond and Dambrun [9] manipulated relative gratification as the likelihood of future gratification and are thus concerned with the behaviour of prospective group members [11]. In our studies, we examine the effects of relative gratification as a current state, rather than as one that can be expected in the future. We also build on this work by examining whether different forms of relative gratification relate differently to opposition to immigration. In an attempt to gain a better understanding of the processes underlying relative gratification effects, we examine the mediating role of a number of different variables. First, similar to Dambrun et al. [65] we explore the extent to which identification with the wealth group mediates the effects of relative gratification on opposition to immigration. Second, we examine whether relative gratification effects are mediated by collective self-definitions. Specifically, and in line with stereotype content model [67,68] we propose that the stereotype of the wealthybecause of this group's ability to acquire wealth-tends to revolve around attributes associated with competence. However, wealthy groups are also seen as relatively cold . These stereotypes are not just ways in which others perceive the affluent, they are also internalized by the wealthy themselves as ways to understand the group's identity, affecting the content of group norms [11]. We expect that the relatively prosperous groups will self-stereotype as competent but cold, the more group norms will emerge that condone and justify hostility and opposition towards those aspiring to become part of society . Finally, we examine the extent to which relative gratification effects are due to so-called 'fear of falling' [69]. In line with Moscatelli and colleagues [66] we predicted that with increasing levels of relative wealth, anxiety related to the belief that one's earlier economic gains might be lost may feed discontent and hostility towards minorities [61]. This reasoning is consistent with social identity theorizing that anxiety about the security of high status will enhance status protection behaviour and enhance the motivation to justify the current status quo [55,56,70]. --- Overview of Research We conducted four studies to examine support for our hypotheses. First, in all studies, we examined support for the v-cure hypothesis. We predicted that compared to moderate levels of wealth, lower and higher wealth will be associated with more opposition to immigrants and towards immigration policies. In a first study, involving a Swiss referendum, we examined whether percentages of votes in favour of curbing immigration would be higher in cantons that are relatively poor and cantons that are relatively prosperous . In Study 2, in an attempt to provide evidence for causality, we tested the same vcurve prediction using an experimental paradigm where we systematically manipulated the position of the group in the hierarchy . In a third study, we examined whether features of the socio-structural context affect the prevalence of the v-curve. Specifically, we examined whether the extent to which relative gratification enhances opposition to immigration is affected by growing inequality in society. Inequality has been associated with the perception that society is hostile, and, building on previous work that has shown that relative gratification is only associated with more hostility towards immigrants when group norms condone such harshness [11], we predicted that the vcurve will be more pronounced in conditions where societies face growing inequality between wealth groups . In a final study, we examined the v-curve in a representative sample of Australians. This study had several additional aims. First, we examined different forms of relative gratification and examine for instance whether effects of past, current and future gratification on opposition to immigration differ and whether perceived personal relative gratification effects are similar to those of collective relative gratification. No strong predictions were formulated, but we were open to the finding that because relative gratification is felt more strongly when it is a current than an anticipated state, the former relative gratification form would be stronger predictor of opposition to immigration than the latter. Furthermore, in line with classic reasoning that collective level responses are best predicted by fraternalistic deprivation [71,8], we expected that collective relative gratification would be a better predictor of opposition to immigration than personal gratification . In addition, in this study we systematically examined support for the proposed mediators of the relationship between collective relative gratification and opposition to immigration. We focused on wealth group identification, collective self-definition as competent but cold, and fear for the future wealth of Australia as mediators to the relationship between relative gratification and opposition to immigration. --- Study 1 Ethics Statement Because no data were collected among human participants, and because data were obtained from publicly available data bases, it was not necessary to obtain ethical clearance for this study. --- Method On February 9, 2014, a referendum was held in Switzerland asking the Swiss to vote on the question whether they agreed or disagreed that immigration into the country should be curbed. A small majority of Swiss citizens agreed with the proposal requiring the government to set an upper immigration limit. However, despite the acceptance of the proposal nationally, levels of endorsement within the 26 cantons varied considerably with 'only' 38.90% "yes" votes in Vaud and 68.20% "yes" votes in Ticino . A data set was created bringing together information on anti-immigrant voting by canton and indicators of the canton's economic performance . --- Results Descriptive statistics, bivariate correlations and partial correlations controlling for total number of votes in a canton and seats in the parliament can be found in Table 1. Inspection of the partial correlations shows relative gratification relationships: a higher percentage of "yes" votes was negatively correlated with unemployment, r = -.50, p = .012, and positively correlated with relative disposable income, r = .57, p = .004. The relationship between percentage of "yes" votes and the quadratic term for unemployment was also significant, r = -.47, p = .019, but non-significant for relative disposable income, r = -049, p = .820. Inspection of the curvilinear effect for unemployment showed that the percentage of "yes" votes was higher in cantons with relatively low and relatively high levels of unemployment . --- Discussion In sum, it was in cantons with relatively lower levels of unemployment and relatively higher disposable income that the percentage of "yes" votes was higher. For the relationship between percentage of "yes" votes and unemployment, we also found evidence of a v-curve whereby this negative relationship was relatively speaking more pronounced in the poorer and the wealthier cantons compared to the moderately wealthy cantons. However, when exploring canton effects, the sample size is by definition small and this affects the certainty with which we can draw conclusions. Furthermore, although this case study provides preliminary evidence for a v-curve, this evidence needs to be interpreted with caution because a host of other variables could potentially have affected "yes" votes other than those relating to unemployment and relative disposable income. Therefore, in Study 2 we used an experimental design allowing us to assess whether relative deprivation and relative gratification causes differences in opposition to immigration. --- Study 2 We developed an experimental paradigm that would allow us to isolate effects and examine causality. We again tested support for a v-curve between wealth and opposition to immigration in an experiment in which participants joined a stratified virtual society, and were subsequently asked for their attitudes towards 'newcomers' who would be joining their society. --- Participants and Design Participants were 61 undergraduate students at a large Australian university . Australian citizenship was a prerequisite for participation in the study. Participants were randomly assigned to one of three wealth conditions: as poor , moderate wealth, or above average wealth. --- Ethics Statement This study obtained ethical clearance from the Behavioural and Social Sciences Ethical Review Committee at the University of Queensland. Before completing the questionnaire, participants were informed about the aims of the study. After this, they were informed that continuing with the survey indicated informed consent. --- Procedure and measures Participants were informed that they would become part of a virtual society, Bimboola, and that they were starting a new life as a member of this society . Participants read that, just like any other society, there are differences in income within this society. They were informed that Bimboola consisted of 5 income groups, with group 1 earning less than 5,000 Bimbolian Dollars per year , group 2 earning between 5,000 and 10,000 BD, group 3 earning between 10,000 and 100, 000 BD per year, group 4 earning between 100,000 and 1 million BD per year and group 5 earning more than 1 million BD per year. Participants were allocated to income group 2, 3, or 4 and they were told that allocation was random. Income groups 1 and 5 were primarily included to provide a broader frame and to avoid that income group 2 and 4 would also be the poorest or the wealthiest within the society. In this study, participants were allocated to income group 1and 5 but these results will not be further discussed because in these conditions income level is confounded with an extreme income position . Once participants knew which income group they belonged to, they were invited to start their new life. Their first task was to purchase basic requirements such as a house, a car, and a phone. When choosing a house, participants were shown pictures of houses ranging from old and run-down dwellings to brand-new luxurious mansions. The houses on offer were displayed on screen, and listed by income group . Participants were advised that they could only buy a house within or below their income bracket. The same procedure was used when buying a car and phone. Participants were then told they had the opportunity to select a holiday destination. Whereas the holiday options for income group 2 were quite basic , income group 3 could go on a 4-wheel drive adventure, a week-long campervan trip, or a one week stay in a lakeside cottage. Participants in income group 4 could choose from any of three more extravagant options such as a one-week shopping trip to Paris. Participants completed a manipulation check asking them to indicate to what extent they agreed with the statements "my group is poor" and "my group is rich" on a 7 point scale ranging from 1 = Strongly disagree to 7 = Strongly agree. Participants were then informed that a new group was about to join Bimboola. They were told that these newcomers wanted to rebuild their lives in Bimboola and that they may need some assistance from existing members of the community of Bimboola. After this, opposition to immigration by these newcomers was measured. After completing all measures, participants were debriefed and thanked for their participation. Opposition to immigration. To measure opposition to immigration by the newcomers, 21 items measuring both symbolic and realistic threat perceptions and well as general items were included. Responses were measured on a 7 point scale from 1 = Strongly disagree to 7 = Strongly agree . --- Results Manipulation check. The two manipulation check items were combined after reverse scoring the statement "I am poor" so that higher scores on this measure indicate the perception that the own income group is wealthy . One-way analysis of variance on the wealth manipulation check revealed that the experimental conditions differed significantly from one another, F = 70.60, p<. 001. In line with the manipulations, above average wealth income group participants agreed most with the statement that their group was rich , followed by moderately wealthy income group participants , followed by poor income group participants . Post-hoc tests confirmed that all conditions differed significantly from each other at p < .001. Opposition to immigration. A one-way analysis of variance was set up testing the linear as well as the curvilinear condition effect on opposition to immigration. The linear effect was not significant, F = .07, p = .79, but the quadratic term was significant, F = 4.18, p = .045. Opposition to immigration was highest for the poor income group and the above average wealth income group and lowest for the moderately wealth group . --- Discussion In sum, results of Study 2 showed that the developed experimental design is effective in manipulating wealth perceptions and that the own group's wealth perceptions cause opposition to immigration differences. Specifically, we found a v-curve relationship whereby it is those in poor and above average wealth income groups who were most opposed to immigration compared to those in a moderately wealth group. --- Study 3 Taking Study 2 findings further, in Study 3 we examined the way the broader structural context affects the v-curve, using a similar 'new life' experiment as used in Study 2. In particular, and building on research by Postmes and Smith [11] who showed that the v-curve was only found when the broader social context endorsed a hostile normative climate, we predicted that inequality would amplify the v-curve. In line with findings that societal inequality is associated with low trust, low cohesion, a harsher and more hostile society [74] and a perceived loss of control [75]-all factors that impact on the perceived stability of wealth-we predict that in particular those assigned to poor and above average wealth groups would be affected by the instability that inequality brings . Importantly, we manipulated perceived inequality in such a way that the level of wealth of the participant's own groups remained unchanged, thus making it possible to disentangle effects of income and inequality. --- Participants and Design Participants were 151 MTurk workers . The design was a 3 x 2 . Wealth was manipulated at three levels: as lower than average wealth , moderate wealth , and above average wealth . Participants were randomly assigned to conditions. --- Ethics Statement This study also obtained ethical clearance from the Behavioural and Social Sciences Ethical Review Committee at the University of Queensland. Before completing the questionnaire, participants were informed about the aims of the study. After this, they were informed that continuing with the survey indicated informed consent. --- Procedure and measures As in Study 2, participants were told that they would become part of a hypothetical society, Bimboola. After they had been assigned randomly to one of three wealth groups in a fiveincome group society, they were again asked to buy items that would help them to get started in their new life. Participants were subsequently exposed to the growing or declining inequality manipulation. Specifically, in the declining inequality condition, participants were told "Imagine that over the next 20 years, Bimboola is affected by a change in economy. As a result, the wealth gap in Bimboolean society has decreased . Status differences have decreased : the poor have become richer , the moderately wealthy earn about the same, and the rich have lost some of the wealth and become poorer ." After this, participants were also presented with a graphical representation of this change. Importantly, in the graph, the wealth of the own income group was presented as unaffected over time and this was emphasized in a clarifying note to the graph. Participants then completed the same two-item wealth manipulation check as used in Study 2 . In addition, the growing versus declining inequality manipulation was checked with four items. On 7 point scales ranging from 1 = Strongly disagree to 7 = Strongly agree, participants were asked "Over time, the gap between the poor and the rich has become wider in Bimboola", "Over time, the gap between the poor and the rich has become narrower in Bimboola", "Over time, inequality between the income groups has increased in Bimboola", and "Over time, inequality between the income groups has decreased in Bimboola". After recording of the second and fourth item, responses on the four items were averaged with higher scores indicating growing inequality perceptions . To ensure that our inequality manipulation had not also affected participants' own wealth ratings, we added a question asking them to what extent they agreed that "the income of my group has stayed the same over time". After this, participants were again informed that a new group would be joining Bimboola. After completing all measures, participants were debriefed, received a code they used for payment and they were thanked for their participation. Opposition to immigration. To measure opposition to immigration by the newcomers, most of the items used in Study 2 were included and a few general opposition to immigration items were added, totalling 20 items. Responses were again measured on a 7 point scale from 1 = Strongly disagree to 7 = Strongly agree . --- Results Manipulation check. A 3 x 2 ANOVA on the wealth check revealed a main effect for wealth, F = 72.81, p<. 001, η 2 = .49. In line with the manipulations, above average wealth income group participants agreed most with the statement that their group was rich , followed by moderately wealthy income group participants , followed by poor income group participants . Post-hoc tests confirmed that all conditions differed significantly from each other at p < .001. Albeit much weaker than the wealth condition effect, we also found a main effect for inequality, F = 4.04, p = .046, η 2 = .03, whereby participants in the growing inequality condition felt wealthier than those is the declining inequality condition . It is possible this effect is the result of the graph that participants were asked to study whereby there is a greater gap between the own group and the poorest income group in the growing than in the declining inequality condition. However, this unexpected effect is small, and may well be a fluke effect. The interaction between wealth and inequality was not significant, F = 2.14, p = .122, η 2 = .03. A similar ANOVA on the inequality check revealed only a main effect for inequality, F = 82.45, p< .001, η 2 = . 35. In line with the manipulation, participants perceived income inequality to be higher in the growing inequality condition , than in the declining inequality condition . Finally, we found no significant effects of wealth or equality on the judgement whether wealth had stayed the same over time, all Fs < 2.16, suggesting that the equality information had not affected perceptions of own wealth differently across the conditions. In line with the manipulations, participants perceived that their wealth had stayed the same despite changes in the equality of society. In general, we found that the manipulation of wealth and inequality worked as intended. Opposition to immigration. Analysis of variance revealed that the linear effect for wealth was not significant, F = .12, p = .78, η 2 = .03, but the quadratic term was significant, F = 5.34, p = .022, η 2 = .04, providing evidence for a v-curve: Opposition to immigration was highest in the poor income group condition , and the above average wealth group condition . Opposition to immigration was lowest in the moderate wealth income group . We also found a significant effect of inequality, F = 5.61, p = .019, η 2 = .04. Opposition to immigration was higher in the growing inequality condition than in the declining inequality condition . The interaction between income level and inequality was not significant, F = .10, p = .91 . --- Discussion In sum, Study 3 replicated Study 2 findings and showed that participants in the poor and above average wealth income groups were more opposed to immigration compared to those in a moderately wealthy group. We also found that societal equality affected opposition to immigration, but not in the way that we had expected. Whereas we had proposed that inequality would amplify the v-curve because in particular participants in poor and above average wealth conditions would perceive inequality as more threatening and anxiety provoking than those in the moderately wealth condition, we found instead that participants in all three wealth groups showed more opposition to immigration when their society was expected to become more unequal compared to equal. Even though this finding was not predicted, it is in line with observations by Wilkinson and Pickett [74] that inequality affects everyone in society-the poor as well as the wealthy. We will discuss this finding in greater detail in our General Discussion. Finally, we also found that growing inequality led to a perception of being wealthier and overall greater opposition to immigration than in the declining inequality condition. Even though the higher wealth perceptions in unequal societies may be a result of the way inequality was manipulated, the latter finding on opposition to immigration is consistent with a growing body of work revealing the negative effects of societal inequality on tolerance and well-being more generally . --- Study 4 Even though Studies 1 to 3 provide good evidence for a v-curve using both correlational and experimental designs, it is not clear yet what forms of relative gratification are most conducive to promote opposition to immigration and why this relationship emerges. These questions were the focus of the fourth and final study where we examined the relationship between perceptions of relative gratification and opposition to immigration in an Australian community sample. To address the first question, we examined a range of different indicators of relative gratification including personal level , and collective level relative gratification . In an attempt to assess whether effects of current relative gratification are similar to effects of past and expected future relative gratification, for both personal and collective level relative gratification, we also examined relative gratification relating to current wealth, current wealth compared to the past, and expected future relative gratification. At the personal level, we also included a measure of class as a relatively fixed indicator or relative gratification. Finally, at both the collective and personal level, we examine the extent to which participants felt relatively gratified when considering the threat posed by the Global Financial Crisis . More specifically, we focused on whether participants minimized the impact of the GFC both personally and at the country level, either because they or their country had recovered quickly from the GFC or had not been affected as much by it compared to other individuals or countries. Another important aim of Study 4 was to examine processes underlying relative gratification. We examined the extent to which identification, collective self-definition as competent but cold, and fear for future wealth can account for the relationship between past, present and future collective relative gratification and opposition to immigration. --- Method Participants. Participants were 621 Australian residents with an average age of 40.90 . In terms of ethnicity, 76.9% reported to be White Caucasian, 15.8% were Asian, 1.4% were Middle Eastern, .8% were Aboriginal and/or Torres Strait Islander, and 3.9% ticked the 'other' category. To the question "what is your highest level of education", 89 indicated they did not complete year 12, 101 completed year 12, 146 had a post-secondary qualification, 155 had finished an undergraduate degree and 117 participants held postgraduate qualifications . Analyses revealed that controlling for ethnicity or education did not change the interpretation of the results reported below. Ethics Statement. This study also obtained ethical clearance from the Behavioural and Social Sciences Ethical Review Committee at the University of Queensland. Before completing the questionnaire, participants were informed about the aims of the study. After this, they were informed that continuing with the survey indicated informed consent. Procedure and Measures. The study was part of a larger survey examining participants' perceptions of their life in Australia and their views of Australian society more generally. Here, we focus on a subset of measures. A professional research company collected the data. Participants were sampled from the company's research panel, which covered all Australian states and territories. Collective relative gratification. Collective relative gratification was measured with a number of items that tapped different forms of relative gratification relating to collective wealth. Specifically, we measured current, past and future gratification at the collective level using the items: "Please think of the economic situation in Australia at the moment. How would you describe the current economic situation in Australia?", "Now please think about Australia's economic situation 3 years ago. To what extent would you describe Australia's economic situation three years ago to be worse, the same, or better than it is now? "and "Now think about Australia's economic situation in the next 3 years. To what extent do you expect Australia's economic situation to be worse, the same, or better in the next 3 years?" Responses to the first item were recorded on a scale ranging from 1 = Very bad, to 7 = Very good. The scale for the second and third item assessing past collective gratification and future collective gratification ranged from 1 = A lot worse, to 7 = A lot better. Two items were included asking about the extent Australia was unaffected and had recovered well from the Global Financial Crisis: "Our country has already recovered quite well from the negative impact of the global financial crisis"and "The effect of the global financial crisis on our country's economy has been negligible" . Personal relative gratification. Personal relative gratification was measured by asking participants to indicate their level of satisfaction with their current personal financial situation with the item "Please think of your personal economic situation at the moment. How would you describe your current economic personal economic situation?" . We also asked participants about past and future personal gratification with the items "Now please think at your personal economic situation 3 years ago. To what extent would you describe your personal economic situation 3 years ago to be worse, the same, or better than it is now?" and "Now think about your personal economic situation in the next 3 years. To what extent do you expect your personal economic situation to worsen, remain the same, or improve in the next 3 years?" . Finally, two questions asking about relative gratification as being personally unaffected or having recovered well from the Global Financial Crisis: "The effect of the global financial crisis on my personal economic situation has been negligible "and "I have personally already recovered quite well from the negative impact of the global financial crisis" . We also added one item asking about participants' social class: "How do you perceive your own family's class/social background?" . As a measure of personal wealth, we also asked participants to record their annual household income. However, 40.5% of respondents indicated that they did not know or did not want to provide this information. Given the high number of missing values, we decided not to analyze responses to this measure. Group identification. Four items, adapted from Doosje, Ellemers, and Spears [77] were used to measure identification with Australia . Higher values indicated more identification with the own income group . Collective self-definition as competent and cold. Participants were asked to indicate on a 7-point scale from 1 = Strongly disagree to 7 = Strongly agree the extent to which they perceived Australians to be warm , and competent ["competent", "capable", "weak" ]. The reliability of the warmth , and competence scales was satisfactory. Fear about future wealth. One item, adapted from Jetten and Wohl [78] was included to gauge fear about future wealth at the collective level: "I feel anxious about the future wealth of Australia". We also included the other items of the collective angst scale that this item is taken from. Similar effects were found for this scale as for the one item measure assessing fear about the future wealth of the country. Opposition to immigration. Opposition to immigration was measured using 6 items assessing both realistic threat perceptions and symbolic threat perceptions . Responses were recorded on 7 point scales ranging from 1 = Strongly disagree to 7 = Strongly agree. Participants did not differentiate between realistic and symbolic threat and we combined the items to form one scale, α = .95. --- Results A series of curve estimation analyses were conducted exploring the extent to which each of the collective and personal level relative gratification measures predicted opposition to immigration . Given the fact that we conducted multiple tests, a conservative significance level of p < .01 was used. Inspection of the linear and quadratic effects showed a number of significant relationships. First, focusing on the collective relative gratification predictors of opposition to immigration, only higher current collective relative gratification was linearly associated with lower opposition to immigration, β = -.17, p < .001. There was evidence for curvilinear relationships, such that opposition to immigration was higher among those with lower and among those with higher wealth levels . Interestingly too, the quadratic term of minimizing the impact of the GFC on Australia also predicted opposition to immigration: it was those who felt Australia had not recovered from the GFC and those who felt Australia was hardly affected by the GFC , minimizing the effects of the GFC that was associated with higher opposition to immigration, β = .49, p = .007. There were also a number of significant relationships between different indicators of personal relative gratification and opposition to immigration. Specifically, higher current and future personal relative gratification was linearly associated with lower opposition to immigration, β = -.10, p = .01 and β = -.18, p < .001, respectively. In both cases, the quadratic effect was also significant. Similar to the collective level analyses, and in line with the v-curve hypothesis, higher opposition to immigration was found for those that indicated lower and higher current and future wealth, β = .65, p < .001 and β = .51, p = .002, respectively. Minimizing the personal effects of the GFC or class did not predict opposition to immigration, neither linearly nor curvilinearly . Mediators. A further series of curve estimation analyses were conducted examining the power of identification, collective self-definition as competent and cold, and fear for the future wealth of Australia to explain the relationship between collective relative gratification and opposition to immigration. These analyses were conducted using only present collective relative gratification measures. Results of these analyses are reported in Table 2. Inspection of the linear relationship between collective relative gratification and each of the mediators revealed a significant linear relationship for identification, β = .11, p = .006, collective self-definition as competent, β = .13, p = .001, and fear for future wealth, β = -.23, p = .001. Collective relative gratification was associated with higher identification, a higher collective selfdefinition as competent and less fear for future wealth. Inspection of the curvilinear relationships revealed only a significant effect for fear for future wealth, β = .76, p < .001. In other words, it was those who were highest and those who were lowest in collective gratification that feared most for the future wealth of Australia. Further analyses revealed that only a few potential mediators predicted opposition to immigration. Higher identification was linearly related to greater opposition to immigration, β = .16, p < .001, and higher fear for the future wealth of Australia was linearly associated with higher opposition to immigration, β = .26, p < .001. Because only national identification and fear for future wealth met the conditions for mediations, it was only these variables that were included as mediators in bootstrap analyses [79] using 5000 resamples. This revealed that group identification was not a significant mediator . However, fear for future wealth turned out to be a significant mediator . Importantly too for our purposes, the quadratic future wealth term was also a --- Discussion In sum, even though we found an overall negative relationship between relative collective gratification and opposition to immigration in Australia, in addition to these linear effects we found the predicted v-curve relationship between relative gratification and opposition to immigration. More specifically, we found evidence of a v-curve whereby the relationship between perceived relative gratification and opposition to immigration was most pronounced for those who felt collectively relatively deprived or relatively gratified at present or expected to feel so in the future. A similar v-curve was found for the relationship between personal relative gratification and opposition to immigration: it was those who felt relatively poor or wealthy at present or in the future who were more opposed to immigration than those who felt moderately wealthy. Interestingly too, we found that relative gratification by minimizing the negative effects of the GFC on Australia also produced the predicted v-curve relationship. Oppose to immigration to Australia was relatively high among those who denied and those who endorsed the statement that these negative effects were no longer of concern. Study 4 also provided evidence for the underlying processes. At the collective level, it was only fear for the future wealth of Australia that mediated the relationship between collective gratification and opposition to immigration. Interestingly, the relationship between collective gratification and fear for the future wealth of Australia was curvilinear: it was those lower and higher in relative gratification who feared future wealth decline of Australia. This finding is consistent with findings by Moscatelli and colleagues [66] and suggests that relative gratification, because it is associated with relative fear for the future wealth of Australia, predicts opposition to immigration. --- General Discussion We set ourselves the task of exploring whether there is evidence for a v-curve shaped relationship between the economic performance of a group or society and opposition to immigration. That is, we predicted that opposition to immigration would be highest among groups that are relatively deprived and poor as well as among those with above average wealth . We found evidence for such a relationship in four studies that examined support for the v-curve in different contexts and using different research designs. In Study 1, we found that voting in favour of a national policy to curb immigration was highest in Swiss cantons where unemployment was lowest and highest . Even though we only found a positive linear relationship and not the predicted curvilinear relationship between canton relative disposable income and voting in favour of this anti-immigration policy, the failure to find evidence for a v-curve was due to the failure to find greater support for this policy among the poorest cantons, not because of the lack of support for higher levels of endorsement of the policy in the wealthier cantons. Even though this finding supports relative gratification theorizing [9,11] it is at odds with classic relative deprivation reasoning [15,17]. Studies 2 and 3 examined the v-curve using an experimental design. Here too, we found that in both studies opposition to 'newcomers' joining a hypothetical society was higher among poor and above average wealth group members than among those in a moderate wealth group condition. Finally, in a cross-sectional study, in line with the v-curve hypothesis, opposition to immigration spiked for those who experienced lower and higher personal or collective levels of current and future wealth. We also found a v-curve between minimizing the effects of the GFC at the collective level and opposition to immigration. In sum, across the four studies, we found consistent support for the v-curve, whereby both wealth deprivation and wealth gratification are associated with greater support for policies aimed at curbing immigration , greater opposition to immigration by newcomers to a hypothetical society , and greater opposition to immigrants seeking to migrate to Australia . --- Implications In order to better understand the processes underlying the v-curve and the conditions under which outgroup hostility will be most pronounced, it is useful to review other findings that emerged from the studies. First, in our examination of the way societal inequality affects the vcurve , we did not find that the v-curve would be more pronounced with growing than with declining inequality. Instead, in addition to a v-curve, we found that all wealth groups became more opposed to immigrants when inequality was growing rather than declining. This finding is interesting and suggests that growing inequality is equally threatening for those at the bottom, middle or top of a wealth hierarchy. The finding is consistent with observations by political scientists and sociologists that growing inequality leads to greater status competition whereby everyone experiences greater status instability and status anxiety [50,74]. Whereas the finding for growing inequality effects is interesting and consistent with Postmes and Smith's findings , equally interesting is that we still found evidence for a v-curve among participants who were informed that inequality in their society was declining. It may be that the way we manipulated declining inequality is responsible for this finding. In our declining inequality condition, income differences between groups became smaller and this may have threatened distinctiveness between the income groups. There is a large body of work showing that threats to distinctiveness leads to a motivation to restore intergroup distinctiveness, often by showing enhanced intergroup discrimination towards other groups [80][81][82]. The extent to which distinctiveness threat associated with smaller differences between income groups enhanced perceived competition for those who felt relatively gratified should be examined in future research. Second, in an attempt to better understand what forms of relative gratification are associated with opposition to immigration, in Study 4 we examined the relationship between collective and personal relative gratification as well as the relationship between current, expected future and past relative gratification with opposition to immigration. Interestingly, past relative gratification did not predict opposition to immigration. However, and in line with the v-curve hypothesis, opposition to immigration was highest for those who experienced high personal and/or collective levels of current and future relative gratification. We can only speculate why current and expected relative gratification predicted opposition to immigration whereas temporal comparisons with past wealth did not. We suspect that a sense of current and expected future gratification is associated with greater wealth stability [70] and that this may feed into a sense of deservingness or entitlement [58,59] thereby justifying the exclusion of immigrants. In contrast, awareness that one was less wealthy in the past than the present may evoke a sense of humbleness, reminding people that they were once less well-off themselves, and this should attenuate hostility and prejudice. However, these are speculative suggestions that should be tested in future research. We also found that relative gratification emerging from a sense that one's country is relatively more robust against the global financial crisis than other countries also predicted opposition to immigration. This should serve as a reminder that it is not so much absolute wealth, but feeling wealthier than others that matters when predicting outcomes of relative gratification [65,9]. It is interesting that we found that both personal and collective level gratification predicted opposition to immigration. Even though this appears to go against classic reasoning that collective level outcomes are predicted by collective level processes relating to fraternalistic instead of egoistic deprivation or gratification perceptions [71,8], it may well be that these judgements are not always all that different. Indeed, because self-definition is in important ways defined by group membership, perceptions of personal achievement are often bound up with the position of ones group in society [70,8]. Finally, Study 4 provided insight in the processes underlying the relationship between relative gratification and opposition to immigration. Even though self-definitions as competent predicted more opposition to immigration, we did not find that relative gratification was associated with greater competence. We also found no support for the idea that self-definitions as competent triggered self-definitions of being cold. Indeed, self-definitions as warm/cold were not related to either relative gratification or opposition to immigration. It may well be the case that we would find more support for a mediating role of self-definitions as competent and cold in context where wealth is attributed to competence . We suspect that under such conditions wealth may trigger feelings of deservingness and entitlement and these may also be the conditions where groups will self-define as cold, both providing some moral justification for holding negative attitudes of immigrants. This should be examined in future research. We found no support for the prediction that group identification mediates the relationship between relative gratification and opposition to immigration . Even though relative gratification predicted identification with Australia linearly, and identification, in turn, predicted higher opposition to immigration, bootstrapping analysis showed that identification did not mediate this relationship. Theoretically, this is not surprising. The prediction that greater commitment to the group enhances opposition to immigration is based on the assumption that group identification is positively correlated with ingroup bias and outgroup prejudice. While this may be true in certain contexts, some have warned against assuming a straightforward relationship between the two [83]. For example, there is evidence that group identification only promotes ingroup bias to the extent that this is normatively prescribed in a group [81]. We did however find that fear for future wealth mediated the relationship between relative gratification and opposition to immigration and accounted for the observed v-curve. What this suggests is that relative gratification is associated with fear that future wealth may be lost and it is this 'fear of falling' notion [69] that helps to explain why those who feel relatively gratified become more negative towards minorities such as immigrants. Opposition to immigration thus results from anxiety and fear that the relative gratified position may be lost in the future, and this undermines the security of the wealth position. Our findings thus echo research showing that for high status groups, status instability and insecurity can lead to a stress response [56] and increased discriminatory behaviour in intergroup settings, . --- Limitations and future research Our studies are not without limitations and they raise new questions to be examined in future research. First, we suggest that it may be instructive to examine the effects of societal inequality in greater depth in future research. Interestingly, our findings suggest that everyone becomes harsher when income inequality grows compared to when income inequality declines. However, because we needed to ensure that the income of the own wealth group did not change with growing or declining inequality , our inequality manipulation may also have been rather weak. This is because the stability of the own group wealth may have taken away some of the uncertainty that everyone experiences when boundaries between groups become more permeable due to growing inequality, and when all wealth groups start to fear for their future wealth. This too should be explored in future work, whereby societal inequality should be manipulated in different ways. It is also essential in future research to examine the processes underlying growing societal inequality effects. Indeed, while the effects of inequality have been well documented [74], less progress has been made in understanding why inequality is associated with such negative outcomes. The paradigm that was presented in Studies 2 and 3 might be useful to examine these underlying processes experimentally. As a final suggestion for future research, we note that we found evidence for the v-curve in contexts dominated by relative gratification relationships in Study 1 but relative deprivation relationships in Study 4 . Put differently, evidence for the v-curve was found against the backdrop of a general relative gratification effect and a general relative deprivation effect . This suggests that in most contexts, multiple processes may be at play at once, simultaneously co-determining the relative strength of prejudice and hostility due to relative deprivation as well as relative gratification related processes . These processes may differ from laboratory to natural context due to a host of extraneous factors, making it very difficult to predict a priori which process will dominate or to make strong claims about the generalizability of findings. This reasoning is consistent with observations that there is no straightforward relationship between economic downturns and opposition to immigration [43,31,13] and that what is required is great sensitivity to the context and the socio-structural context that may help to inform researchers which processes will dominate. But notwithstanding these variations in the tilt of the v-curve pattern across contexts, it should be noted that in none of these studies was the curvilinear effect reversed. In other words, we are satisfied with the robustness of the underlying effect. --- Final thoughts Although our work is not the first to demonstrate the v-curve empirically, by examining this relationship in new ways, we have advanced our understanding of the way economic performance of a group affects opposition to immigration. We contribute to this body of work by providing correlational as well as experimental evidence for the v-curve, by examining different forms of relative gratification and by providing greater clarity on the processes underlying the relationship between relative gratification and opposition to immigration. Although wealth undoubtedly reduces certain kinds of stress, we find that relative gratification and wealth also triggers fears that wealth will be lost in the future. Ironically, then, the perception that one is relatively wealthy is therefore not always more comfortable than the perception that one is relatively deprived. In some sense, relative gratification and relative deprivation are not the polar opposites they might seem to be. Feelings of gratification might be pleasing and comfortable at one level, but like all delights such relative advantages might well bring latent insecurities of loss and fears of decline. --- Data files from Study 1 to Study 4 are available from the Dryad repository database . ---

Previous research has shown that negative attitudes towards immigrants and support for anti-immigrant parties are observed both among those experiencing relative deprivation and those experiencing relative gratification (so called v-curve). Whereas the effect of relative deprivation is intuitive, the effect of relative gratification is more difficult to explain. Why would economic prosperity provoke negative attitudes towards immigrants? We first present correlational (Study 1) and experimental (Study 2) support for the v-curve. In Study 1, in a national Swiss referendum, a higher percentage anti-immigrant voting was found in cantons with relatively lower and relatively higher relative disposable income. In Study 2, in a hypothetical society, more opposition to 'newcomers' joining society was found among poor or above average wealth group members than among those in a moderate wealth group condition. In Study 3, we replicate this finding and also show that opposition to immigration is higher for all wealth groups when societal inequality is growing rather than declining. In a final study, we examine different forms of relative gratification and mediators of the relationship between relative gratification and opposition to immigration (i.e., identification, collective self-definition as competent and cold, and fear about future wealth). Only fear about future wealth mediates this relationship. We conclude that, paradoxically, relative gratification effects are partly due to the fear of future deprivation.

Introduction Over the past decade, there has been increased scrutiny on high rates of psychotropic use and restraint practice in Australian residential aged care from researchers [1][2][3][4], the media [5], policymakers [6,7], human rights [8] and advocacy groups [9]. This attention led to restraint being highlighted as a key area of focus for the Royal Commission into Aged Care Quality and Safety which commenced in 2018 [10]. Further, the Aged Care Act was amended in 2019 to include Australia's first legislation regulating the use of restraint in residential aged care. From 1 July 2019, aged care providers have explicit obligations in relation to restraint use [11]. The use of restraint must be the strategy of last resort after rigorous assessment and other non-restraint approaches have been trialed. When judged as appropriate, the use of restraint must be the least restrictive form only after 2 of 15 informed consent is gained. Moreover, all use must be monitored and reviewed on a regular basis [11]. 'Restraint' is defined by the Australian Medical Association as 'a device or medication that is used for the purpose of restricting the movement and/or behavior of a person' [12]. The use of restraint in people receiving aged care is often justified based on reducing risk or preventing harm to the person or others [3]. Yet, restraint use is associated with detrimental consequences, including cognitive decline, increased falls, pressure injuries, lessened activities of daily living and death [3,4,[13][14][15]. Despite these adverse effects, restraint is used commonly in residential aged care. A recent systematic review and meta-analysis cited the average prevalence over the last two decades of physical restraint as a third of all residents and chemical restraint use as 32% of residents [16]. Ideally, aged care providers, wherever situated, should strive for a restraint-free environment. However, in practice, it is often difficult to balance risk management with the promotion of autonomy for older people needing care. Similarly, it can be challenging to provide a safe environment but at the same time enhance a person's quality of life [17]. There will be situations when difficult decisions regarding restraint need to be made. Recognizing this, in 2012, the Australian Department of Health developed a set of resources, the 'Decision-Making Tools' , to guide providers, nursing and care staff, care-givers and, wherever possible, relatives and residents, to make informed decisions about restraint [18,19]. At the start of 2020, our research group was commissioned to update these resources to align with legislative changes and contemporary best practice. To inform and develop clear, practical and influential resources we conducted qualitative research aimed to explore the attitudes, beliefs and experiences of a diverse group of interdisciplinary stakeholders towards restraint use in aged care. The current interpretations of what constitutes 'restraint' were scoped, along with views on amendments to the Australian Aged Care legislation relating to restraint [9]. --- Materials and Methods --- Design and Sampling Qualitative research can be defined as the study of the nature of phenomena and strives to understand why something is, or is not, observed [20]. To achieve our research aim we conducted interviews with a variety of care providers and relatives and two dedicated focus groups with management staff. The triangulation of interview and focus group qualitative data was intended to achieve an in-depth understanding of restraint practice, incorporating both individual perspectives and views of homogeneous groups with relevant expertise and experience [20]. Participants of the interviews and focus groups were purposely selected to represent the key stakeholders involved when restraint is proposed and used in aged care. The semistructured interviews were held with health practitioners working within, or those with relatives living in aged care settings, including residential, community and day care. The first focus group included senior nurse managers and clinical directors based in residential aged care; the second focus group was comprised of community care managers. --- Recruitment For the semi-structured interviews, we recruited participants working in various roles and aged care settings to capture a wide range of interdisciplinary health practitioner perspectives, including registered nurses , enrolled nurses , personal care assistants , physicians, physiotherapists and pharmacists. We also sought to obtain viewpoints from relatives of people receiving aged care. Most of the interviews were held in Hobart, Tasmania, where the research was based, but participants from other Australian States were also sought. Potential participants were identified by all members of the interdisciplinary research team across their professional networks, e-mailed an information sheet and invited to take part. Those who responded were phoned by J.B. or C.S. who outlined the study and arranged an interview after gaining verbal consent. Written consent was obtained before each interview and focus group. The focus group participants were recruited by J.B. who sent emails to potential candidate aged care home provider groups and community care providers inviting them to be part of the study. The focus groups were initially intended to be held before the interviews; however, due to workload and uncertainty associated with COVID-19 they were conducted after the majority of the semi-structured interviews were completed. Both focus groups were moderated by J.B. and conducted remotely via video internet platforms. Approval for this research was obtained from the Tasmanian Health and Medical Human Research Ethics Committee . As part of the ethics approval, all identifying information was removed from interview data and details of participants were kept confidential. All participants were assured that they could withdraw from the interview or the study at any time. The interviews were conducted between 23 April 2020 and 1 June 2020 and the two focus groups were held in the final week of May 2020. After their interview or focus group, each participant was offered a $100 book voucher to compensate them for the prereading and their interview time. A third of interviewees declined this incentive. --- Qualitative Semi-Structured Interview and Focus Group Process Prior to the semi-structured interviews, all participants were sent a link to the 2012 DMT resource and told they would be asked for their opinions on this resource during their interview [18,19]. Originally, we planned to conduct most of the semi-structured interviews face-to-face; however, due to COVID-19 restrictions, all interviews were conducted by J.B. and C.S. remotely through internet video-meeting platforms or by phone. The interviews were conducted using an interview guide developed by all members of the research team which consisted of three pharmacists, two nurses, a physiotherapist, physician and a speech pathologist. Standard demographic questions were followed by a series of closed and open-ended questions, which were adapted as the study progressed and new areas of enquiry emerged [20]. Participants were free to express their views and experiences and diverge from the interview guide. Likewise, the interviewers were free to ask additional questions or omit questions when not considered relevant. During the interview, participants were asked to recall a case where restraints had been proposed or used, so their real-life experiences were described. These are all the questions I have listed-but would you like to add any other comments about this topic? As with the individual interviews, each focus group participant was given a link to the 2012 DMT resource [18,19] to enable content review and completed an on-line demographic questionnaire. A topic guide based on the semi-structured interview guide was customized for residential aged care or community home care for the first and second focus group, respectively. The Focus Group moderator began each session with an introduction to encourage an open environment for participants to share their opinions and experiences [20]. The moderator then initiated group discussion using open-ended questions from the topic guide and ensured that each participant was given several opportunities to speak during the session. The interviews and focus groups were recorded, stored confidentially and transcribed verbatim by a professional transcription company. Returned transcripts were anonymized and missing/unclear words clarified by C.S. and J.B. by listening to the original recording. Both focus group transcripts and a sample of semi-structured interview transcripts were checked by J.B. and C.S. to verify the accuracy of the transcription. Interview participants were also offered the opportunity to check and amend their transcribed interviews. Nineteen participants were sent transcripts and three made slight amendments. All interview data files were uploaded onto NVivo 12 software for analyses [21]. --- Data Analysis Data were thematically analyzed using a pragmatic approach in which the most appropriate research methods were chosen to investigate the topic as opposed to a single paradigm based on a philosophical doctrine [22]. The 'Framework Method' [23], often used in applied health care research, was used to answer the research question about what constitutes 'restraint'. A bubble plot was used to visually represent the frequency of each theme, along with their conceptualized interconnections, creating a framework of these elements grouping codes into sub-themes [23]. Qualitative analysis was undertaken using Braun and Clarke's six-step process of thematic analysis which involved data familiarization; interim code generation; seeking themes; reviewing themes; defining and naming themes; write-up [24]. To start, three authors familiarized themselves with the semistructured interviews, made notes and discussed findings. Likewise, J.B. and B.W. read through the focus group transcripts several times and met to review findings. Then, data were independently coded by J.B. and B.W. using a hybrid inductive and deductive approach, with emerging themes hierarchically coded utilizing the NVivo 12 platform [21]. The data were organized into themes and sub-themes, mapped, and interpreted. As themes were identified, they were cross-checked and debated. Any differences in interpretation were resolved by discussion and adjusted until consensus was reached. Exemplar quotes supporting each theme were captured by J.B and B.W. --- Results --- Participant Demographics Participant demographics for the semi-structured interviews are listed in Table 2. A total of 28 participants were interviewed over a six-week period from April to June 2020. Most were females and 6 were males, with ages ranging between 28-68 years. Ten interviewees were RNs , 5 were PCAs and 1 EN. The other participants included 3 physicians, 3 physiotherapists, 3 pharmacists and 3 relatives. The majority of health practitioners interviewed worked in residential aged care, 4 worked exclusively within community care and 2 in day-care centers. Three of the participants, all physicians, worked in both residential and community aged care settings. The majority of participants had worked in aged care for ten years or longer. Of the three relatives, one had a sibling living in residential care, one had a parent receiving community care and the final relative's parent attended a day-care and respite center. Sixteen of the participants lived in Tasmania and 12 were based elsewhere in Australia. The semi-structured interviews lasted between 35 to 85 min. Demographic data for each focus group are as follows: Six residential care managers, all RNs, participated in the first group . Four were based in Tasmania and two in Victoria, all were female and aged between 48 and 64 years. For the second focus group , three community managers participated . All were female and aged between 42 to 53 years. Two other community care managers from Tasmania had verbally agreed to participate in the second focus group but withdrew without explanation on the day of the meeting. The residential care focus group ran for 87 min and the community focus group for 65 min. --- Qualitative Themes Following analysis, five themes relating to restraint practice in aged care settings were identified. They were: "understanding of restraint", "support for legislation", "a restraintfree environment is not achievable", "low-level restraint" and "community restraint use is uncharted". These themes are described further below. --- Theme 1: Understanding of Restraint All the participants, except two, were able to define what restraint meant to them. Several participants provided more than one interpretation. One of the physicians refused to give a definition, claiming that to do so was a "meaningless circular pursuit". Over half of those interviewed defined restraint under the sub-theme of 'limiting what people do' stating that restraint involved stopping people doing things, restricted their movement or impeded their freedom. "It is the act of stopping someone from doing something they want to do. Whatever they want to do, be it a decision or action." --- Physio 1 The next most common sub-theme is related to ensuring the safety of the older person or others. Notably, those defining restraint as needed for safety or to reduce harm were predominantly RNs working in residential care. A few participants, all pharmacists, defined restraint as a means to control a person's behavior. Several interviewees defined restraint more broadly as also impeding the ability to make choices. "It's about restricting movement, restricting rights, anything about the person's ability to retain their independence or choice." --- FG Residential Three participants stated that restraint altered the mind of a person or the way they thought. Finally, all the PCAs and one of the relatives defined the term in literal terms as either 'physical' or 'chemical' forms of restraint. The codes and sub-themes relating to the overriding theme of 'understanding of restraint' are presented as a bubble plot below . A summative approach was used to calculate the total number of definitions including a coded element. The size of the bubble is proportional to the number of definitions coded to each sub-theme [23]. of those interviewed defined restraint under the sub-theme of 'limiting what people do' stating that restraint involved stopping people doing things, restricted their movement or impeded their freedom. "It is the act of stopping someone from doing something they want to do. Whatever they want to do, be it a decision or action." --- Physio 1 The next most common sub-theme is related to ensuring the safety of the older person or others. Notably, those defining restraint as needed for safety or to reduce harm were predominantly RNs working in residential care. A few participants, all pharmacists, defined restraint as a means to control a person's behavior. Several interviewees defined restraint more broadly as also impeding the ability to make choices. "It's about restricting movement, restricting rights, anything about the person's ability to retain their independence or choice." --- FG Residential Three participants stated that restraint altered the mind of a person or the way they thought. Finally, all the PCAs and one of the relatives defined the term in literal terms as either 'physical' or 'chemical' forms of restraint. The codes and sub-themes relating to the overriding theme of 'understanding of restraint' are presented as a bubble plot below . A summative approach was used to calculate the total number of definitions including a coded element. The size of the bubble is proportional to the number of definitions coded to each sub-theme [23]. Many participants felt that definitions of restraint varied widely depending on the aged care organization, the aged care setting and between staff working within an organization. "I think a lot of it comes down to sometimes how people define it, and obviously that changes hugely. Even in one facility, you talk to maybe the manager and they say one thing, and then the RN thinks something different, and then someone else thinks something different; so it can be quite confusing." --- Pharm 2 Several participants mentioned inconsistencies in restraint definitions used in different government publications, the aged care sector and the National Disability Insurance Scheme . This lack of clarity and consistency about what restraint means confuses staff and other health practitioners working within aged care. "In the residential context it's even more complicated, because the national quality indicator program defines 'restraint' differently to the Legislation. And in providers where there's mixed circumstances like ours, where potentially you could have staff providing support to community and residential aged care, it's diabolical . . . .. we've also got NDIS consumers in the community where the restrictive practices obligations are different again." --- FG Community --- Theme 2: Support for Legislation Participants were very supportive of the new restraint legislation that had been introduced for residential aged care, claiming that it had heightened awareness, made staff seek alternative strategies, ensured greater accountability and had already impacted use. "I think it's great. It's reined people in. It's made everybody think about what we're doing as opposed to just this person is disruptive on the evening shift, we don't have time to deal with this so let's just give him something to shut him up. Because that's what was happening." --- RN 9 Some participants felt the legislation had directly enhanced interprofessional collaboration, particularly around the use and review of chemical restraint. "It has made us focus on the chemical restraints a lot more, and we've had a lot more conversations, with General Practitioners , around ceasing, than we would have had before." --- RN 8 Although most people interviewed were supportive of the tightened regulation many admitted that they were not entirely sure what it involved. Several nursing staff said the legislative changes had imposed additional workload, such as increased documentation and ensuring informed consent had been gained. Yet, despite increased obligations on providers, most felt the additional reporting and safeguards were worth it to reduce restraint practice. "It's a massive pain, but I think it needed to be done and I'd rather them go ridiculous and way over-report and me have to deal with the paperwork nightmare for the next two years and then slowly reduce it and actually catch out some of the people that were doing the wrong thing." RN 1 --- Theme 3: Restraint-Free Environments Not Possible All participants were asked to read the DMT restraint resources [18,19] which incorporate the title: 'supporting a restraint-free environment'. Yet most queried the feasibility of 'restraint-free' practice; expressing the view that locked doors and gates, both forms of environmental restraint, were crucial to have in aged care, particularly when many clients were highly cognitively impaired. If locks were categorized as restraint use, then using restraint in most settings was unavoidable. "I don't know that restraint-free practice is-I don't know that it's possible. I mean, we're talking about me, here. Restraint-free means my door's open. It's not ideal. It's not safe, it's not possible. Well, it's possible, I can do it, but what do I say? Well, he got run over yesterday, told him he shouldn't have gone out the gate." day care 1 Similarly, many participants expressed the opinion that chemical restraint could never be completely eliminated but instead it was more important to ensure they were used appropriately when prescribed. "I don't believe it's possible to have zero antipsychotics in a facility, or zero psychotropics in a facility, but I definitely think that it should be possible to have only those who have a clear diagnosis, a clear plan, and it's all monitored." --- Pharm 3 There was consensus from most stakeholders that restraint was sometimes needed and that the overarching emphasis should be on minimizing use, not to condone all use. "I think that needs to be clear from the get-go with recognition that sometimes, restraint is necessary to prevent people from harming themselves or coming to harm or harming other people." Relative 2 "The care staff know that it's never going to be a restraint-free environment. To minimize the impact of restraint, you're minimizing them and having as little as possible." EN Some of the managers in the residential aged care focus group commented that some homes were catering to the 'restraint-free spin'. They felt that homes voicing they were 'restraint-free' demonstrated a limited understanding of what restraint meant and the types of practices it entailed. "People want to say, "We don't have restraint here," and that's such a big aspirational target. I think in some instances, there's real ignorance about what restraint is and what it looks like." --- FG Residential --- Theme 4: Low-Level Restraint During the residential care focus group and in many of the semi-structured interviews, participants referred to the use of low-level or less obvious forms of restraint. From the residential focus group: "All of our facilities say they don't use any physical restraint, but we found physical restraint: the pushing the chair under the table, the locked doors to outside areas, so a whole lot of things that aren't seen as hard physical restraint but are definitively restraining." FG Residential Participant 1 "I see things like call bells that are dropped on the floor or not in positions to allow the person to get assistance." --- FG Residential Participant 2 "Even just simple things like leaving a tray-table across a chair that is being used for having a meal or an activity but then not removing it, so the person is free to move around." --- FG Residential Participant 3 Participants also referred to practices such as tucking in bed sheets tightly to restrict a person's movement, taking cushions away from deep armchairs and the use of low beds or princess chairs as forms of restraint. One physician mentioned that not accommodating for hearing and/or sight impairment could also be viewed as restraint. Some felt that in most cases this 'low-level' restraint use was unintentional and spoke to a lack of awareness and the need for more education on this issue. "It's making people aware of what is considered to be a restraint is really important too. So things like the princess chairs that they use. Or even someone who's got some sort of incapacity, so they can't hear everyone, or they can't see everyone, they're not able to access help when they want to access help." --- Physician 1 In contrast, a few participants implied that low-level restraint was used commonly to compensate for a lack of staff or to allow staff to assist other residents. "Everyone's got to try and put other people to bed . . . one resident can sit there for half an hour after dinner's finished all by themselves, with the wheelchair locked, because they don't want them to get away. But no-one's there to take them back to their room and help them out." PCA 3 "I suspect, across the board, there's a lot of, what I would call, low-level restraint to be able to implement the care of anyone in a facility like ours, which is for people with dementia . . . It's one of those things where you end up in this argument . . . ."Well, if we can't do that, we can't actually implement any care." Physio 2 --- Theme 5: Community Restraint Use Is Uncharted Community-based nursing staff and PCAs faced additional barriers when it came to identifying and minimizing restraint. In response to the question, "do you think that the practice of restraining people occurs commonly in community aged care?" the community focus group members replied: Participants working for community service providers said that they had limited control over what happened in a client's own home. If the family installed a bed-rail they were unable to prevent its use when they were not there. In addition, they stressed that many people with home care packages opted not to use their funding for clinical care, including medication management, meaning that the use of chemical restraint could not be ascertained. Adding to the complexity was the use of multiple care providers by the same client, the lack of home visits made by GPs and the need to manage relationships with certain clients who were resistant to having assistance with care: "I think you have to step so carefully with some people in the community. Even if we've got concerns, we've got to be really careful how we manage that, so we don't affect the relationship, the provisional relationship with the client." FG Community Participant 3 "Absolutely. We go into homes and there's medications all over the floor, and we can't do anything about that, we just have to report it." --- FG Community Participant 2 Several community nurses and PCAs stressed that they fostered client independence and encouraged family involvement, rather than let the service 'take over'. Paradoxically, several relatives of older people receiving community aged care reported situations where they were not consulted when restraint was proposed and subsequently used: "The community nurse called Dad's GP and to her credit, the GP was reluctant to put Dad on this antipsychotic but everybody else was pushing for it, so she did write a prescription and they used it and then they let me know by email after it was done." --- Relative 3 Interviewer: "So, they didn't ask your permission at all, it was just more informing you that it happened?" "That's exactly it." --- Relative 3 All three relatives interviewed agreed there is was a strong need for those receiving community-based care and their families to receive guidance on restraint practice. --- Discussion This study provides insight into the attitudes, beliefs and experiences of a diverse group of interdisciplinary stakeholders towards restraint use in aged care. 'Interdisciplinary' is a term used to describe healthcare practitioners from different professional disciplines who work together to manage the care needs of a person. Whenever possible the person and their family should be an integral part of this group [25]. Our interdisciplinary stakeholders were nursing staff, PCAs, physicians, physiotherapists and pharmacists, as well as relatives of people receiving aged care. Aged care clients, also known as consumers or residents, were not directly involved in this research due to ethical considerations around capacity to consent, alongside restrictions and uncertainty associated with COVID-19. However, a group of Australian researchers was able to interview community-based older people about restraint, reporting that they were conscious of this issue and concerned about being on the receiving end of such practice [26]. Those interviewed were most averse to the use of physical restraint and sedation, which were perceived to have the greatest impact concerning limiting choice and self-expression {26]. Several recent systematic reviews have reported that definitions of restraint in the research literature are highly variable, with interpretations differing according to country, aged care setting, and the timeframe in which studies were conducted [3,27,28]. When we asked our participants to define what restraint meant to them we noted similar differences in understanding. The definitions given by our stakeholders tended to vary according to the professional background of the participant; for example, all the physiotherapists defined restraint as restricting the movement of a person. Likewise, all three pharmacist participants referred to restraint as a way of managing behavior. This would be expected given physiotherapists specialize in movement and pharmacists provide advice on medication that affects mood and behavior. Yet, in spite of a degree of professional variation in interpretation, more than half of the participants defined restraint as 'limiting what a person can do' with regards to how they act, move and their overall freedom. This broad definition aligns with the 2019 legislative definition of restraint as 'any practice, device or action that interferes with a consumer's ability to make a decision or restricts a consumer's free movement' [11] and provides some indication that most people in our study conceptualize restraint in line with the new legislation. The second most common definition of restraint cited by participants, predominantly RNs, was that restraint ensures the safety of older people receiving care and reduces the risk of them harming themselves or others. This is not, in essence, a definition but rather provides justification for why restraint is used. The rationalization for restraint; 'under the premise of risk minimization and prevention of harm to self or others', has been reported in Australia as far back as 2005 [28]. It has also been reported in several government enquiries conducted about Oakden, an older person's mental health service in Adelaide, South Australia [29], despite a lack of evidence that restraint, either physical or chemical, protects residents against injuries and falls [3,14,15,28]. Participants were far more likely to define restraint as restricting movement and actions than to control behavior, an aspect which relates to the definition of chemical restraint as 'a practice or intervention that involves the use of medication for the primary purpose of influencing a care recipient's behavior' [30]. This may speak to the difficulty of determining if a psychotropic medication is used for a medical or mental health condition as opposed to influencing behavior. Some older people receiving care may present with a combination of behavioral symptoms and mental illness. The reluctance to define chemical restraint was also observed in a recent systematic review examining the prevalence of restraint practice in residential aged care [16]. The authors of this review could locate only four studies that provided a definition for chemical restraint, compared to 51 studies that explicitly defined physical restraint [16]. The mental health sector has also found it challenging to define chemical restraint, acknowledging 'use remains controversial with different understandings of what it is and its role in care' [31]. It appears more education is needed around what chemical restraint entails in the aged care sector so that it is identified, and when proposed and/or used, practice accords with the requirements set out in the Aged Care Legislation [11] and the Aged Care Quality Standards [32]. Evidence is mounting that legislation appears to be one of the most effective approaches to reduce restraint practice for older people receiving residential aged care. Countries that have introduced legislation in response to high rates of restraint, including the USA and Canada, have subsequently reported significant reductions in use [16,33,34]. One consistent finding in our study was the high level of support for the recent legislative amendments that have been introduced around restraint use in Australian residential aged care [11]. Although some participants were frustrated with the increased reporting associated with the new Legislation, they agreed that documentation and enhanced oversight was necessary to stamp out poor practice and reduce reliance on restraint use. Interestingly, although highly supportive of enhanced regulation, many participants admitted they did not know the specifics of the legislative amendments relating to restraint, pinpointing a need for additional information and training for staff, healthcare practitioners and informal caregivers. Another common theme raised in this research relates to the concept of ensuring a 'restraint-free environment'. The 2012 DMT resources state: 'with a restraint free approach, the use of any restraint must always be the last resort ' [18,19]. Some of the participants expressed the view that in literal terms, 'restraint-free' meant that restraint should never be used, as opposed to being permitted in certain situations. As with restraint, the meaning of a 'restraint-free environment' appears to vary in different settings and between countries. For instance, a group of researchers based in a hospital in the USA claimed they were able to achieve a 'restraint-free environment' in their delirium unit [35]. However, they specifically defined 'restraint-free' as meaning no physical restraint. Antipsychotic and benzodiazepine use was still permitted and there was no reference to the use or absence of environmental restraint [35]. Many of our participants felt that a completely restraint-free environment was unattainable especially when providers require gates to be locked or keypads used to prevent people with severe dementia leaving services unattended. Others stressed that it was important to acknowledge there were certain times when restraint was needed but instead there should be significantly more emphasis on ensuring use was appropriate and that legislative obligations were followed. To circumvent confusion and skepticism around the term 'restraint-free environment' it may be more appropriate to focus instead on using terms such as 'minimizing restraint' or 'restraint as the last resort'. 'Low-level' restraint was another practice over which many participants expressed concern, citing examples such as wedging a wheelchair under a table at mealtimes or placing a walker or care-bell out of reach. The use of these less obvious methods of restraint has also been reported by others [17,36,37]. An ethnological study conducted in Norway, consisting of a mixture of field observation and staff interviews, found that low-level restraint practices were often used to avoid using more overt forms of restraint. These low-level practices also allowed staff to 'get the care work done'. Similar to some of our participants, the Norwegian researchers reported that many of the care staff were not aware that certain practices could be construed as restraint [36]. Training the staff in what restraint means within a human rights framework, detailing the practices it entails, and encouraging open discussion around restraint is vital to mitigate this issue. Participants based in community aged care reported additional challenges with regards to restraint. They reported that due to the ethos of 'consumer directed care' in home care, clients now control the types of care and services they receive [38]. This means that for many clients, clinical care is not provided as part of their home care package. Community care nurses and PCAs would often only visit a client's home briefly so had limited knowledge of the medication they were taking or if they were subject to other forms of restraint. It should be stressed that the new legislative obligations on restraint [11] only apply to residential aged care providers, although the Aged Care Quality Standards [32], which apply in all federally-funded aged care settings, do require providers to minimize restraint and report on restraint as part of governance. Our health professionals reported that families, in general, were more involved with the care of their loved one when they were living in the community and stressed the need to work collaboratively with them. However, several of our community-based participants recounted experiences where their relatives with dementia were commenced on psychotropic medication for behavior control and they only found out after the medication had been administered. Similar experiences of chemical restraint use without informed consent were recounted in a 2019 Human Rights report [8] and at the Royal Commission [10]. Additional training for health care practitioners around the legal requirement for obtaining informed consent is urgently needed to address this issue. In a recent research study [2], also presented as evidence to the Royal Commission [10], rates of psychotropic use in the community were shown to increase markedly in the year prior to aged care admission, providing an indication that the issue of chemical restraint is not just confined to residential care. More research is needed to gauge the extent of restraint practice, both chemical and other forms, in community aged care, particularly given the rapid expansion of the Home Care sector. Further, the knowledge, experiences and opinions of relatives around restraint use in all aged care settings need to be investigated in greater depth as their voice is rarely captured [39]. Community care nurses and PCAs also reported that they had little influence if the client or their family decided to use devices such as bedrails or low beds, both forms of restraint, as they were only responsible for practices that occurred while they were present providing care. Similar barriers were reported in a Dutch qualitative study of restraint use in community settings that concluded that informal caregivers, especially relatives, have a dominant role in the use of restraint [40]. They also reported that relatives were less aware of the harms associated with restraint use and had limited knowledge of alternative strategies. Due to their findings, the researchers have instituted a training program on restraint for informal caregivers, with promising results to date [40]. Similar education and training may be needed in Australia. There are several limitations associated with this study. First, we found it challenging recruiting participants based in community aged care. Although we approached several large community care providers in Tasmania, requesting names of potential staff to interview, volunteers were not forthcoming. Two participants from a large Home Care provider also withdrew on the day of their focus group. We theorize that this lack of recruitment was due to several factors, including staff shortages, limited experience with aged care research, and the emerging COVID-19 situation at the time. As there were fewer participants from the community compared with the sample from residential aged care, their experiences and opinions regarding restraint may not be representative of the community aged care sector overall. Second, we need to note that all participants were asked to read a 2012 resource on restraint [18,19] before being interviewed or involved in one of the focus groups. This pre-reading may have increased knowledge and influenced viewpoints around restraint use, meaning that their opinions and familiarity about this topic may not be representative of key stakeholders overall. A final limitation involved the online nature of the interviews and focus groups owing to restrictions imposed at the initial stages of the pandemic. We inevitably experienced technical difficulties with sound and vision with some participants, resulting in several interviews having to be rescheduled, and some conducted solely using mobile phones. Participants may not have been as willing to express their opinion due to the lack of direct personal engagement with the interviewer, and with the lack of interaction with other participants in focus groups. Despite these limitations, we felt that all participants were very accommodating of the situation and that rich and insightful data regarding restraint in aged care settings were obtained. --- Conclusions This study was conducted to inform and update Government resources on restraint to align with legislative changes and contemporary best practice. The findings suggest that many nursing and care staff, health practitioners and relatives have a broad understanding of what restraint means. However, additional education is needed for these stakeholders on a more conceptual level about what restraint involves, particularly how to recognize and minimize low-level forms and chemical restraint. Likewise, resources need to contain information about the new Aged Care Legislation relating to restraint and what this means for providers and consumers of aged care services. Resources for providers advocating restraint-free environments were considered aspirational. Instead more practical guidance on what to do to prevent and minimize restraint, as well as what to do when restraint is judged appropriate was sought. Research is needed to investigate restraint practice in community aged care as use is largely unknown. --- Data Availability Statement: Full anonymized data supporting reported results are available by contacting the lead author via [email protected]. ---

Restraint use in Australian residential aged care has been highlighted by the media, and investigated by researchers, government and advocacy bodies. In 2018, the Royal Commission into Aged Care selected 'Restraint' as a key focus of inquiry. Subsequently, Federal legislation was passed to ensure restraint is only used in residential aged care services as the 'last resort'. To inform and develop Government educational resources, we conducted qualitative research to gain greater understanding of the experiences and attitudes of aged care stakeholders around restraint practice. Semi-structured interviews were held with 28 participants, comprising nurses, care staff, physicians, physiotherapists, pharmacists and relatives. Two focus groups were also conducted to ascertain the views of residential and community aged care senior management staff. Data were thematically analyzed using a pragmatic approach of inductive and deductive coding and theme development. Five themes were identified during the study: 1. Understanding of restraint; 2. Support for legislation; 3. Restraint-free environments are not possible; 4. Low-level restraint; 5. Restraint in the community is uncharted. Although most staff, health practitioners and relatives have a basic understanding of restraint, more education is needed at a conceptual level to enable them to identify and avoid restraint practice, particularly 'low-level' forms and chemical restraint. There was strong support for the new restraint regulations, but most interviewees admitted they were unsure what the legislation entailed. With regards to resources, stakeholders wanted recognition that there were times when restraint was necessary and advice on what to do in these situations, as opposed to unrealistic aspirations for restraint-free care. Stakeholders reported greater oversight of restraint in residential aged care but specified that community restraint use was largely unknown. Research is needed to investigate the extent and types of restraint practice in community aged care.

Introduction Fluctuating and debilitating muscle weakness involving facial, bulbar, cervical, axial, and limb muscles is the hallmark of generalized myasthenia gravis -a rare, autoimmune, chronic neuromuscular junction disorder . gMG affects more than 50,000 individuals in the United States and is more commonly diagnosed in women <40 years and men >60 years of age . Although symptomatic immunosuppressive medications used as standard-of-care treatments in the management of gMG can be effective , individuals living with gMG continue to experience a high burden of disease , with up to 20% experiencing a myasthenic crisis requiring intensive care and respiratory support over their lifetime . Impairment of daily activities including eating, driving, walking, and housework due to clinical symptoms of gMG can be compounded by common comorbidities such as hypertension and diabetes . Moreover, mood disorders, which can lead to clinical worsening of gMG, affect 41% of diagnosed individuals , contributing to low health-related quality of life . Substantial economic burden is attributed to high annual medical costs, challenges in maintaining employment, and approximately one-third of diagnosed individuals requiring frequent care from their primary caregiver . While multiple recently launched biologic treatment options offer hope for improved outcomes , the aforementioned challenges still remain in gMG management as the landscape of care continues to evolve . Critically, a major knowledge gap exists in understanding the burden and impact of social determinants of health , defined as nonmedical, nonbiological, and nongenetic factors of the environment or individuals' lives that affect health, functioning, and HRQoL outcomes-including conditions in which people are born, grow, work, live, and age -in gMG. SDOH has emerged as a key consideration across healthcare as SDOH barriers can be associated with increased health risks and implicit bias that can affect patient-provider interactions and patient health outcomes . Age, employment, and education status were shown to affect HRQoL in gMG in Germany and Poland , and both treatment utilization and clinical outcomes significantly differed by race among individuals who were hospitalized for gMG in the United States . While SDOH-based variances have been reported around clinical phenotypes of gMG , further evidence demonstrating the impact of SDOH on day-to-day experiences in gMG is limited. Better understanding the impact of SDOH on individuals' support needs is vital for gMG support networks to reach those who may benefit the most. In gMG, support is commonly offered through patient advocacy groups , which are nonprofit organizations that offer a variety of resources including, but not limited to, support groups, educational events, and health and wellness resources for patients throughout the diagnostic journey and beyond. PAGs can be regional or national, with a broad range of scope including representing the patient voice with policymakers, promoting the needs of patients and families, and raising awareness of the condition to the general public. Patient support in gMG is also offered through patient support programs , often sponsored by providers, pharmacy benefit managers, payers, PAGs, or pharmaceutical companies. Offering interventions that aim to improve access, usage, and adherence to treatments and disease self-management, PSPs can have a positive impact on clinical, humanistic, and economic outcomes , especially when designed to match the needs of individuals who are seeking support. For individuals living with gMG and SDOH barriers, targeted PSP approaches may alleviate common challenges associated with gMG, such as receiving a correct diagnosis, finding a specialized care team, accessing treatment , and other unique unmet needs . Although identifying the needs of historically marginalized populations should be a priority in rare diseases, evidence is still scarce. Those who may need increased support-including people of color, lower socioeconomic status, those on public insurance, and/or with limited healthcare options due to geography-are still critically underrepresented in a majority of studies, in part due to difficulties in communicating and engaging this subpopulation in research initiatives using conventional recruiting approaches . To address these questions, we conducted a sequential mixedmethods study aimed to better characterize how SDOH barriers impact individuals from diverse racial, ethnic, educational, and socioeconomic backgrounds living with gMG across 4 phases of the diagnosis and treatment experience: diagnosis, accessing treatment, initiating treatment, and continuing treatment. Additionally, we investigated their awareness and utilization levels of existing resources, with the aim of identifying and prioritizing interventions that can help mitigate specific challenges associated with SDOH barriers. --- Materials and methods --- Study design and ethics This mixed-methods study included qualitative interviews followed by a cross-sectional multimodal quantitative survey. All participants received compensation for their participation in this study. Records of included participants were held strictly confidential using standard protected health information security guidelines under the Health Insurance Portability and Accountability Act of 1996 . Protocols and materials used in the study received Institutional Review Board Approval . quotas were implemented to include a balanced spread of various baseline demographic categories including age, gender, employment status, ethnic and racial background, living environment, education status, and annual family income . For caregivers and PAG representatives, separate inclusion criteria were also defined . --- Qualitative interviews 2.2.1. Participant recruitment Individuals living with gMG, caregivers of those living with gMG, and representatives from gMG PAGs were recruited. The identity of the study sponsor was masked. To overcome the inherent challenges of recruiting a heterogeneous cohort of individuals experiencing SDOH barriers, recruiting strategies included partnering with multiple third-party vendors with robust rare disease networks and collaborating with several gMG PAGs to broaden the reach. --- Interviews Participants who met the inclusion criteria and provided written informed consent were invited to a double-blinded 45-min in-depth qualitative web-assisted phone interview. The moderator was an academic researcher identifying as a woman of color, with expertise engaging and facilitating conversations around SDOH and intersectionality. No identifying information was collected during the interviews. --- Survey design De-identified interview recordings and transcripts were used to categorize, analyze, and curate descriptive themes. Additional information was extracted from a previously described screener , existing literature, and a landscape assessment of currently available gMG support resources. These insights were combined to develop questions and relevant response options to identify common concerns and support needs of individuals living with gMG and SDOH barriers overall and at 4 distinct phases throughout the diagnosis and treatment experience defined by: diagnosis, accessing treatment, initiating treatment, and continuing treatment. --- Quantitative survey and analysis --- Respondent recruitment Recruitment of individuals living with gMG occurred for a total of 8 weeks, with 2 independent third-party vendors recruiting individuals through their rare disease patient panels and outreach networks, with additional support from US-based gMG PAGs . Potential respondents were directed to a screener questionnaire . Similar to recruitment for the interviews, quotas were predefined to include a balanced spread of various baseline demographic categories including age, gender, employment status, ethnic, and racial background, living environment, education status, and annual family income to ensure a diverse sample of respondents . --- Data collection Those who fulfilled the inclusion criteria were redirected to the online survey, designed to be completed in approximately 20 min. The survey was offered in English or Spanish. As not all individuals living with gMG may be physically and/or emotionally comfortable or capable of completing the survey independently, optional telephone assistance in English or Spanish was offered. --- Data analysis De-identified data were aggregated and analyzed. Exploratory subgroup analyses were conducted using pre-defined SDOH definitions. Results were exported to generate corresponding tables or charts. Quality assurance of data was conducted during fielding, pre-analysis, and post-analysis. S4). Most interviewees identified as non-White/Caucasian , were covered by public insurance , and were currently unemployed . --- Results --- Qualitative interviews --- Interview insights Overall, financial constraints and profound lifestyle shifts were most commonly expressed by interviewees as challenges throughout their experience of receiving a diagnosis and treatment for gMG. Additional barriers such as navigating insurance coverage, maintaining employment, and accessing resources were also commonly mentioned. Interviewees expressed that the logistical and administrative burden of addressing these challenges exacerbated their clinical manifestations of gMG, such as fatigue, especially when combined with treatment side effects and comorbidities. Another critical theme centered around detrimental effects on mental health, including anxiety and fear of exacerbations, social isolation, long periods of emotional distress or despair, and/or suicidal ideation. This mental health burden shifted over time from diagnosis to post-treatment . Interviewees discussed that the quality of educational resources they were aware of or were currently using to help understand their diagnosis and treatment options was poor; as a result, feeling misunderstood or burdening others often led to communication challenges. When interviewees were presented with several existing gMG support programs and offerings, overall awareness of these resources was low. Individuals living with gMG, and their caregivers, often sought gMG information and support through social mediabased online community groups, with low awareness of other available support such as PAGs. --- . Respondent demographics and characteristics Thirty-eight individuals living with gMG met the inclusion criteria and completed the web-based survey. Among the 38 respondents, age, racial/ethnic background, gender, living environment, education level, insurance type, and employment status were diverse . Notably, 60.5% of respondents identified as non-White/Caucasian, and at least 1 participant in every pre-specified racial/ethnic background subgroup completed the survey. Eleven respondents reported high school or General Education Development as their highest education status, 13 respondents were enrolled in Medicaid, 17 respondents were enrolled in Medicare, and 23 respondents were unemployed. Despite efforts to actively recruit male respondents, significantly more respondents identifying as women were represented , with one participant identifying as binary gender nonconforming . --- Overall concerns, resource usage, and support needs Overall, financial concerns were highlighted as the most common challenges experienced in daily life, as 76% of respondents reported problems making ends meet at the end of the month, and 50% reported worry or concern that they may not have stable housing . While the majority of respondents reported high levels of agency in healthcare decision-making with their doctors , a distinct cohort of up to 34% of respondents expressed low to neutral confidence and comfort across facets of their relationship with their doctors, most notably including feeling empowered in managing their gMG and trust in their doctor's gMG expertise . Further, 50% of respondents reported low to neutral confidence in navigating their healthcare and treatment experience with ease, and 39% of respondents expressed unmet needs around social support and disease education. Most respondents reported to seek gMG information through their HCPs , followed closely by internet websites and social media-based communities . Regarding the quality of currently available gMG information, most respondents had neutral impressions . Consistency of information had the most negative perceptions, with 29% of respondents expressing that currently available information from different sources was conflicting rather than consistent . --- Concerns and support needs by phase of the diagnosis and treatment experience 3.2.3.1. Diagnosis The most common concerns at the time of diagnosis were how the diagnosis may change their lifestyle and managing fear and confusion due to not understanding gMG , which highlighted feelings of uncertainty and potential negative effects on mental health . Approximately one-third of respondents also expressed logistical challenges, including finding the right specialist and extended time taken to receive a correct diagnosis . Most respondents were unaware of many of the support offerings and resources available through PSPs at diagnosis, aside from general disease information . Reflecting concerns around disease education and mental health, respondents commonly reported that the most valuable resources at this phase would have been nurse navigator support ; general disease information resources ; information around disease symptoms, testing, and treatment options ; and customized mental health services . Most respondents preferred to receive these support resources from their doctor, followed by patient support groups and other individuals living with gMG . --- Accessing treatment When seeking access to treatment, the most common concerns were financial, including qualifying and paying for treatment and navigating the logistics of assistance programs . Anxiety related to financial concerns was common in this phase . Despite high levels of financial and access-related concerns, awareness and utilization of available financial resources such as copay cards and free medication was extremely low . Respondents most commonly expressed that a list of funding assistance with guidance on how to obtain support and free medication would have been valuable . In addition, nurse case managers trained in various areas such as navigating insurance , identifying resources for Medicaid or Medicare , and/or connecting individuals to mental health services and community --- N = 38 You have had problems making ends meet at the end of the month. 76% You have felt worried/concerned that you may not have stable housing that you own, rent, or stay in as a part of a household. 50% You have felt like you received unequal treatment due to your socioeconomic status . 32% You have had to skip buying medications or going to doctors' appointments and/or treatments to save money. 32% You have put off or neglected going to the doctor because of distance or transportation. 32% The electric, gas, oil, or water company has threatened to shut off services to your home. 29% You have felt judged due to not understanding something new or unfamiliar to you by someone with more education. 24% You have felt worried that the place you are living now is making you sick or unsafe . 21% You have felt like you received unequal treatment due to your race/ethnicity/skin color. 13% You have had difficulty communicating with someone due to English not being your fist language. 0% gMG, generalized myasthenia gravis. Respondents were asked to select all statements that were relevant in their day-to-day lives over the past 6 months. Statements are ordered from largest to smallest proportion of respondents reporting the statement as a day-to-day concern. Statements were shown in a randomized order during the questionnaire. 10.3389/fpubh.2023.1147489 Frontiers in Public Health 06 frontiersin.org resources were also expressed as resources that would have been valuable. Most respondents preferred to obtain these resources from their doctor or patient support groups , followed by PAGs and manufacturers . --- Initiating treatment Preparing for potential side effects , acceptance of lifestyle changes , and managing their current lifestyle while on treatment comprised the most common concerns at treatment initiation, highlighting the sensitive nature of initiating treatment and its potential effects on anxiety . The majority of respondents had greater awareness and utilization of available treatment-and lifestyle-related resources when initiating treatment; however, awareness was low around nurse case managers and treatment option discussion guides . Resources that would have been the most valuable during this phase were most commonly reported as educational resources , nurse case managers , and a guide to managing comorbidities . Most respondents preferred to obtain these resources from their doctor or patient support groups , followed by PAGs . --- Continuing treatment When continuing treatment, common concerns included consistent fear of experiencing an unexpected crisis and considering long-term lifestyle adjustments , signifying that treatment-related challenges and anxieties continue even after the treatment initiation phase . Most respondents engaged with patient communities in this phase for support, with lower awareness of other available tools such as guides for finding a qualified gMG neurologist and search tools to find the closest infusion centers . Consistent with common concerns in this phase, resources that would have been the most valuable were quick emergency resources to bring during a crisis or exacerbation and resources providing tips for maintaining physical and mental health in gMG . Respondents preferred to receive these resources from their doctors or patient support groups , followed by other individuals living with gMG and PAGs . --- Discussion In this study, we aimed to better characterize the experiences of individuals living with gMG who belong to historically underrepresented communities based on various SDOH challenges. We utilized a unique approach to capture their awareness and utilization of currently available PSPs, with the aim to better tailor PSPs to their support needs. This study begins to address a critical knowledge gap, highlighting previously underexplored perspectives to understand how support for diagnosed individuals, their social environment, and the healthcare system can be further improved for rare diseases such as gMG. 4.1. Key learning 1: Individuals living with gMG and SDOH barriers face a complex set of challenges across diagnosis, accessing treatment, initiating treatment, and continuing treatment; most of these challenges centered around financial concerns and mental health Among the diverse study cohort, financial concerns and mental health were expressed as the most common gMG-related --- N = 38 --- Diagnosis --- How your gMG diagnosis might change your lifestyle 61% Fear and confusion due to not understanding gMG and its symptoms 45% Managing your gMG diagnosis and care planning due to reduced energy/fatigue 39% --- Accessing treatment Determining if you are qualified for your treatment and how you will pay for it 47% Managing the logistics of applying for disability and other assistance programs 45% Mental stress due to uncertainty of how treatment would be paid for 39% --- Initiating treatment Preparing for potential side effects from treatment 71% Acceptance of the many life changes resulting from gMG diagnosis 66% Managing your current lifestyle while on treatment 55% --- Continuing treatment Consistent fear of experiencing an unexpected crisis 50% Quality of life/lifestyle adjustments required to live with gMG for the long term 45% Challenging maintaining lifestyle given reduced energy levels/fatigue 34% gMG, generalized myasthenia gravis. Respondents were asked to select 3 statements that were most relevant in each phase. The 3 most commonly reported concerns for each phase are shown in this table; the full dataset is available in the Supplementary Figure S2. Statements were shown in a randomized order during the questionnaire. concerns. As expected, financial concerns were emphasized while accessing treatment. Negative effects on mental health were common throughout all phases but associated with different underlying concerns at each phase, demonstrating the multifaceted and dynamic nature of living with gMG. These core concerns evolved from fear and confusion toward understanding gMG at diagnosis, to financial concerns when accessing treatment, to short- and longterm lifestyle changes when initiating and continuing treatment. Although these challenges have been reflected in more general populations of individuals living with gMG , financial and mental health challenges may be further amplified for those living with SDOH barriers. As depressive symptoms can negatively affect HRQoL in gMG , mental health should be particularly monitored in underrepresented communities on a consistent basis. --- Key learning 2: Well-publicized and easy-to-access educational and financial resources would be most valuable for individuals living with gMG and SDOH barriers Our results highlighted an overall low awareness and utilization of existing support services among individuals living with gMG and SDOH barriers. This could be attributed to not only the general scarcity of gMG-specific resources and PSPs as in many rare diseases, but also systemic barriers to finding, accessing, and/or receiving existing services for those experiencing SDOH barriers. Although educational resources were commonly perceived as the most valuable resources throughout the diagnosis and treatment continuum, some negative perceptions were captured on the consistency of available information, highlighting the importance of continuing to improve exposure and user-friendliness of accurate and well-curated resources. --- N = 38 --- Diagnosis --- Most commonly used Resources that provide general disease information about gMG 61% used Resources on commonly used gMG terms to help understand gMG 55% used --- Most unaware Resources that explain diagnostic testing options and how to get tested 63% unaware Periodic updates sent with gMG information, developments, and resources 63% unaware --- Accessing treatment --- Most commonly used Services to verify insurance benefits and coverage 37% used Financial support from a foundation or advocacy group 32% used --- Most unaware Free medication given while waiting to hear if insurance will cover costs 84% unaware Copay card to help cover out-of-pocket costs 74% unaware --- Initiating treatment --- Most commonly used Packets or brochures with gMG treatment information 63% used Periodic updates sent with gMG information, developments, and resources 47% used --- Most unaware Nurse case manager to support the beginning of treatment journey 79% unaware Discussion guides to help discuss treatment options with doctor 61% unaware --- Continuing treatment --- Most commonly used Forum where stories of people living with gMG can be shared with others 76% used Forum where stories of people living with gMG on treatment can be shared 68% used --- Most unaware Resource guide on the steps to take to find a qualified gMG neurologist 66% unaware Tool to search for nearby infusion centers for treatment using your zip code 58% unaware gMG, generalized myasthenia gravis. Respondents were asked to select 1 of 3 choices for each of the provided statements describing currently available resources: "Resources used, " "Resources aware of but not used, " and "Resources unaware of. " Statements shown here are the 2 most commonly used and 2 most commonly expressed as unaware of and are paraphrased from actual statements; the full dataset is available in the Supplementary Figure S4. Statements were shown in a randomized order during the questionnaire. The high perceived value of general disease information that discusses symptoms, diagnostic testing, and treatment options at diagnosis and treatment initiation illustrates the need to deliver high-quality information during this critical time. Particularly when accessing treatment, financial support was expressed as most valuable. Despite the high perceived value associated with financial support resources such as funding assistance and free medication, respondents had a low awareness of existing programs that offer such support. To better meet the needs of individuals living with gMG and SDOH barriers, potential systemic barriers to access these resources should be further explored and addressed. --- Key learning 3: "High-touch" and tailored support programs may be better suited to improve outcomes for underserved populations Throughout the diagnosis and treatment experience, respondents preferred to receive support resources from HCPs, closely followed by patient support groups. While the preference for HCPs is not surprising, heavy reliance on patient support groups could reflect unmet needs for diverse disease-related, cultural, and personal support, especially in underserved and underrepresented communities . Notably, nurse case managers-who offer personalized guidance for individuals living with gMG in a variety of realms, including disease education, navigating insurance, tracking symptoms, and sources for financial assistance-were commonly perceived as a valuable resource. However, their current awareness and utilization were low, which could in part be due to the lack of their availability at healthcare facilities where individuals receive their care. PSPs offering nurse case managers and similar services that are high in "human touch" have been shown to increase medication adherence, satisfaction, HRQoL, and/or lower total medical costs in many chronic diseases including diabetes, hypertension, cancer, and autoimmune conditions . Across these therapy areas, case management has been associated with decreased emergency care utilization, hospital admissions, readmissions, and length of stay, reducing healthcare system costs . In one study examining patientcentered coaching interventions in 321 discharged Medicare users, over $3,700 of cost savings per patient were reported over 6 months compared to controls, with no cost shifting observed to other types of healthcare utilization . Moreover, in individuals living with chronic autoimmune diseases, participation in a PSP with personalized 1-to-1 support was associated with 29% higher medication adherence, 22% lower discontinuation rates, and 35% lower disease-related medical costs compared with non-PSP controls . Similar PSP benefits can be expected in gMG, as our results highlight unmet needs for additional empowerment and education to allow individuals to be better equipped to manage their diagnosis and disease. Such resources can be especially valuable in rare diseases as HCPs may not always be well-equipped to identify and address SDOH challenges with limited time and resources available to support individuals' holistic needs beyond the clinical realm. Improved visibility and access to "high-touch" PSPs in gMG can provide additional support for social workers and community resources who HCPs often rely on to address these needs. For individuals with diabetes, involvement of case managers increased activation of social --- N = 38 --- Diagnosis Nurse navigator to provide support for newly diagnosed individuals, including referring to resources for mental health 42% Resources that provide general disease information about gMG 42% Resources developed by clinical gMG experts that explain gMG symptoms, testing, and treatment options 37% --- Accessing treatment List of funding sources and assistance programs, with guidance on how to obtain support 39% Free medication given while waiting to hear if insurance will cover costs 34% Nurse case manager who can help navigate the insurance process and help understand potential financial assistance programs 34% --- Initiating treatment Educational resources developed by top clinical gMG experts, explaining gMG symptoms, testing, and treatment options 42% Nurse case manager available to support individuals living with gMG starting their treatment journey 39% Guide on how to manage gMG and additional health concerns 34% --- Continuing treatment Quick emergency "takeaway" resources for individuals diagnosed with gMG to carry/bring during an exacerbation or crisis 45% Collection of resources providing tips for practicing better physical and mental health in daily life with gMG 42% Tools to track symptoms/nurse case manager/social community 34% gMG, generalized myasthenia gravis. Respondents were asked to select 3 statements that would have been the most valuable in each phase, regardless of whether they were available or utilized. First, awareness and utilization of existing support services should be increased. To accomplish this, education on available support services should be improved throughout the larger gMG ecosystem, including HCPs, PAGs, social workers, pharmacies, infusion centers, and health plans. This can help provide timely access of support services and resources, such as fact-checked educational materials, trained nurse case managers, and financial assistance, to individuals living with gMG who need them. For HCPs, forming dedicated multidisciplinary care teams or clinics may also enhance gMG knowledge-sharing between clinicians, nurses, rehabilitation specialists, physical/occupational therapists, dieticians, speechlanguage pathologists, and other professionals so that individuals can be connected with available resources as soon as they are diagnosed with gMG. Secondly, updated, high-quality educational resources offered by support programs such as PSPs should be disseminated via an increasingly multichannel approach to accommodate a diverse range of communication preferences. These can include treatment information telephone hotlines, digital symptom trackers, educational webinars, online nurse case managers, and social media campaigns. Importantly, updated offline materials such as pamphlets and printed guides should also be offered at healthcare sites and by mail for those who may prefer them or have limited digital literacy. Increased collaborations throughout the gMG ecosystem can be leveraged to share these resources robustly across communities of individuals living with gMG. To continue better addressing the needs of individuals living with gMG and SDOH barriers in the longer term, financial investments to support initiatives that can improve access to treatment for vulnerable populations living with gMG should be prioritized. These can include increasing availability of multilingual assistance, localized educational programs, and public and private funding assistance. In addition, future research should better identify individuals who may need additional and/or specific types of assistance related to SDOH on a larger scale with the use of large databases such as electronic health records. To gather more data, enrollment of historically underrepresented populations should be increased in gMG research. More broadly, the importance of early detection, early diagnosis, and better screening of mental health symptoms in gMG should be further studied. --- Limitations This cross-sectional study was designed to characterize experiences of a specific cohort, and these data are not intended to be comparative with that of a general population of individuals living with gMG. Expanded longitudinal observational studies are needed to better understand its implications, to further contextualize how SDOH barriers may contribute to health inequities in gMG, and how these data may also be relevant to broader rare diseases. Although SDOH-specific differences are of high relevance, subgroup analyses were exploratory. The sample size was not powered for statistical testing and considerable variability of responses among the sample led to the absence of any strong patterns discerned between SDOH subgroups. The heterogeneity of responses could be in part attributed to the variable and fluctuating nature of gMG, as well as the complex intersectionality of SDOH factors for each individual among the limited study sample. Finally, we used primarily web-based data collection for both recruitment and study phases. Although measures were taken to minimize selection bias for digital literacy through offering telephone assistance to complete the survey , the study cohort may have been enriched with individuals with access to basic web/phone-based services at the minimum. --- Conclusion Our results highlight distinct concerns and unmet needs among a diverse population of individuals with gMG living with various SDOH barriers. Financial and mental health concerns were common overall; additionally, barriers to access and poor disease education were also common, depending on the phase of the diagnosis and treatment experience. Although resource and support needs reflected their concerns, awareness and utilization of existing PSPs were low, emphasizing the need for further targeted, specific, accessible, and well-publicized support. To improve the experiences of historically marginalized individuals in the healthcare system, the larger gMG support network should continue to spotlight these communities to work together to provide further evolved customized and localized support focused on the specific needs of these communities. --- Data availability statement The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. --- Ethics statement The studies involving human participants were reviewed and approved by WCG IRB, Puyallup, WA, United States. The patients/ Author contributions TH, KP, CB, SS, CZ, DM, AW, and GP were involved in developing the concept and designing the study methodology. SS, CZ, DM, and AW oversaw data collection and analysis, and all authors were involved in interpretation. MS drafted the manuscript. All authors contributed to the article and approved the submitted version. --- --- Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. --- Supplementary material The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpubh.2023.1147489/ full#supplementary-material

Introduction: Social determinants of health (SDOH) are important contributors to health outcomes, and better understanding their impact on individuals diagnosed with rare, chronic diseases with high burden and unmet need is critical. Characterizing SDOH burden can help improve the design of patient support programs (PSPs), using targeted approaches to remove barriers to access. Methods: This study used a mixed-methods strategy employing a quantitative survey, which was designed based on qualitative interviews, to understand the unmet needs and awareness/utilization of PSPs among individuals living with generalized myasthenia gravis (gMG) and experiencing SDOH barriers. The survey was completed by 38 individuals living with gMG, of which the majority were non-White/Caucasian, unemployed, low income, and enrolled in public insurance. Common SDOH challenges, awareness/utilization of available PSPs, and unmet needs were identified. Results: Financial and mental health concerns were the most common among individuals living with gMG and experiencing SDOH barriers throughout diagnosis, accessing treatment, initiating treatment, and continuing treatment. Awareness and utilization of existing support services were low, especially when accessing treatment. Educational, financial, and personalized support with high "human touch" were commonly perceived as the most valuable resources. Implications: To better serve the needs of individuals with gMG experiencing SDOH barriers, PSPs should use a targeted approach to offer services tailored to harder-to-reach populations. Further, providers, advocacy groups, manufacturers, and public organizations in the gMG ecosystem should strengthen collaborations with PSPs to enable individuals living with gMG to access the services they need to improve their health outcomes. KEYWORDS patients, caregivers, myasthenia gravis, social determinants of health, burden, patient support services, mixed methods

F or the past century, the US Forest Service has maintained a network of experimental forests and ranges that provide a land base where longterm research, experimental and otherwise, is performed. The network is currently comprised of 81 sites nationwide that range in size from 47 to 22,500 ha . EFRs have largely been viewed as places to conduct biophysical research. For example, research at the H.J. Andrews and Wind River Experimental Forests has shaped understanding of the structure and function of old-growth forest ecosystems . Research at Caspar Creek and Fernow Experimental Forests on the relationship between forests and water has informed the development of watershed management strategies. However, EFRs are also sites where it is possible to perform social science research and, indeed, where social science has a valuable contribution to make, as we show in this article. The advantages to social scientists of working at these sites are many, including the ability to undertake long-term, comparative, and experimental research; the avail-ability of ecological data sets; the presence of a community of researchers with whom to interact and pursue interdisciplinary studies; and built-in research facilities and infrastructure in many places. However, surprisingly, little social science research has been conducted at EFRs to date. The purpose of this article is to encourage researchers to take advantage of the untapped potential at EFRs for social science and for interdisciplinary research across social and natural science disciplines. We describe the human dimensions of EFRs, summarize the social science research that has been conducted at EFRs to illustrate what is possible, present suggestions for studies desired by EFR site coordinators, and examine barriers to social science research at EFRs and suggest how these barriers might be overcome. --- Methods In 2009, we collected information from 80 EFRs about human interactions and the social science research that has been conducted at these sites. [1] First, we conducted a review of the published and gray literature resulting from social science research at or associated with EFRs. Second, we contacted EFR coordinators and scientists using e-mail correspondence and telephone interviews to request information about public uses of EFRs, allowable activities, the social science research associated with each site, proposed or possible future studies, and public outreach activities. Typically, a US Forest Service scientist at each research station serves as the Officer in Charge of one or more EFRs. A total of 80 individuals from five research stations representing 58 EFRs provided information during this data gathering phase. Third, we administered a formal, online survey to US Forest Service personnel affiliated with each site using on online survey tool called Survey Monkey. Its purpose was to systematically verify our findings about human activities, citizen outreach, and social science research at EFRs. We invited key contacts for the 80 EFRs to participate. A total of 68 surveys were completed, representing an 85% response rate. Survey data were entered into an Excel spreadsheet and analyzed. --- EFRs: Human Landscapes EFRs should not be viewed as isolated, empty, untrammeled places. Nor should they be conceptualized as gated or discrete outdoor laboratories accessible only to scientists. Humans have long been a part of these ecosystems and have influenced present con-ditions there. Lands currently managed as EFRs have strong links to the past and often are sites of historical or cultural significance. They are also places used by neighboring residents and nonlocal visitors for recreation, hunting, fishing, and gathering for subsistence and for connecting to sacred or special places. In addition, EFRs serve as natural outdoor classrooms for students, educators, scholars, and resource professionals. All of these uses suggest fertile ground for social science research. Many EFRs contain important cultural resources, indicating their rich human history. Our survey of EFR contacts found that historical, archaeological, and cultural heritage sites occur at many EFRs . More than one-half contain historic structures, and many have archaeological, cultural, or sacred sites. Examples include historic structures installed by the Civilian Today,35.4% of all EFRs in the network lie within 50 mi of an urban area, and just over 15% are located within 25 mi of an urban area, suggesting the potential for urban visitors to use these sites . Another 20.1% of EFRs are fairly remote, located more than 100 mi from an urban area. Human uses of EFRs vary, depending on their unique geographic setting, topographic features, and accessibility. Only two EFRs in the network are completely closed to the public, although one of these is leased by private hunting clubs. Sixteen other EFRs implement special limitations on human use, such as closed areas, restrictions on certain activities, or seasonal closures. Typically, management for human uses is consistent with and determined by the national forest or grassland or other government agency with which it is affiliated. Table 2 shows the number of EFRs at which different human uses occur. The most common uses are walking/hiking/running and hunting, which take place at nearly 90% of EFRs. Other common uses occurring at over 75% of EFRs are recreation activities, including wildlife viewing, vehicle driving, and scenic viewing. Timber harvesting occurs at many sites , although some of this harvest is conducted as part of the research mission of the site. Less common are motor boating, mining, grazing, commercial recreation, and outdoor events, which take place at fewer than 25% of EFRs. These data suggest that EFRs are currently used by local and nonlocal visitors in a variety of ways. Conversations with EFR contacts revealed that often these visitors are unaware of jurisdictional boundaries and may not know about the research mission of the site. They range from being places where it is simply convenient to walk a dog, to places such as Bent Creek Experimental Forest in North Carolina, which has one of the most premier urban-proximate mountain bike trail networks in the United States. When EFRs are adjacent to national forests with high visitation, or specially designated management areas such as wild and scenic rivers, human activity on the EFR may be affected. Many EFR scientists and site managers conduct public outreach activities by involving stakeholders, communities, other scientists, students, and citizens in their scientific endeavors ( --- Social Science Research at EFRs Although EFRs have largely been viewed as natural laboratories for conducting ecological research relating to forest and rangeland ecosystems, more social science research has occurred in association with these sites than is evident at first glance, and several studies are currently ongoing or proposed. In some cases, social science research has been one component of a larger study focusing on biophysical research questions. In other cases, social science research has been conducted on the national forest or rangeland, or in the larger region, where the EFR occurs and has included it, but has not exclusively focused on it. We group the so-cial science research that has occurred at or in association with EFRs into six topic areas: human uses; economics; studies of the past; human dynamics at the wildland-urban interface; human values relating to forests, rangelands, and their management; and interdisciplinary social-ecological research. --- Human Uses People use EFRs in numerous ways that potentially influence ecosystem processes and functions. Understanding the nature of these human uses, their ecological effects, and how to manage them has emerged as one area of research at EFRs. One reason the EFR network was originally established was to study the effects of grazing and silvicultural practices on forest and rangeland ecosystems and to use this understanding to inform management . For example, research on grazing, rangeland ecology, and management have been ongoing for nearly 100 years at the Great Basin Experimental Range and Fort Valley Experimental Forest . Similarly, the Bent Creek Experimental Forest and the Crossett Experimental Forest are sites where nearly 100 years of silvicultural research has provided a foundation of knowledge for forest management . Recreation research began at EFRs in the 1950s with an experimental study at the San Joaquin Experimental Range that tested a permit registration card developed for regulating the number and timing of sport hunters . EFRs have since served as sites for experimental research relating to recreation and its impacts, as well as sites where ongoing recreation activities have been studied. For example, an experimental study at Hubbard Brook assessed the effects of hiking and associated trampling on ground flora along trails . In at least five EFRs, there is ongoing or proposed research on how recreation uses affect wildlife and forest and range conditions. At four additional sites, scientists expressed a desire for future research on hunting and off-highway vehicle use because of concerns about the effects of these activities on natural resources and on study plots. One example of this type of study is highlighted in Box 1. Administrators and scientists contacted for this study expressed particular interest in future research to help them learn more about EFR visitors and the experiences and activities they seek. Interestingly, much of the research on human uses at EFRs has been conducted by natural scientists rather than social scientists because the main object of study has been the environmental effects of human activities rather than the people engaged in these activities. Humans are typically treated as a disturbance factor. This is particularly true for studies using experimental designs. Future interdisciplinary research on human uses that includes social scientists could be undertaken to gain insight into the motivations and behaviors of forest and range users. --- Economics Economics research at EFRs dates back to the 1930s and represents the longeststanding area of social science research conducted at EFRs. From the 1930s to the 1950s, economics research took place at the Desert Experimental Range to evaluate the economic impacts of alternate grazing strategies . Studies to assess the economics of different silvicultural treatments in the context of forest management for timber production have also occurred for decades, having been performed on at least eight EFRs nationwide . More recent economic studies at EFRs have focused on recreation. For example, current studies at Bent Creek Experimental Forest are investigating the economic impacts of recreation activities on the local economy and the broader economic value of recreation there . EFR scientists and administrators identified ecosystem services as a desirable area for future economics research. Rarely has comparative social science research of any kind been conducted across multiple EFRs. One exception is a study of forest-economy relations at six long-term ecological research sites , four of which are EFRs. [2] This study is highlighted in Box 2. --- Studying the Past A third area of social science research consists of studies of historic and prehistoric uses of areas now being managed as EFRs and how these activities have shaped ecosystems there. The earliest historical study comes from the Bent Creek Experimental Forest and contains a detailed reconstruction of the land use history of individual ownership tracts that now make up the forest . Several At the Starkey Experimental Forest and Range in Oregon, a 25,000-ac ungulate-proof enclosure was constructed in 1987 that makes it possible to study interactions between humans and ungulates using controlled research experiments. Scientists have been doing just this since 1991. A 10-year experimental study of the effects of hunting pressure on elk and mule deer found that elk respond to hunting pressure by fleeing disturbance, and that hunter density and traffic counts were positively correlated with elk velocity. Hunting poses significant energetic costs to elk. In contrast, deer responded to hunting pressure by hiding, which entails much lower energy costs. Habitat use by both species did not shift, but daily patterns of use were disrupted. Archery hunts affected animal movements for longer periods during a day than rifle hunts. Another study at Starkey evaluated the impacts of off-road recreation-including all-terrain vehicle use, horseback riding, mountain biking, and hiking-on elk and mule deer using controlled experimental treatments. Scientists found that off-road recreation activities have a substantial effect on elk behavior, causing their movement rates and flight responses to become highly elevated. ATV use and mountain biking provoked the greatest travel response. ATVs caused elk to decrease their feeding time, and mountain biking and hiking caused them to be restless. However, mule deer exhibited little response to off-road recreation activities. The results of these studies can inform hunting regulations and management guidelines for off-road recreation in national forests throughout the west where recreation, elk, and mule deer co-occur. studies that examine how historic settlement patterns and land-use activities have shaped present forest conditions at experimental forests have been conducted since the 1990s . One study looked at how past forest management practices on an EFR shaped the social and economic history of the surrounding region . Other studies simply document the history of individual EFRs and the research endeavors that have occurred there . The only archaeological study we are aware of comes from the San Joaquin Experimental Range . Be-tween 1978 and 1995, students in the Anthropology Department at Fresno City College conducted surveys of cultural resources from the Indian and early ranching and mining periods. The students learned archaeological survey methods, and the Range gained valuable information about its cultural resources. This information included an inventory and description of the sites and their archaeological, ethnographic, and historic backgrounds and maps of their geographic locations. Another research area looks at how indigenous peoples used and managed natural resources on and around EFRs in the past. One area of inquiry concerns Indian burning practices and the use of fire for forest management in historic and prehistoric times . Box 3 highlights a study of past natural resource use practices by American Indians at Cascade Head Experimental Forest in coastal Oregon. --- Human Dynamics at the Wildland-Urban Interface A fourth area of inquiry looks at the effects of nearby development and changing demographic patterns on EFRs and the broader landscapes of which they are a part. Population growth and associated development have occurred around many national forests and rangelands, driven by an influx of urban residents, retirees, second-home owners, and telecommuters who wish to live closer to natural areas. On average, population density within 50 mi of an EFR was higher in 2000 than the US average ; and population growth within 50 mi of the five EFRs that saw the most growth ranged from 43.8 to 67. 8% between 1990 and2000 . Although few studies have been conducted around EFRs specifically that focus on human dynamics at the wildland-urban interface, two have been proposed, both in the southern United States. People interviewed for this study expressed a high level of interest in better understanding the implications of population and land-use change around EFRs. Some scientists worried that the arrival of new residents could threaten the research mission of EFRs. --- Human Values Studies of peoples' values and beliefs relating to forests, rangelands, and their management were initiated in association with EFRs in the 1970s and 1980s. These early studies involved showing photographs to people living in the surrounding area to examine aesthetic preferences regarding alternative silvicultural treatments . Scientists were apparently motivated by public concern over the visual impacts of clearcutting, a harvest method that was used in experimental treatments at these forests. The goal of federal forest management shifted from sustainedyield timber production to ecosystem management in the 1990s. Human values research shifted accordingly, focusing on treatments associated with the goals of ecosystem management and peoples' scenic preferences. Study findings provided information about the social acceptability of different harvest practices, an important consideration in determining which practices to implement. The research highlighted in Box 4 is one example of this kind of project. Another kind of human values study that has been performed at EFRs solicits input from members of the public on specific management issues, such as forest road clo- Comparative research across six LTERs, four of which are experimental forests, was undertaken to develop a framework for assessing the multiple relationships between forest ecosystems and regional economies and to improve understanding of the variables that influence these relationships. At three EFRs, timber harvesting was the most economically important forest management activity. Although demand for timber was increasing, jobs and income associated with converting a unit of forest resources to timber products were either stagnant, and likely to decline , or were decreasing . At all sites, researchers found that allocating forest resources to intensive timber production in the future would likely generate fewer jobs and lower incomes than in the past. Furthermore, at all sites, competing demands for forest resources-such as other commercial uses and intrinsic forest values such as biodiversity and other ecosystem services-are likely to increase relative to timber industry demands in the future. Thus, communities that continue to depend on the timber industry are predicted to stagnate or contract economically. The framework for evaluating forest ecosystem-economy relations developed at these case study sites can be applied elsewhere to help managers evaluate the economic impacts of forest management decisions. sures or restoration activities. The findings can be used for planning and management. Several of the EFR scientists and administrators mentioned research on human values as being particularly desirable. They expressed interest in continued research on attitudes toward silvicultural treatments. More broadly, the EFR network provides a unique opportunity to bring people on site and gather information about their perceptions of forest treatments, as well as postfire conditions, road characteristics, wildlife distribution, recreation density, and other features. EFRs could also be used to learn more about how people and children connect with the natural world by exposing them to different kinds of forests and natural places. --- Interdisciplinary Social-Ecological Research Finally, a number of integrated social and ecological studies have occurred or are underway, especially at EFRs that are LTERs. Six EFRs participate in the LTER network: Baltimore , Bonanza Creek , Hubbard Brook , H.J. Andrews , Coweeta , and Luquillo . The best examples of interdisciplinary socialecological studies come from these sites. The most prolific site in terms of interdisciplinary research has been the Baltimore Ecosystem Study, where numerous studies that measure interactions between the social, physical, and ecological components of the urban ecosystem have been published over the past 10 years . Around Bonanza Creek the emphasis has been on studying adaptation and resilience in arctic socioecological systems in response to global social and environmental change . New research being proposed will focus on "moose-human systems," investigating place-based feedbacks between ecological disturbance-moose interactions, human uses of moose, and human responses to change. Since the mid-1990s, interdisciplinary research at Coweeta has included studies of the social, economic, and environmental factors driving land cover change in the southern Appalachians region and the consequences of these changes for water quality and quantity, terrestrial and aquatic biodiversity, and regional carbon cycles . Box 5 describes integrated research at Luquillo that focuses on interactions between human and natural disturbances and how they affect forest characteristics. Interdisciplinary social-ecological research still has to materialize in a substantive way at Hubbard Brook and H.J. Andrews, although social science has been conducted at both locations. --- Overcoming Barriers to Social Science Research at EFRs The extent of human uses of EFRs documented here, together with the range of social science research topics that can be addressed, suggests that these sites hold enormous potential for contributing to our understanding of human-natural resource interactions-past, present, and future. Despite these potential contributions, the volume of social science research conducted at EFRs, to date, is quite small compared with that of biophysical research. One recent overview of 79 EFRs lists for each site the reasons they were established, the key research topics pursued, major research accomplishments, and the long-term databases they maintain . Are forest management practices such as timber harvesting and prescribed burning that try to mimic past natural disturbance regimes socially acceptable? H.J. Andrews has been a partner in a study designed to develop and evaluate disturbance-based management systems. Thus, residents of the watershed that contains the H.J. Andrews Experimental Forest and residents living in nearby cities were surveyed to find out their knowledge about and perceptions of disturbance-based management and their level of support for it. The survey found that over one-half of the respondents supported disturbance-based management, although about one-quarter did not fully understand it. Concerns revolved around timber harvesting and associated road building. Favorable past interactions between citizens and US Forest Service employees was the variable most highly correlated with the tendency to support it. The research findings will help managers communicate with and engage the public in developing ecosystem management strategies and overcome barriers to implementing disturbance-based management regimes. Box 5. The effects of interacting natural and human disturbances on forests, Luquillo Experimental Forest, Puerto Rico. At the Luquillo Experimental Forest, scientists are studying how natural and anthropogenic disturbances interact to influence forest characteristics. Economic globalization caused a shift from agriculture to manufacturing in Puerto Rico beginning in the late 1940s. This shift caused people to abandon their farmlands and migrate to urban centers. Farm abandonment, in turn, led to forest recovery in places where agriculture had been marginal. Scientists have reconstructed the land-use histories of different forested areas and found that processes of forest recovery varied, depending on the land-use legacy of a particular area. For example, forests that reestablish on land used for pasture versus coffee plantations have different compositions and structures, which are different from those of primary forests that lack a history of anthropogenic disturbance. Moreover, the effects of anthropogenic disturbance last a long time. These forests are also subject to natural disturbances. Hurricanes are the leading natural disturbance agent in Puerto Rico. Researchers investigated how hurricanes interact with human disturbance to influence forest characteristics. They found that human land-use legacies had a greater influence on forest characteristics than hurricanes, although this influence diminished over time once human disturbance ceased. Forests with a history of human land use were predicted to change to resemble forests lacking a history of anthropogenic disturbance over a 200-year period, regardless of hurricane regime. The research findings indicate that forest dynamics can not be understood without considering how both natural and anthropogenic forces influence ecological processes. science research. The majority of social science studies that have been undertaken have been discrete, terminal studies as opposed to long-term research trajectories. In addition, rarely has comparative work been undertaken between sites. Why have social scientists not taken advantage of the unique opportunities to conduct research at EFRs? Part of the explanation may lie in the history of social science research relating to public lands in the United States. Social scientists concerned with forestry and grazing on public lands have typically focused on communities or individuals as the unit of analysis. Doing so may explain why they have overlooked EFRs as places to conduct research. EFRs that are remote, closed to the public, or receive little use may have no pressing social science research questions to address. Furthermore, studies of the socioeconomic impacts of forest and range management-another area of social science research relating to public lands-typically look at the management of national forests and grasslands at a broad scale. When we asked EFR scientists and administrators why more social science had not occurred, they offered several explanations. A widely shared view pertained to limitations on budgets and staffing. US Forest Service Research and Development has always been dominated by biophysical scientists. Many research laboratories lack social scientists on site. With few social scientists on staff and limited research budgets, biophysical research projects have taken precedence. Biophysical scientists, who have dominated both the research and the administrative aspects of these sites, may also be unaware of what social science has to offer and therefore have not made an effort to encourage social science research at EFRs, although many with whom we spoke were open to working with social scientists. Even at sites where there is a strong interest in incorporating social science, natural scientists sometimes do not know who to engage because of a lack of people having social science expertise nearby or lack of awareness of existing social science networks. Social scientists may also be unaware that EFRs exist and of the research opportunities they present. An inquiry we conducted of US Forest Service R&D social scientists and economists found that only one of those currently on staff had conducted research at an EFR. Finally, biophysical and social scientists alike may be unclear on how to link their work in interdisciplinary research projects. Although there has been emphasis recently on integrated research about coupled human-ecological systems, the practice of working across scientific disciplines can pose challenges. It is not always evident how to integrate social science with the biophysical studies that take place at EFRs. How to overcome these barriers? We believe that highlighting the broad range of human activities and social science research topics associated with EFRs is an important first step to making social and natural scientists alike aware of the untapped potential these sites offer for social science and interdisciplinary research. It is also important to make scientists and site administrators aware of the potential contributions social science research can make to the body of knowledge produced at EFRs, so that they will support and allocate funding for it. These contributions are many. EFRs are places where natural scientists test and experiment with alternative forest and range management techniques related to fuels reduction, timber harvesting, grazing, watershed restoration, and carbon sequestration, to name a few. However, natural resource management goals are socially defined; and implementing effective management approaches to achieving these goals will depend on whether they are economically feasible and socially acceptable . Integrating social and economic components into these kinds of studies can help managers understand what values people want public lands to be managed for, and help research results be successfully applied. EFRs are also places where people currently engage in a variety of human uses, and where people have lived and used natural resources in the past. Forest and range ecosystems have been shaped by their human histories, and they continue to be shaped by people today. EFRs are places where the ongoing dynamics between social and natural processes can be studied to better understand how they interact to influence forest and rangeland conditions, and socioeconomic conditions. Natural scientists may be concerned about the ecological effects of human uses at EFRs and the impacts of these uses on scientific research there. Social science can help document the nature, extent, and motivations behind different forest and range uses and their importance to people. This information can be used to develop solutions to management issues and to promote sustainable human uses of EFRs. It can also generate information on how the public views and values the science mission of EFRs to ensure the relevance of the research conducted there. Social science research can also improve understanding of the changing social environment surrounding EFRs, and the implications of this change for EFRs. This knowledge can help EFRs develop effective education and outreach programs, establish positive collaborative relationships with local communities, engage citizens in their science, address management issues, and garner support for the science mission at their sites. Regarding science integration and working across disciplines, much can be learned from scientists involved in the LTER network. These scientists have discussed how to promote interdisciplinary social-ecological research and how to enhance social science research at LTERs . Suggestions include collaboratively building a conceptual framework for integrated research, focusing research on the interactions between the social and ecological components of ecosystems, developing a multiscale approach to research questions, and identifying tools and models that both natural and social scientists can use. Inviting social scientists to participate in all-scientist meetings at EFRs , as occurs at meetings of LTER scientists, is one way to promote communication, networking, and relationship building that could result in future studies. Social scientists could also be involved in the strategic planning process associated with EFRs in which future science questions are developed. One excellent example, again from the LTER network, is the Maps and Locals project. This comparative research project is being developed across several LTER sites to study patterns of landscape change over time, the anthropogenic versus nonanthropogenic drivers of ecosystem change, and the implications of this change for human livelihoods. --- Conclusions We encourage social and natural scientists alike to use their knowledge of the social science research opportunities at EFRs described here to reach out to one another and promote more social science research at EFRs. We are optimistic that the next 100 years at EFRs will include a substantial social science research component. Several signs point in this direction. The need and desire for social science research at EFRs is recognized by scientists and administrators associated with these sites. Some funding organizations that support EFR research, including the US Forest Service and the National Science Foundation, are increasingly calling for interdisciplinary research that addresses both the biophysical and the human dimensions of forest and rangeland ecosystems and their management. One of the newest EFRs in the network-Heen Latinee, or "River Watcher" in Tlingit -was created in part because it is accessible to Juneau where there is an opportunity for US Forest Service social and natural scientists to work together to develop integrated studies and to involve the community in outreach and education. Social science will thus be incorporated at the outset, providing a model for future EFR sites. One hundred years of research at EFRs has yielded valuable insights about the workings of forest and rangeland ecosystems. We hope that the next 100 years will do the same, with a greater contribution from the social sciences. --- Endnotes

For a century, US Forest Service experimental forests and ranges (EFRs) have been a resource for scientists conducting long-term research relating to forestry and range management. Social science research has been limited, despite the history of occupation and current use of these sites for activities ranging from resource extraction and recreation to public education. This article encourages researchers to take advantage of the rich, though largely untapped, potential EFRs offer for social science by describing their many human dimensions and providing an overview of potential research topics. These topics include human uses, economics, historical studies, population and land-use change, human values, and interdisciplinary social-ecological studies. Lack of awareness among social scientists, limited budgets and networking, and the predominance of biophysical scientists who administer and conduct research at EFRs appear to be inhibiting the development of social science research there. We suggest ways of overcoming these barriers.

Introduction Marriage is a solid inner and outer bond between a man and a woman to create an eternal and happy family in the dunya and the hereafter. 1 Prospective partners must be aware of the law and their responsibilities to achieve the marriage's goals. Therefore, all provisions mandated by religion and the State must be viewed as an attempt to attain happiness in the afterlife. 2 Marriage laws in Islamic law and regulation must be implemented as proof that a person is a Muslim and a good citizen. 3 Marriage registration is one of the provisions that Muslims as citizens frequently ignore because it is not clearly regulated in Islamic law. However, this is governed by Law Number 16 of 2019 concerning Amendments to Law Number 1 of 1974 Concerning Marriage. According to Article 1 paragraph 1, marriage is lawful if it is conducted in accordance with the laws of each faith and belief. According to Article 2 paragraph 2, every marriage is documented in accordance with the applicable laws and regulations. 4 The community views marriage solely as a means to fulfil fiqh requirements, ignoring the administrative interests of both partners as citizens and new families. This has resulted in widespread unrecorded marriage and unrecorded divorce. This is permissible from a religious standpoint, but it violates state regulations. Muslims, as citizens, must comply with state regulations in order to receive the State's intended benefits. Indonesia adheres to European-Continental law, which mandates that all matters be recorded and governed by applicable rules. Particularly within Sukajadi Sub-district, the public's understanding of marriage and divorce registration must be enhanced. This is due to a lack of understanding, so it does not result in the implementation of legislation-required awareness. Unrecorded marriages and divorces will significantly impact spouses and children born of the marriage. These impacts include marriages and divorces that are invalid in the eyes of the law and are, therefore, incapable of protecting the rights and fulfilling the obligations of the husband, wife, and children. When a party violates their rights and responsibilities, the injured party has no legal recourse, and the violating party is not legally required to fulfil these responsibilities. 5 For instance, if the husband does not provide for his wife's maintenance or if the wife does not fulfil her responsibilities, this cannot be prosecuted. In other instances, the fulfilment of rights cannot be pursued, such as when one of the spouses dies, the heirs of both the wife and children born from an unrecorded marriage are not entitled to inheritance property in addition to joint property at the time of divorce. Those whose divorces are not recorded are considered to be legally married to their former spouse. It will be easier for them to remarry if their status as the husband or wife of the other party remains intact. According to the Marriage Law, children born from unregistered marriages only have a blood relationship with their mother. Under state law, they are not regarded as legitimate children. In addition, marriage and divorce should be recorded from the start to avoid the problems listed above, and marriage and divorce will be considered illegal. As previously stated, marriage registration is still an important issue, particularly in Sukajadi sub-district, which is 730 metres above sea level and encompasses five Urban-villages, including Cipedes, Sukagalih, Sukawarna, and Pasteur, the largest village. Geographically, Sukajadi sub-district is bordered by Sukasari and Cidadap Sub-districts to the north, Cicendo to the south, Coblong to the east, and Cimahi City to the west. In addition, Sukajadi Sub-district contains 49 Community Units and 332 Neighbourhood Units per sub-district. Pasteur is once again the administrative region with the most Neighbourhood Units, with 13 Community Units and 92 Neighbourhood Units. 6 In 2018, the total population of Sukajadi Sub-district was 110,356, equivalent to 23,594 families. The Sub-district with the largest population, 32,840, was tallied as Cipedes. This makes Cipedes the most populous village in Sukajadi Sub-district, with a population density of 624.31 inhabitants per hectare. Sukabungah also received the most rice for low-income people with 525 Family Cards out of 2,138 families in Sukajadi Sub-district. 7 It can be concluded that Sukabungah Urban-village, with approximately 25 cases of unregistered marriages and divorces, is the jurisdiction with the most cases of unregistered marriages and divorces, based on the results of field research. Subsequently, the data was confirmed by Sukajadi Religious Affairs Office to indicate that at least 25 couples did not register their marriages between January 1, 2019, and June 30, 2021. This information was only what has been disclosed. This case is like an iceberg; it requires a more profound and broader investigation to uncover the truth. 8 In addition to the community's lack of legal awareness, the registration of marriages and divorces must be higher. This impacts the administrative issues faced by couples and children born of unrecorded marriages and divorces. 9 In the community of Sukajadi Sub-District, which contains 332 Neighbourhood Units, unrecorded marriages and divorces are still rife. In Sukabungah Urban-village, Cipedes, for instance, every child who enrols in school has trouble registering because they lack a birth certificate to prove they are a legitimate child. Regarding inheritance rights, divorced widows and widowers cannot be determined as heirs because their marriage was never registered. Similarly, a widow or widower of a living divorce has difficulty registering a subsequent marriage due to the unrecorded divorce. 10 Field research conducted in Sukajadi Sub-district revealed at least three undocumented marriages and divorces. Imagine the effect on legal actors if the information was collected from every Urban-village. Researchers are interested in studying the phenomenon of unrecorded marriages and divorces in Sukajadi Sub-district, as indicated by the preceding description. This study employed a library research whose data sources were books, scientific papers, and other documents categorised in the literature, with an empirical normative approach, namely research that utilised case studies in certain legal regions in the form of legal behaviour products. 11 This research began with written legal provisions applied to actual societal events. 12 --- Legal Consequences of Unrecorded Marriages and Divorces Currently, under-armed marriages are still prevalent; this is due to several factors, one of which is that the community understands the law but disregards it, and some people are very unfamiliar with the law; this affects various facets of life, including marriage and divorce. The legal consequences of unrecorded marriages and divorces based on interviews with stakeholders in Sukajadi District area conducted on March 20, 2021, at 11.00, and the results of these interviews are: Inquirying about the existence of people whose marriages and divorces were not recorded; initially, the head of 11 Community Unit stated that all marriages of 11 Community Unit residents were registered at the Religious Affairs Office and were in compliance with applicable regulations; however, b) the head of 11 Community Unit subsequently The chairman of 11 Community Unit was surprised to learn of the residents' confession because communication was not going well. The researchers subsequently inquired as to what efforts the head of 11 Community Unit would make to resolve the issue; thus, the head of 11 Community Unit urged his community to complete the administration of marriage and divorce at the Religious Court immediately. 13 The interviews indicated that the community's and local officials' legal knowledge in Sukajadi Subdistrict must be improved, particularly in marriage and divorce. In addition to interviewing the head of 11 Community Unit, the interview was also conducted with the head of Sukajadi Sub-district. According to the results of the interviews, the community's reluctance to undergo the procedure of recording marriages and divorces due to financial constraints was the most significant obstacle. Thus, this became an issue in the administration of the relevant documents, such as the absence of a marriage certificate; it is challenging to create a birth certificate for a child; even if a birth certificate is created, it cannot contain the father's name. 14 10 Farid,D.,Muhammad Husni Abdulah Pakarti,Mohamad Hilal Nu'man,Hendriana Hendriana,Iffah Fathiah. -Talak Perspektif Kesetaraan Gender: Perintah Tuhan Menerapkan Egaliter Di Dalam Rumah Tangga.‖ Tahkim, Jurnal Peradaban Dan Hukum Islam 6, no. 1 : 1-18. https://doi.org/10.29313/tahkim.v6i1.10849. 11 Soerjono Soekanto, Pengantar Penelitian Hukum, ., 10. 12 Abdulkadir Muhammad, Hukum dan Penelitian Hukum, , 52. 13 Wawancara dengan H. Taufik Joesman pada tanggal 20 Maret 2021 Pukul 11.00 14 Wawancara dengan Drs. Inci Dermaga Mustawan A,., M.A.P, Kepala Kecamatan Sukajadi, tanggal 25 Maret 2021 Pukul 11.00 Mr Suhada, the Regional Leader of Muhammadiyah, was interviewed regarding marriage and divorce registration in the context of orderly legal administration. Mr Suhada stated that people must realise that Indonesia's legal situation and culture differ significantly from the past to the present. Under Law No. 1 of 1974 about marriage, the administration of related matters must be recorded with the state administration, as it is directly related to the needs of children, inheritance, and the rights of husband and wife. 15 In addition, the legal consequences of unrecorded marriages and divorces include the inability to apply for a birth certificate with the father's name. In addition, the Education Office does not accept birth certificates without the father's name. When a divorce occurs, it cannot be processed by the Religious Court, so the couple cannot obtain a divorce certificate. Women do not have autonomy due to unrecorded marriages; for instance, they cannot present a marriage certificate if they want to borrow money and mention their marital status. Similarly, it is challenging to create a birth certificate; she cannot manage her own or her husband's pension, report domestic violence, or manage an inheritance. In addition to issues related to women's autonomy, couples, particularly women, experience additional consequences. Unregistered marriages are stigmatised as "to live together without being married", and this stigma is undoubtedly unsettling for both partners. 16 Article 2 of government regulation number 9 of 1975 concerning the implementation of Law No. 1 of 1974 concerning marriage authorises two institutions to record marriages: the Religious Affairs Office for Muslim couples and the Population and Civil Registry office for non-Muslim couples. Unrecorded marriages impact women because a woman's status as a wife is not recognised by the State, which impacts the recognition of her children as their father's children. 1. Not being able to apply for a birth certificate with the father's name on it. 2. Not being able to apply for a divorce and obtain a divorce certificate from the religious court. 3. Women do not have autonomy; for example, they cannot show their marriage certificate when borrowing money and mentioning their marital status. 4. Unable to apply for a pension for herself or her husband's pension, 5. Cannot report when she has experienced a domestic violence case. 6. Not being able to apply for a passport or visa for Umrah and not being able to use it for inheritance. 7. Being stigmatised as a couple living together without marriage or often called " to live together without being married". --- Factors Affecting Unrecorded Marriages and Divorces in Sukajadi Sub-district Community Religious leaders generally view marriage registration as unnecessary, as no Ulama has ever mandated its requirement. Some religious leaders, on the other hand, view marriage registration as a positive endeavour. In addition, they lament the fact that religious figures are 15 involved in unregistered marriages. There are unrecorded marriages for religious, legal, sociocultural, and economic reasons. 1. Religion a. The view that registration of marriages is not mandated by religion; b. The view that religion does not require the permission of the first wife for polygamy; c. The view that religion permits the practice of mut'ah marriage; d. The belief is that if a state regulation conflicts with religious teachings, the religious teachings should prevail. Likewise, government-determined marriage registration is not considered to be commanded by religion. e. The belief that widows could marry without the permission of their guardians resulted in many widows secretly marrying outside of the State-required marriage procedures. f. Efforts to criminalise individuals other than marriage officers criminalise Sharia because marriage is part of Sharia guidance. g. Existence of religious figures as the primary actors who marry unregistered couples. --- Law Better public legal awareness so that every member of society knows, understands, realises, and carries out their rights and responsibilities as citizens and citizens of society and respects human rights daily. Community legal awareness must begin in the home with young children. In reality, however, particularly in Sukajadi Subdistrict community, some people do not know the marriage and divorce registration rules, others know the rules but disregard them, and others do not know the registration laws; however, follow them.17 Many married couples, particularly those who have never interacted with the administration, view their unregistered marriages as equivalent to other marriages due to lax law enforcement. --- Socio-cultural Insufficient Age Underage marriage is an additional contributor to unrecorded unions. According to Kustini, marriages between minors are still prevalent in the community. This is demonstrated in the underage marriage perpetrators who petition the local religious court for a marriage dispensation. Parents assume that a girl's marriage will reduce the family's economic burden because her husband will now be responsible for her. --- Polygamy The difficulty of polygamous rules for second, third, and subsequent marriages stems from the fact that the conditions for polygamy outlined in the marriage law, namely obtaining permission and consent from the previous wife, must be satisfied. In addition, because permission from the wife is difficult to obtain, the husband performs a religious marriage ceremony; however, unregistered known as Siri marriage. --- Economy The primary factor influencing the non-recording of marriages and divorces is economic difficulties. This is because, in the past, the community believed that registering a marriage required much money and complicated procedures. Moreover, modern society would benefit from a custom requiring the bridegroom to pay the dowry and the cost of the wedding feast, thereby requiring people to conduct halal marriages. Initially, the community viewed the recording of marriages and divorces as unimportant, but when faced with administrative requirements, the community realised the significance of recording. By analysing some of the data in the table above, it can be concluded that unrecorded marriages continue to occur in Sukajadi Sub-district. One of the reasons is the need for a greater understanding of marriage and divorce registration laws. --- Legal Handling Efforts for Unrecorded Marriages and Divorces --- Counselling Legal counselling activities improve society so that each community member understands and abides by his or her rights and responsibilities as a citizen and realises a legal culture in a manner that is conscious, obedient, law-abiding, and respectful of human rights. Legal counselling activities include disseminating information and understanding the applicable legal norms, laws, and regulations. With legal counselling, it is hoped that the community will become aware of all applicable laws and regulations. Once aware, the community will comprehend the material and content contained in laws and regulations, motivating the community to respect and obey all applicable laws and regulations. 19 The community is educated on the significance of marriage registration and the necessity of registration for achieving a sakinah family. The following steps are taken when conducting counselling: Coordinating with local authorities, providing qualified resource individuals, and inviting the community. 20 In addition, there are unique regulations regarding legal counsel in Indonesia. Furthermore, legal advice is part of the legal fiction principle. According to the Regulation of the Minister of Law and Human Rights No. M.01-PR.08.10 of 2006 Concerning Legal Counselling Patterns, legal counselling is one of the activities of disseminating information and understanding applicable legal norms and laws. The objective is to realise and cultivate public legal awareness to establish a legal culture compliant with legal norms and legislation. 21 According to Muhammad Yunus Affan, director of the National Law Development Agency , Legal Counselling and Assistance Centre, the legal counselling mechanism has evolved. The public may directly visit BPHN, the Regional Office of Law and Human Rights in their place, or use the provided online system to submit questions. Additionally, BPHN utilises additional communication channels. "Our online legal counsel is provided through legal-savvy channels with legal consultation capabilities," he explained. Legal aid consultations, lectures and discussions, legal awareness meetings, simulations, socialisation, and legal awareness competition exhibitions are examples of direct legal aid-19 https://lampung.kemenkumham.go.id di akses pada tanggal 22 Oktober pukul 22.00 WIB. 20 Farid,D.,Muhammad Husni Abdulah Pakarti,Usep Saepullah, --- Consultation Legal consultation is a legal service consisting of advice, explanation, information, or direction provided to people with legal issues to assist them in resolving those issues per applicable laws and regulations. People who are eligible for legal consultation have legal problems. The phases of the consultation activity are as follows: a) A person litigating or has a problem that must be resolved legally through a religious court has made a request. b) The principal will explain the entire problem during the consultation. c) After understanding the issue, the mediator must provide a response and legal justification for his opinion on behalf of the principal. --- Mediation Article 2, paragraph 2 of Supreme Court Regulation No. 1 of 2008 stipulates that the mediator judge and parties must follow the dispute resolution procedure through mediation before proceeding with the main case procedure. Since this process is also an integral part of the court, the judge must follow the settlement procedure through mediation; if the judge does not follow the procedure, the judge's decision may be declared null and void. 1. Handling of Legal Aid to the Court and Religious Affairs Office Based on the issues that have been previously mentioned. The economy and the cost of processing cases primarily influence community disobedience in marriage and divorce registration administration. Case-handling assistance for religious courts is the actual effort that the community can directly perceive and receive. Each religious court has a prodeo quota for those who wish to resolve civil cases within the religious court's jurisdiction but are financially constrained. In addition to the courts that provide special quotas for resolving these cases, there are Legal Aid Institutions that have been accredited by submitting case-related documents and a certificate of poverty, in collaboration with the Ministry of Law and Human Rights, in order to assist those who are constrained by economic issues. Legal assistance is required even during the judicial process to assist the community in resolving their legal issues because not all levels of society understand the litigation procedure. An advocate is a party who can provide legal assistance until the conclusion of the judicial process. Law No. 16/2011 on Legal Aid emphasises that everyone has the right to legal aid, ensuring that only people experiencing poverty are eligible for free legal aid. Access to legal assistance is no longer restricted to the wealthy. --- 2. Marriage determination Marriage determination is a legal remedy for the provisions governing Muslim marriages in Indonesia in addition to the provisions on legislation that require marriage registration so that the marriage will receive an official marriage certificate; there are additional provisions governing Marriage determination, namely; is a legal effort that can be made to obtain state recognition for unrecorded marriages by submitting a marriage determination. 22 Article 7, paragraph of the Compilation of Islamic Law states: "If a marriage certificate cannot prove a marriage, an Marriage determination may be submitted to the Religious Court," and paragraph states: "The Marriage determination that may be submitted to the Religious Court is limited to matters about the marriage." The court's jurisdiction is limited to the following: a) Marriage status in the context of marriage settlement; b) The loss of a marriage licence; c) Existence of doubts regarding the validity of a marriage. d) The existence of a marriage performed before the enactment of Law No. 1 of 1974; and e) Marriages performed by those who do not have a marriage impediment according to Law number 1 of 1974. Article 4 of the Compilation of Islamic Law states, "Husband or wife, their children, marriage guardians, and interested parties may apply for itsbat nikah." The provision that allows Marriage determination, as stipulated in Article 7 of the Compilation of Islamic Law, indicates that the regulations in force in Indonesia recognise the validity of an unrecorded marriage and, for the reasons listed in the details of paragraph of the Compilation of Islamic Law, the marriage can be recorded and be determined; administratively, this situation recognises the marriage. 23 Article 7 of the Compilation of Islamic Law has also provided opportunities for unrecorded marriages to register themselves correctly; this condition is also considered to benefit the perpetrators of unrecorded marriages; 24 in addition, it is the responsibility of the authorised institutions to provide opportunities for people who want to realise that their marriage is legally recognised so that it has superior legal force. The Marriage determination procedure is the existence and independence of the Religious Courts since the birth of Law Number 3 of 2006 concerning amendments to Law Number 7 of 1989 concerning Religious Courts, where the Religious Courts have the same position as other courts and one roof under the Supreme Court; consequently, the Religious Courts are required to implement the law as effectively as possible as an institution that serves the public in seeking justice. Marriage determination is a case in which there is neither a dispute nor a volunteer, and the Religious Courts have the authority to handle it if the law requires it. In general, the court does not actively seek out cases. Article 7 of the Compilation of Islamic Law has granted the Religious Court broad, absolute jurisdiction over the Marriage determination case. Marriage determination is a petition to legalise and recognise a previously performed marriage contract without authentic proof of marriage. The procedures that the applicant must complete are as follows: a. As applicants, the husband or wife, widow or widower, children, marriage guardian, and other parties interested in the marriage process file a written petition with the court;25 b. The application is submitted to the Religious Court in the applicant's place of residence, and it must include the following information: a) The party's identity ; b) Posita or the reasons/arguments supporting the petition; and c) Petition or the matter for which a court decision is sought. Article 39, subsection of Minister of Religious Affairs Regulation No. 3 of 1975 contains the provision regarding Marriage determination. This provision only applies to marriages before the passage of Law Number 1 of 1974 and not to marriages after the law's enactment. This provision permits the Religious Courts to exercise their jurisdiction in this matter for reasons other than Marriage determination that occurred before the enactment of Law No. 1 of 1974, based on the individual interpretation and judgment, to provide applicants with a sense of justice. Often, the submission of Marriage determination is accompanied by the submission of the legitimacy of children born from unrecorded marriages to obtain the same validation as Marriage determination; this is due to the emergence of phenomena in society if children born cannot be recognised by civil records as children born from legal marriages, so careful consideration is required because many phenomena in some areas of Indonesia do not accept if their children are declared baptised. There is only a civil relationship with the mother, and the father's name is not mentioned. 26 Because it is common for a man to deny a child conceived by a wife, as explained in Article 44 of Marriage Law Number 1 of 1974 and Article 102 of the Compilation of Islamic Law, the submission of children included in the Marriage determination must be authentically proven by the person concerned. Paragraph of the law authorises the court to decide whether the child is legitimate at the request of an interested party. 27 The court requires a procedure for determining whether a child is legitimate under existing law's applicable procedures and rules. Article 55 of Marriage Law Number 1 of 1974 states: a. The only way to prove the origin of a child is with an authentic birth certificate issued by an authorised official. b. If the birth certificate referred to in paragraph of this article does not exist, the court may determine the origin of a child after a careful examination of qualified evidence; and c. Under the court's order specified in paragraph of this article, the birth registration agency within the court's jurisdiction shall issue a birth certificate for the child in question. One of the supporting pieces of evidence is that the child is only in the custody of a woman who knows whose seed she is carrying. However, this condition will lead to discrimination and put women in a difficult position because men can deny it. The problematic nature of the woman's proof causes women to despair. Demanding the husband's responsibility is another way to prove that the child is a legitimate offspring of an unrecorded marriage; this can be done through legal means as outlined in the Law on Child Protection Number 23 of 2002 article 13: "Every child while in the care of parents, guardians, or any other party responsible for care is entitled to protection from: a. Discrimination; b. Economic or sexual exploitation; c. Neglect; d. Cruelty." Cruelty, violence, and other forms of abuse, e. Injustice; and f. Other forms of mistreatment. Suppose parents, guardians, or carers of children administer all of the treatments mentioned above. In such cases, the perpetrators face a maximum sentence of 5 years in prison and/or a maximum fine of IDR 100,000,000. Kustini continued that if it is not recorded, the marriage has no legal standing in the eyes of the State. Legally, the wife is not considered a legal wife; therefore, she has no right to maintenance or inheritance in the event of divorce or death. --- Conclution Many still need to comprehend the law governing marriage registration and unrecorded divorces, which impedes administration processes such as birth certificates, joint property, and inheritance. In addition to issues related to women's autonomy, couples, particularly women, continue to experience additional consequences. Couples with unregistered marriages face the stigma of living together without marriage, also known as "living together without being married." This stigma is certainly not reassuring. The non-registration of marriages and divorces in Sukajadi Sub-district is influenced by financial and legal awareness, insufficient age, and polygamy factors. Litigants are assisted with legal matters through legal counseling, consultation, mediation, and handling legal aid for the court and the Religious Affairs Office. The suggestions for the government or other stakeholders on how these problems can be resolved include conducting legal counseling so that the community can gain insight into marriage and divorce; thus, the community understands that living in a legal state means obeying the rules of the state not only about religion but also the rules of the state; subsequently, what has happened immediately facilitates and provides legal assistance to the court in order to resolve the problem; and finally, break up the practice of unregistered marriages and divorces from an early age and do not transmit it to our descendants.

Differences in marriage and divorce law in Islamic and state law one of the triggers for the initial causes of the disorderly implementation of marriage and divorce in the Sukajadi Bandung community was an understanding regarding the separation of religious and state law as well as social welfare factors related to law enforcement. This study aims to analyze the legal consequences, factors, and legal remedies that can be taken in unregistered marriages and divorces. The method used in this study is empirical-juridical with a qualitative approach. This approach is carried out to obtain information that occurs in the Sukajadi Bandung community regarding unregistered marriages and divorces. The results of the study show that the factors behind the rampant practice of unregistered marriages and divorces are not recorded because people's understandings of the dichotomy between Islamic law and that of the state are different. There is a link between the legal awareness of the community and its understanding of religion and welfare. In addition, unregistered marriages and divorces have an impact on the absence of marriage, inheritance, custody, and maintenance, which are detrimental to children and wives. Several efforts need to be made to minimize the occurrence of unrecorded marriages and divorces, including counseling, legal assistance for divorce victims, mediation, and consultation. This effort does not only involve the government and law enforcers but also preachers, lektors, and social organizations.

Introduction In their review of the connections between ecotourism and conservation, Stronza et al. identify a number of research elements that are frequently missing; sometimes these are conducted independently, but it is necessary to conduct them together for rigorous evaluation. These elements include: gathering longitudinal data , addressing issues of scale , studying community outcomes beyond economic impacts , participatory evaluation , and addressing the larger social context driving land-use change and deforestation . Special issues on systems and resilience approaches to protected area management and nature-based tourism and the articles therein have advocated for the further development of social-ecological systems and resilience frameworks, and for research that explicitly considers hierarchical dynamics and feedback loops and incorporates Morse . /frsut. . analysis that considers protected areas and surrounding landscapes where tourism and conservation occur. This article builds on these frameworks and links the bodies of literature on tourism, protected areas, and landscape change through a Social-Ecological Complex Adaptive Systems framework. The SECAS framework was originally developed to enable an interdisciplinary team to assess the social and ecological impacts of Costa Rica's Payments for Ecosystem Services program ). Ecosystem services are the benefits that people receive from ecosystems, including production , regulation , and cultural services . In 1996, Costa Rica passed a Forestry Law that prohibited converting natural forests to other land uses and established one of the first programs that paid landowners directly for providing several environmental services, including watershed protection, biodiversity conservation, carbon sequestration, and aesthetic values . Costa Rica targeted the PES program toward a system of biological corridors that linked national parks and other conservation areas. These corridors generally consisted of areas with high forest cover and agricultural land use that were privately owned but located in poorly developed areas of the country. The PES program was designed to enhance conservation and improve local household and community livelihoods in the regions outside of protected areas. Our team research was conducted in the San Juan-La Selva Biological Corridor in northern Costa Rica, where some of the highest concentrations of private forests mixed with agricultural lands connect the highlands of the central volcanic range, including Braulio Carrillo National Park, Volcan Poas National Park, Juan Castro Blanco National Park, and several forest reserves through lowland areas to the Indio Maiz Biological Reserve in Nicaragua along the San Juan River . A framework was needed to organize our project, which examined how a social conservation policy could influence landowners' decisions on land use , which would then change the land cover across the landscape over time to have an impact on the desired ecosystem services . The framework was clearly required to incorporate social and ecological system factors and hierarchical multi-scale considerations that changed over time. We needed a SECAS framework. The initial development of the complex adaptive system concept came from ecology . CASs are characterized as dynamic, unpredictable, non-linear, multi-scale systems with multiple interacting components, and a lack of central control . A CAS is defined by the presence of a network of interactions and relationships among the multiple components . CASs adapt over time through recursive interactions and feedback between components, and between components and their environment, leading to emergent or novel patterns . CASs are open systems, and dynamic interactions occur across multiple scales, allowing them to self-organize, often into nested hierarchies . CASs are considered to be non-linear, meaning that cause and effect are not always proportional, and small changes can lead to bigger impacts on other components or on the whole system . Interactions can take the form of slow or fast variables and can occur across spatial scales . Non-linearity leads to complexity, unpredictability, and uncertainty within and about the system . The term adaptive indicates that a CAS can change, evolve, and self-organize over time in response to feedback . Similar to ecological systems, social systems have multiple interacting components across multiple scales, are dynamic, and change over time . SESs are considered to be inextricably linked, and together, these systems are considered to be CASs . The concept of resilience is a way to frame SECAS that explicitly recognizes uncertainty, complexity, and change . Resilience has been defined as the capacity of a system to absorb disturbance and maintain the same identity or the same function, structure, and feedbacks . Resilience also describes the degree to which a system can self-organize and its ability to build its capacity to adapt or learn . Resilience has become a goal in managing CASs . Managing protected areas and tourism as complex adaptive systems Social-ecological systems, complex adaptive systems, and resilience have been promoted as frameworks for research on and management of protected areas and for tourism based in protected areas . To address biodiversity conservation and protected area management, Cumming et al. proposed a framework to capture the multi-scale SESs that extend beyond the boundaries of protected areas into the "functional landscapes" necessary for conservation and support of the protected area. The authors build on Ostrom SES framework and address some of the concerns for application by adding five hierarchical levels and highlighting temporal dynamics and cross-scale interactions . Research from an ecosystem conservation perspective expands the interests in protected area management beyond the administrative boundaries of the area into humandominated landscapes, as linked SESs focus on cross-scale feedback , ecological solidarity , and resilience . In a seminal article reconceptualizing theoretical frameworks in tourism, Farrell and Twining-Ward specifically identify the need to fully consider SESs and frame research around the process, transition, or journey of dynamic complex adaptive systems. The authors draw parallels from CASs in ecology with tourism systems, introduce the concept of resilience, and develop their own Complex Adaptive Tourism Systems model to address tourism systems more comprehensively . Strickland-Munro et al. also assesses protected area tourism and local community interactions as multi-scale embedded CASs with two case studies in national parks in South Africa and Australia. Following others, the author emphasizes the importance of resilience thinking in understanding continually adapting tourism systems . Strickland-Munro et al. develops a four-step model for research that includes system definition, past system change, current system state, and monitoring of change. Lew and Lew et al. emphasize the importance of spatial scale and of an understanding of fast and slow variables; they also emphasize how a resilience perspective will help in placing focus on adaptive management within ever-changing tourism CASs. McCool et al. ) and McCool and Bosak argue that framing protected area management and tourism research from a systems perspective is essential in order to counter past reductionist perspectives and provide managers with meaningful leverage points to target resilience-building in these systems. These articles also discuss the difficulties involved and the need to work with the public and use systems frameworks to make sense of dynamic and complex contexts , address the challenges of systems work , and identify bridges and barriers to conducting interdisciplinary research . McCool et al. provide a set of six "complexity practices" to help frame CASs and manage them toward resilience, namely, building situational awareness, investing in personal relationships, appreciating the power of networks, identifying and using leverage points, employing different forms of knowledge, and learning continuously. --- The social-ecological complex adaptive systems framework The Social-Ecological Complex Adaptive Systems framework was designed based on the fundamental principles of the CAS framework . It was designed to be multi-scale and to integrate across dynamic and non-linear social and ecological systems, with inputs and outcomes across scales and systems . Visually and conceptually, the framework was based on research by Grimm et al. on change in land use and land cover, and on research by Ostrom Theoretically, our research group used structuration theory from the social sciences to explain social CASs , because humans can and do act with foresight and intent, meaning that social and ecological systems are fundamentally different in terms of the drivers of self-organization . Structuration theory had been identified by others as suitable for linking social and ecological systems , and we elaborated on and updated their contributions to include revisions to structuration theory made by Stones . "A defining characteristic of structuration theory is that through recursive social practice or action, social systems influence the activity of individuals, who in turn, produce, transform, or otherwise reaffirm those same structures constantly producing and reproducing society" . We retain the descriptors "social" and "ecological" in front of "CAS" in order to highlight the differences in terms of drivers of self-organization. On the ecological side of the SECAS framework, we applied the theory of hierarchical patch dynamics , where each patch is nested in a dynamic patch mosaic , which is again nested in a higher-level patch mosaic . A base SECAS model that demonstrates the linking of social and ecological systems across scales is presented in Figure 1. The left-hand side of the model represents hierarchically nested social systems, and the right-hand side represents ecological systems in terms of nested patch mosaics. The top of the model illustrates the inputs to an action, and the bottom half represents the outcomes of that action. Actions are modeled as having outcomes that impact both systems and all levels simultaneously, as each is a nested part of the other. In the CAS framework, our knowledge of external social and ecological systems is seen as incomplete, and the outcomes of our actions may be intended or unintended . Since the inception of the SECAS framework , I have collaborated with others to place existing recreation models into a systems perspective and to integrate them with a cultural recreation ecosystem services perspective . McCool et al. recognize that many of the tools used to manage outdoor recreation are linear and reductionist and do not take a systems approach. The SECAS model has been applied to outline how a number of these recreation tools and constructs, such as the recreation experience model, beneficial outcomes, the recreation opportunity spectrum, limits of acceptable change, and constraints theory, could all be framed together into a unified systems perspective . A second application of the SECAS model to recreation is in examining how the field of outdoor recreation research and the concept of recreation ecosystem services could be better integrated . This work has further integrated components of recreation management into the SECAS framework, extended the framework to consider outdoor recreation and the corresponding tools and theory as they apply to nature tourism, and added protected area and protected area management as a third dimension. Furthermore, the article presents the idea of transformation at the center of the recreation experience to highlight the experiential and dynamic nature of outdoor recreation and nature tourism . While this last application of the SECAS framework does address protected areas and their management, it still considers the entirety of the tourism system in individual boxes on the social side of the model. The current article conceptualizes the tourism system in accordance with the literature on tourism systems and protected area systems, and integrates this with a meta analysis of the drivers of land-use and land-cover change to further frame the ways in which the landscape changes around a protected area with tourism. --- SECAS for protected area management and nature tourism Once the general model is understood, it must be populated with variables that are important to the relevant research questions across scales and systems. If i want to understand the interactions between tourism, conservation, protected area management, and the environment as a SECAS, i need to understand the drivers of agriculture and forest management in the functional landscapes outside of protected areas, how the tourism system impacts local communities and protected areas, and even how the tourist navigates the system through components of the traditional tourism industry. I began by identifying and consolidating the major subsystems identified in the literature on land use and land cover outside protected areas , items mentioned as critical for ecotourism as a form of tourism closely associated with protected areas , and items mentioned in the protected area and tourism CAS literature that was reviewed. The major change to the SECAS model is to move beyond generic two-dimensional representations of social and ecological systems and identify the many other social systems that are important for conservation and tourism around a protected area. I identified 12 major component categories of social systems from the literature ; these are presented in Figure 2. Each social subsystem could be modeled as a nested hierarchy with inputs and outcomes, as in the current SECAS framework . With all the subsystems included together, the model would be visualized as a sphere with a funnel or hourglass through the middle. For example, park management is its own hierarchically nested social system, from the management of an individual setting , to an individual park, to the park system across a country, to its implications at the global level . Governance systems are frequently hierarchically nested. Similarly, tourism accommodations are a hierarchically nested social system with different types and amounts offered at different scales. Ecological systems could similarly be expanded to address watersheds, habitats, and biodiversity as hierarchically nested systems. The side-by-side stepped framework captures the dynamic system with inputs, outcomes, and feedback pathways in two dimensions, while the 12-piece pie chart shows all the different subsystems and how they come together across scales. This view from the top can be imagined as an open hourglass, seen from above: the center is where all the different variables come together to form a tourism experience and where the sand flows down to the next level to produce outcomes for all the different systems. Feedback loops refill the top half with sand, enabling the process to continue recursively, as tourism, park management, and conservation are part of a continually updated SECAS . --- How to study the SECAS Where in the system, or what scale, you want to focus your analysis is dependent on the research question at issue. HPD has a multi-scale analysis protocol of "enveloping, " while structuration theory has "methodological bracketing." Both approaches indicate that multiple levels of analysis are needed to understand a CAS, including the external environment, which provides the conditions for any action/disturbance, and the mechanism that describes how and why things happen at a lower level. Stones developed methods for analysis of actors' conduct and for context analysis from Giddens's methodological brackets, and these approaches help in representing the steps for analysis that we outline below. These steps address items from the four-step model of Strickland-Munro et al. and the six "complexity practices" proposed by McCool et al. . These steps extend these previous models by adding temporal analysis , a purposefully scaled analysis, and multiple viewpoints. The steps also address each of the five components that were identified as lacking in rigorous studies on tourism and conservation systems by Stronza et al. . The steps can be used for both social and ecological systems analysis. --- Step . Context analysis The context analysis is designed to examine enabling and constraining conditions of the external context for actions . This step helps to define the system. Context analysis should be derived from both the researchers' perspectives and the actors' perspectives . --- Past system change: the researcher's historical perspective To understand how systems change and the influence of feedback over time, a more historical perspective is needed. Examination of the "intermediate temporality" would allow reflection on how social systems enabled or constrained or reacted to different actors' actions . This can be done through literature reviews, policy analysis, and other external analyses. Similar historical analysis can be done for land use change, biodiversity trends, and other ecological assessments. --- Building situational awareness It is also important to obtain multiple perspectives of the current situation at the systems level. For example, interviews, focus groups, and group mapping exercises with government agricultural agencies, non-governmental conservation organizations, and protected area managers can provide new insight as to the specific social systems variables that are influencing the system . This level of analysis helps to build personal relationships and understand power relations and social networks . Parallel analyses with many of these same groups can explore environmental issues in that local context identifying underlying and proximate drivers, feedback, and change. --- The actors' perspective It is critical to conduct interviews, surveys, and discussions with individual actors or local groups about what they believe are the external enabling and constraining conditions that are influencing their decisions . This is the perspective of the actor looking out at the system. For example, Lupoli et al. developed a rapid assessment tool to capture local community desires regarding volunteer tourism in their community. This process will also help to build personnel relationships, assess the network of actors, provide an understanding of the relevant power relationships and social norms, and begin to identify leverage points . An understanding of local or traditional ecological knowledge for both farm and landscape management and conservation can be obtained in this phase. Step . Conduct analysis: the actors' perspectives on why they take action A conduct analysis is an assessment of the knowledgeability, motivations, capabilities, and desires of the actors themselves and how these are translated into action . The environmental parallel here is the ecological beliefs and perceptions of ecological outcomes that lead to actors' actions. A protected area manager will be able to explain why they took particular land management actions, and a landowner will likely be able to present the environmental benefits and costs that their actions might cause. This is a critical step in identifying leverage points, or those places in a system where intervention might have the greatest impact toward the development of more resilient systems . --- Step . Monitoring change and learning continuously Understanding the system, developing networks, and identifying leverage points will help with identifying and understanding the relationships among the variables in the system, but understanding the system is only the beginning. A SECAS is characterized by feedback, non-linearity, and uncertainty, indicating that any single assessment at any given time will only provide part of the picture. Adaptive management is a form of continuous learning that can inform management toward the building of more resilience in the target systems, and the steps described above will inform managers of the networks involved and develop the linkages and networks that can facilitate communication and interventions . In the same way that one must look back to understand context, long-term monitoring and an adaptive management approach are needed. --- Conclusion Case studies often "[treat] tourism as a separate enclave from its larger social and environmental system, which is anathema to the complex systems approach of resilience" . To examine protected area management, landscape conservation, and tourism, we need a framework that can capture the entirety of these dynamic and evolving systems, including inputs, actions, outcomes, and feedback. This article has presented the SECAS framework as an organizational concept that can help in framing the multiple systems and subsystems that can drive change and resilience. The major components of the systems of protected area tourism and conservation have been highlighted, along with steps that can help in identification of the specific elements in the systems that can be used to leverage resilience. I hope that the SECAS framework and this article can be used as a springboard for applied analysis and a baseline for further theoretical development. Bebbington, A. --- --- --- Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

) Protected area tourism and management as a social-ecological complex adaptive system.

INTRODUCTION By the 20s of the 20th century, the life of the country's rural population was in a poor state. The severe complications of the First World War, and the policy of brutality and violence that the Bolshevik rulers used against the local population during the establishment of the new Soviet system were visible at every step. Industry has been destroyed, transport and communication facilities are in operation, and countless fields of cultivation have been laid waste. Worst of all, hunger, poverty, and mass unemployment reigned in the country. On top of that, the food shortage developed by the regime of Soviet power and the policy of military communism, which was combined with it, made the life of the people of the country even more difficult. In order to preserve the Soviet power, this emergency policy took everything from the population, strictly limited its will, initiative, and independence, did not allow trade and market relations, and as a result, the socio-political and economic situation here became even more complicated [1]. --- THE MAIN RESULTS AND FINDINGS In the national territorial demarcation carried out in Uzbekistan, in the historical and ethnographic process of the Uzbek people, it began to form its own local character. This has had an impact on the social, economic and political life of the Uzbek people. As a result of economic and political reforms, the class structure and demography of Uzbekistan changed radically by the end of the 1930s. In the 1939 population census, there were 6.2 million people in the Republic. The causes and roots of the crisis situation in the political and socio-economic life of his society lay in the backwardness of his thinking and the inability to adequately assess the specific historical conditions. The concept of "advanced socialism", which entered scientific and political life at the end of the 60s, was the result of a serious theoretical error in determining the level of socio-economic development. By the 70s and 80s, this policy did not justify itself. As a result, the process of disintegration of the USSR accelerated. The economic and social life is becoming more and more evident, and the need to take a new path to eliminate crisis situations and bring Uzbekistan to a higher level of development has been fulfilled. Currently, there are 563,300 families in Kashkadarya region. In connection with the "Year of Social Protection", 9.5 billion soums of financial aid was provided to 114,600 families in the region this year. In order to support more than a thousand young families, 2 billion 134.1 million soums of preferential loans were granted by commercial banks [7]. In In their activities, cultural institutions also held various non-traditional activities that did not correspond to the mentality of the local people. For example, in a series of events held under the heading of "new life", the European way of life was promoted, and Uzbek traditions, family relations and religious ceremonies were promoted as "superstitions". The "New Life" column, aimed at "liberating" women and forming new relationships, naturally caused a number of misunderstandings among the local people in the matter of family relations formed over the centuries. --- CONCLUSION Thus, it can be noted that in the 20th century, fundamental changes took place in the social, economic, and political processes of the Uzbek people.

This article abstract explores the social, economic, and cultural processes that took place in the Kashkadarya oasis during the 20th century and their influence on the lifestyle of the local residents. The Kashkadarya oasis, located in Uzbekistan, has been a historically significant region with a rich cultural heritage. Through an analysis of historical sources, archival materials, and ethnographic studies, this abstract sheds light on the major transformations that occurred in the oasis throughout the 20th century. The article examines the impact of various factors on the lifestyle of the local residents. It delves into the social changes brought about by industrialization, urbanization, and the introduction of modern education systems. These changes resulted in shifts in the traditional social structure, family dynamics, and gender roles within the community. Furthermore, the economic transformations, such as the transition from agrarian to industrial economies, affected the livelihoods and occupations of the local population. The abstract also highlights the cultural shifts that unfolded during this period. It investigates the influence of Soviet policies and ideologies on the cultural practices, beliefs, and traditions of the oasis dwellers. Additionally, it explores the interplay between modernization efforts and the preservation of local cultural heritage, including traditional crafts, music, and folklore.

Moos & Moos, 2007) where recovering peers typically develop social networks based in support for abstinence . Abstinence self-efficacy is regarded as an important resource for relapse prevention that has been found to predict abstinence . Abstinence is more likely to occur with high investment in abstinent social support , and ongoing abstinence has been related to increases in abstinent persons within social support networks . Abstinence social support might increase abstinence self-efficacy because persons recovering from substance use disorders acquire effective coping skills modeled by their peers . Few investigations have examined the relationship between abstinence social support and abstinence self-efficacy and found these recovery resources to be positively related . However, characteristics of abstinence social support networks and the process of developing abstinence self-efficacy was found to differ between women and men recovering from substance use disorders , suggesting that other types of social support might be instrumental in the development of recovery resources. Other types of social support that promote health and psychological functioning include appraisal, belonging, and tangible support . Tangible support refers to instrumental aid one might receive, appraisal support to the availability of someone to talk to about one's problems, and belonging support to the availability of people one can do things with . Research evidence suggests these types of social support are, in part, positively related to health outcomes, whereas a lack of these resources poses greater health risk among women . However, the relationship between these types of social support and abstinence self-efficacy has not been investigated, and there is a need to examine various types of social support concurrently because any one type may not suffice. Thoits's theory of social support emphasizes the importance of examining the match between individuals' needs and appropriate sources of support, and it postulates that there are conditions under which stressors are likely to challenge social support. This would suggest that additional types of social support are needed to compensate for a social support type that is compromised or insufficient. For example, specific characteristics of social support network members, such as their similarity to persons' life experience , might facilitate effective abstinence social support through empathetic understanding on the one hand. On the other hand, stressors associated with incarceration, low socioeconomic status , and trauma histories may pose great challenges to well-being and reintegration among women with criminal justice involvement. For instance, Salina, Lesondak, Razzano, and Parenti found significant negative relationships between social support, and frequencies of various traumatic stressors among a sample of women with criminal justice involvement who reported many unmet needs . However, an examination of various social supports in relation to recovery resources has yet to be investigated. Therefore, examining various types of social support might help clarify why support networks consisting of similar others alone may not be sufficient for understanding predictors of recovery resources such as abstinence self-efficacy; especially among women with criminal justice involvement who typically have high rates of substance use . Rates of substance use among those with correctional histories are approximately the same as inmates' report of substance use prior to incarceration , suggesting that incarceration experiences might threaten the development of self-efficacious behaviors for abstinence. Thus there is a need to also examine whether incarceration histories are related to abstinence self-efficacy among women with criminal justice involvement who have substance use disorders. Understanding predictors of abstinence self-efficacy among this subset of women with substance use disorders would have important research and treatment implications. Although it is reasonable to expect high levels of abstinence social support would be related to increased abstinence self-efficacy as evidenced by research with clinical and community samples, this relationship might not be significant among women with criminal justice involvement who have substance use disorders because of the tremendous stressors and unmet needs these women experience . Overall, there is a need to examine whether abstinence social support predicts increased abstinence self-efficacy among women with criminal justice involvement who have substance use disorders. The present study examined several types of social support and incarceration histories in relation to abstinence self-efficacy among a sample of women with criminal justice involvement who have substance use disorders. We hypothesized that high levels of abstinence social support would predict increased abstinence self-efficacy. In addition, we explored whether appraisal, belonging and tangible social support types in addition to incarceration histories would predict abstinence self-efficacy. --- METHODS --- --- Procedures Participants recovering from substance use disorders who had involvement in the criminal justice system within the preceding 2 years were recruited from multiple sites in metropolitan Chicago and its suburbs from 2008 to 2011. Recruitment sites included Cook County Jail and multiple substance abuse treatment sites throughout Chicago, the surrounding suburbs, and Northern Illinois. Although recruitment staff actively visited these sites, research staff also posted recruitment flyers that were distributed in places that might provide some form of services to women who were formerly or currently justice-involved. Women who were interested in participating in the study contacted our study recruiters. Participants were also recruited using snowball techniques, which permitted other participants to refer women to the study. Most women agreed to participate in the study, with the exception of a few women who declined. No women were refused participation by study personnel. All women in the study were enrolled using Institutional Review Board-approved informed consent procedures. Participants received transportation passes to travel for their interviews and $40 in grocery store gift cards as incentive for their participation. --- Materials ABSTINENCE SOCIAL SUPPORT The Important People Inventory is a measure adapted from the Important People and Activities Inventory and has been used in previous research to assess abstinence social support from drugs and alcohol among those recovering from substance use disorders . Participants were asked to describe important persons from their social network within the past 6 months. This included specifying the quality of their networks relative to drug and alcohol use in addition to the activities they engaged in during the past 6 months. Participants rated members of their social network on a 5-point Likert scale that distinguished substance users and nonusers. This procedure resulted in computing a percentage of important persons identified by dividing the number of nonusing persons by the sum total of all persons identified as important persons, consistent with previous investigations on abstinence social support . The IP has good internal consistency , and the internal consistency of the IP in the present study was acceptable . --- INTERPERSONAL SUPPORT EVALUATION LIST The 12-item version of the Interpersonal Support Evaluation List was administered to assess three distinct types of actual or perceived social support. Tangible support refers to instrumental aid one might receive; appraisal support refers to the availability of someone to talk to about one's problems; belonging support refers to the availability of people one can do things with . The 12-item version of the ISEL was designed to measure these three separate functions of social support. Considerable research has been conducted with the ISEL and good internal reliability has been reported with the 12-item version of the scale . The internal reliability of the ISEL in the present study was good . --- ABSTINENCE SELF-EFFICACY We administered the Drug-Taking Confidence Questionnaire to assess participants' confidence in resisting the urge to use drugs or alcohol across 50 hypothetical situations. The DTCQ is rooted in Bandura's cognitive behavioral self-efficacy theory, and it is based on antecedents of substance abuse relapse . The DTCQ has been used among people with different addiction typologies . Because confirmatory factor analyses support the eight-factor model of the DCTQ's highly reliable subscales , we used a total confidence score in the present study by collapsing the subscale scores and averaging these scores on a scale that ranges from 0% to 100% . This total score approach to calculating self-efficacy for abstinence has been effectively used in previous studies . The DTCQ had excellent reliability with the present sample . --- DEMOGRAPHIC CHARACTERISTICS AND INCARCERATION HISTORIES The Addiction Severity Index-Lite , a briefer version of the Addiction Severity Index , was used to assess demographic characteristics and incarceration histories. The ASI has good internal consistency, excellent predictive and concurrent validity , and the ASI-Lite has been demonstrated as being quite comparable to the ASI with good validity and reliability . The internal consistency for legal status items from which we collected incarceration histories was very good . --- SOURCES OF INCOME We asked participants to identify their major sources of income for the past 12 months. This provided us a contextual understanding of employment status . --- Data Analysis Preliminary analyses were conducted to describe the sample in terms of sociodemographic characteristics in addition to describing rates of incarceration histories, levels of abstinence social support, general social support and abstinence self-efficacy. A hierarchical linear regression was used to examine the influence of predictors on levels of abstinence self-efficacy, entering predictors sequentially in a manner that approximated their temporal relationship to abstinence self-efficacy in the absence of any theory to guide us. This approach enabled us to examine the incremental validity of predictors and how much they individually contributed to the model in terms of variance. Sociodemographic variables and lifetime incarceration histories were included in the first and second steps to control for their variance. Abstinence social support was entered in the third step. The other three general social support types were entered in separate subsequent steps to understand their influence on abstinence self-efficacy apart from abstinence social support. These general types of social support were entered separately because they bore significant and positive moderate relationships to each other, thus justifying our use of hierarchical linear regression. To better understand the effects of this model, we ran the regression model by varying the order of general social support types entered into it and discovered only one type was a significant predictor in each combination of entry whereas the other two were not. In addition, the results were statistically similar when we entered these three general social support types in the same step , therefore the results of this model are presented. Testing these additional models confirmed tangible support was a significant predictor beyond abstinence social support and not a statistical artifact of any one model. --- MISSING DATA A listwise deletion approach was used to evaluate data and calculate analyses. Participants with missing data were excluded from analyses. A missing values analysis of all the independent and dependent variables indicated that the data were missing completely at random, Little's MCAR test, v 2 ¼ 39.54, p ¼ .14. --- RESULTS --- Preliminary Analyses Participants reported an average score of 80.48% for abstinence self-efficacy and an average score of 80.2% for abstinence social support. In terms of general social support, they reported average scores of 3.41 , 3.29 and 3.24 for appraisal, belonging, and tangible types, respectively, with an average of 46.48 months of lifetime incarceration histories . There were proportionately more African American women than those from other racial groupings, v 2 ¼ 284.92, p < .001, in the sample. --- Primary Analyses A hierarchical regression model was employed to test our hypotheses, and results of this model are presented in Table 1. Sociodemographic characteristics in the first step and incarceration histories in the second step were not significant. The inclusion of abstinence social support accounted for 2.8% of the variance in the third step, predicting significant increases in levels of abstinence self-efficacy. The inclusion of general social supports types in the fourth step accounted for an addition 7.7% of the variance, with only tangible support significantly predicting levels of abstinence self-efficacy. We ran this model with the inclusion of interaction terms based on main factors , and these three interaction terms were not significant predictors of abstinence self-efficacy. --- DISCUSSION The significant positive relationship between abstinence social support and abstinence self-efficacy suggests that social networks compromised of recovering peers might help women with criminal justice involvement who have substance use disorders develop important recovery skills to support abstinence. This finding extends previous research that examined these recovery resources in community samples that found a significant pathway between social support and self-efficacy . However, tangible social support was also significantly and positively related to increases in abstinence self-efficacy and accounted for more variance in our regression model, suggesting that social support types that are not abstinence-specific may have indirect effects that foster recovery resources among women with criminal justice involvement who have substance use disorders. Tangible social support was the only type of social support significantly related to abstinence self-efficacy in the present study. This finding is somewhat consistent with one investigation that found tangible support related to abstinence in a sample of persons with co-occurring substance use disorders . The need for tangible social support might be more pressing than abstinence social support among women with criminal justice involvement Social Support Among Substance Using Women who are disproportionately impacted by social and interpersonal oppressions. These include poverty, lack of stable housing, victimization from abusive relationships, and untreated co-occurring substance use disorders . Access to tangible resources is sorely needed among women with criminal justice involvement who are typically single parents, living in poverty, undereducated, unskilled, unemployed, marginally housed, and victims of childhood and adult physical and=or sexual trauma . In light of these stressors, results in the present study are aligned with Thoits' theory of social support either in that tangible support complements abstinence social support or that the needs of women with criminal justice involvement who have substance use disorders outweigh the potential benefits of abstinence social support. Although such claims can only be verified through additional research, findings in the present study are consistent with the conceptual framework of naturally occurring social support among women with criminal justice involvement . In addition, there was no significant relationship between incarceration histories and abstinence self-efficacy in the present study, suggesting that women's involvement with criminal justice systems do not impede their ability to develop this important recovery resource. --- Limitations Although predictors of abstinence self-efficacy might be better understood when accounting for various types of social support, there are some limitations in the present study. For instance, participants' stressors and needs probably had an impact on their levels of social support and abstinence self-efficacy, but these were not directly examined in the present study. Comparison groups would help us better understand the role of various social support types in relation to recovery resources such as abstinence self-efficacy, and findings in the present investigation certainly have implications for research with noncorrectional samples. It is possible that abstinence rates and treatment duration had an impact on levels of abstinence social support and abstinence self-efficacy in the present study, and future investigations should account for these potential influences. Finally, the use of self-reported data at one time-point and chain-referral sampling techniques are other limitations of the present study. A repeated-measures design might have provided more information in relation to changes in social support types and abstinence self-efficacy over time whereas snowball techniques have the potential of negatively affecting sample representation. Although results in the present study have implications for future research, social workers should closely examine the need for various types of social support beyond abstinence-specific networks when working with women who have substance use disorders, especially those who have criminal justice involvement. --- Implications for Research and Practice Most research on abstinence self-efficacy has examined this recovery resource as a predictor, and the present study adds to the growing body of research that is examining abstinence self-efficacy as an outcome variable. More importantly, our findings point to the limitations of abstinence social support among one of the most vulnerable populations of persons with substance use disorders; women who have criminal justice involvement. Findings in the present study emphasize the need for researchers to examine social support types which are specific to the needs of women who have criminal justice involvement. Findings in the present study support the need for social workers to consider context-specific factors such as social capital when assessing and targeting substance use recovery resources. The present study examined predictors of abstinence self-efficacy that have not been examined in previous investigations involving persons with substance use disorders. Our investigation is innovative in that it examined several types of social support in relation to this important substance use recovery resource and found a general type of social support to be a better predictor than abstinence social support among women with criminal justice involvement and substance use disorders. Social workers should assess various social resources among women in criminal justice settings who have substance use disorders. In addition, social workers should examine whether various social resources might be instrumental in meeting these women's diverse needs including substance use recovery resources such as abstinence self-efficacy. Findings in the present study strongly support the assumption that women exiting the criminal justice system need to be linked not only to self-help programming where they will receive peer-based abstinent specific support but also linked to community agencies to get their basic needs met. Social workers should examine how this strategy may increase or potentiate the effects of abstinence social support on recovery efforts and increase the probability that justice involved women will remain abstinent. Overall, findings in the present investigation suggest various types of social support are necessary when providing relapse prevention programs or strategies, particularly among women with substance use disorders who have criminal justice involvement.

Social support types (abstinence, appraisal, belonging, tangible) were analyzed among a sample of women with criminal justice involvement and substance use disorders (n ¼ 200). Hierarchical linear regression was conducted to examine social support types in relation to changes in abstinence self-efficacy while controlling for incarceration histories. Only abstinence social support and tangible social support predicted significant increases in abstinence self-efficacy, with tangible support accounting for more variance in the analytic model. Findings suggest women with criminal justice involvement who have substance use disorders have basic needs that if met would have an indirect effect on their recovery. Implications for treatment and research are discussed. Abstinence self-efficacy is a recovery resource that has been conceptualized as the extent one is confident in effectively engaging in behaviors to maintain abstinence, based on Bandura's (1997) cognitive-behavior self-efficacy theory. Studies have shown abstinence self-efficacy is strengthened through social structures such as 12-step involvement in groups such as Alcoholics

Introduction The coronavirus disease spread across the globe in 2020 resulting in death and long-term health challenges. Subsequent lockdown restrictions resulted in the curtailment of many social freedoms and the removal or scaling down of health and social care services [1]. Families who relied on social care services to provide --- Open Access *Correspondence: [email protected] 1 School of Nursing and Midwifery, Queen's University Belfast, 97 Lisburn Road, Belfast BT9 7BL, Northern Ireland Full list of author information is available at the end of the article care and support experienced significant stress [2] and reduced mental health [3,4] which may have long-term implications for their health. Family carers provide significant and ongoing care for their family members with intellectual disabilities [5]. Profound and multiple intellectual disabilities refers to people with a severe learning disability and cognitive impairment that significantly affects their ability to communicate and live independently [6]. Due to severity of impairment, the burden of care experienced by family carers of those with PMID might be considerable. People with PMID often have severe, life-long impairments including reduced vision, hearing and mobility, with associated complex physical and mental health conditions [7,8]. They are at greater risk of experiencing multiple and complex physical health needs, including epilepsy, coordination disorders, respiratory infections, pneumonia, dysphagia, gastroesophageal reflux disorder, helicobacter pylori, constipation and urinary incontinence [9][10][11] These conditions require on-going management by family carers and services to prevent deterioration and minimise the potential of complications such as chest infection and pneumonia [12]. Services required by people with PMID and intellectual disabilities prior to the pandemic included physiotherapy, speech and language therapy, personal assistants, day activity centres, respite care, special education, mental health and social care [13][14][15][16], which were essential for their continued health and well-being. With the removal of in-person treatment due to pandemic restrictions, UK and Dutch studies showed increases in referrals to psychiatric services for deterioration in the mental health of people with ID [14,17]. The removal of in-person physiotherapy services to treat coordination disorders meant that treatment was delivered online with the rehabilitation burden placed on family carers [3]. Prior to the pandemic many of these services had experienced cuts, or offered limited service provision, leaving people with disabilities feeling their needs were not being met [18,19]. Research has shown that people with ID and their family carers experienced significant strain during the pandemic. An international online survey of carers and support professionals conducted in 12 countries between August and September 2020 showed that carers experienced high levels of stress and depression [1]. The strongest predictor of carer well-being were changes in mood of the person they cared for [1]. Similarly, a survey of 323 Spanish carers of people with ID also documented increased levels of stress which was attributed to an increase in their caring role due to service closures [13]. A qualitative study conducted with eight English parents of people with ID during the first lockdown period of 2020 identified themes of powerlessness, coping , support and reduced well-being [20]. Interviews with 24 people with ID living in England revealed experiences around a lack of consideration in government planning, cuts to social care provision, removal of opportunities for social contact and lack of vaccine prioritisation [21]. While evidence existed to show that people with ID were at increased risk of morbidity and mortality from Covid-19 [22] the UK government did not originally assign them priority status for vaccination [21]. Non-Governmental Organisations provide ongoing care and support for families of individuals with disabilities. Among others, these include supported living, residential services, day care, vocational training, supported employment, respite care, leisure and advocacy [23]. With the removal of services and support structures during the COVID-19 pandemic, NGOs became even more vital with many services moving to online support platforms to provide essential services for families [24]. These included, art therapy [25], sport and fitness [26] mental health support [21], and the provision of technology to enable online interaction [15]. The use of online platforms and telephone support services were a lifeline for many, whilst offering both opportunities and challenges [27]. While in the past families may have expressed a preference for face to face support, the COVID-19 pandemic demonstrated that some services were effectively delivered online [28,29]. For example, videoconferencing was thought to be a useful tool and source of support for carers in staying in touch with day services [20], while online training to provide mental health support was shown to improve carer well-being [19]. However, few researchers have yet developed an evidence-based online programme to meet the needs of family carers. Programmes which include the perspectives of family carers will more closely align with their needs and be better suited in meeting these. Coproduction of such a programme is an approach that offers great promise through recognising that traditional notions of expertise are complemented by the lived experience of service users [30]. Co-production offers opportunity to bring together diverse perspectives of key stakeholders to truly collaborate on a topic that is of mutual benefit [30]. Researchers have successfully employed co-production to investigate diet and weight management in individuals with ID [31], to identify health priorities for families of South Asian children [32] and to create an intervention for carers of people with dementia [33]. In addition to gaining the perspectives of family carers in creating a programme to meet their needs, the views of NGOs as key stakeholders in the lives of carers are an additional source of information. Non-Governmental Organisation workers have played an important role in supporting families of people with intellectual disabilities and had regular contact with families during the COVID-19 pandemic [23]. In addition to continued provision of many of their individual programmes, NGOs advocated for the needs of persons with ID at government; translated COVID-19 guidance within supported living settings and adapted their services to take advantage of new technologies [23]. As a key source of support for family carers during the pandemic, NGOs had regular contact with carers through their online activities and via telephone [23]. They built on already established relationships with carers and formed close bonds with them. They are therefore well placed to provide insights into the challenges family carers experienced. The COVID-19 pandemic has brought about changes to the lives of many, however, to date little research has focused on the experiences of family carers of people with PMID. The research presented here sought to capture the experiences of family carers of people with PMID during the COVID-19 pandemic. Further, the project sought, with the aid of NGOW and family carers to develop an online support programme which would ameliorate the impact of the pandemic on carers. We sought to address two research questions: 1. What do NGOW perceive to have been the particular impacts of the COVID-19 pandemic on family carers of people with PMID?; 2. What features of an online programme would be helpful in alleviating the impact of the COVID-19 pandemic on family carers? --- Methods --- Design This manuscript describes the first phase of data collection in a larger study intended to explore the experiences of family carers of people with PMID and develop a programme to ameliorate the impact of the COVID-19 pandemic. These findings describe the views of NGOW who provided support for these families during the pandemic. Employing a qualitative design allowed us to explore the experiences of NGOW during the pandemic and gain their perspectives on how the pandemic affected their clients. --- --- Data collection Five focus groups were held between September and December 2021. Non-Governmental Organisations who provided care to families of individuals with intellectual disabilities were invited to take part. We sought the views of those with a client facing role and employed convenience sampling for the purpose of recruitment. Letters of invitation were emailed to eligible staff by a gatekeeper in their organisation, followed by a participant information sheet approximately one week later. Those who expressed an interest in participating in the study shared their details with the gatekeeper who arranged a suitable time for the focus group to take place. Focus groups were conducted using an online platform with discussions recorded for later transcription and analysis. Focus group discussions lasted between 60 and 90 min. The focus group guide is included in Table 2. Informed consent and basic demographic details were gathered by means of an online questionnaire. --- Ethical considerations Ethical approval was granted by the Faculty of Medicine Health and Life Sciences ethics review board at the Queen's University of Belfast. Participants were required to provide written consent prior to participation and were fully informed of their right to withdraw, the limits of confidentiality and data protection. All procedures were conducted in accordance with the Declaration of Helsinki. --- Data analysis The six stage process of thematic analysis [35] was employed to capture important features of the data in relation to the research questions. Themes were inductively developed and represent major motifs across the data. --- Rigour To ensure rigour, the research protocol was first reviewed by two independent researchers as part of our application for ethical approval. The same researcher conducted all of the focus groups with recordings transcribed verbatim to ensure accurate accounts of participants' views were captured. The researcher was an experienced psychologist with over twenty-one years familiarity with qualitative methods. Credibility of the analysis was achieved through triangulation by two members of the research team with direct quotations presented below to demonstrate confirmability. --- Findings Thematic analysis resulted in four primary themes and three subthemes which are summarised in Fig. 1 and explained below. --- Mental and emotional health --- I think families were scared and then confused and I think they have went through different stages of emotions since then, stages of grief and there has been different stages throughout the pandemic. There has been anger, frustration and a lot of negative emotion. [Female, NGOW, 45-54 years old] The pandemic was a highly emotive time for all and one of the main themes to emerge from the focus group interviews related to the impact of the pandemic on people's mental health and emotional well-being; both family carers and the individuals they have been caring for, in addition to that of staff within the NGO sector. This theme was apparent in all of the focus groups and incorporated a broad range of ways in which people's mental health had been impacted, for example, loneliness, fear, anxiety and uncertainty. These subthemes are presented in Sects. 1.1 -1.4. One participant commented on the impact of lockdown on their own mental health, and that of their team: 9. Are there any topics that we haven't covered today or that you would like to talk about before we finish the interview? --- I've really struggled with my mental health throughout lockdown like all of us have in differ- --- ent ways and there is not one of us that hasn't been in tears in a meeting [Female, NGOW, 45-54 years old] Concern about the pandemic having a negative impact on mental health largely focused on that of family carers and those individuals they were caring for. --- We were really worried about people, some of our advocate team, we were seriously worried about what was happening to their mental health because of what was going on. I think mental health was definitely one of the worst things. [Female, NGOW, 45-54 years old] A participant in another focus group reflected on how successive events in family carers lives meant that there was a cumulative negative effect on their mental health: --- I think you could also look at all of our family carers as having gone through multiple experiences of trauma because it's just been horrendous [Female, NGOW, 55-64 years old] However, there was an element of hope from some of the NGOW who felt that, while they had gone through difficult times, they also felt supported by their colleagues and were thankful for them. We have all cried, self-advocates have cried and we have had carers or family members in tears and it's trying to just say we are all going to get there, you are not alone, we are all in this together, which is great for our team. [Female, NGOW, 45 --- -54 years old] Concern for the long-term impact of the pandemic on family carers' mental health and that of their family members was evident: These people need the help, these people are in crisis and I feel like I have been alongside my friends and they are on their knees. They are coping, some of them aren't coping and it's really, really tough and there doesn't seem to be a light at the end of the tunnel and the support still isn't coming. [Female, selfadvocate, 35 --- -44 years old] Non-Governmental Organisation Workers, who were also family carers or self-advocates, described the impact of lockdown restrictions on the mental health of family members with learning disabilities. This was often discussed to the exclusion of the impact of the pandemic on their own mental health. One self-advocate said: --- I think for me I suppose it reminded me we talk through our anxieties, our losses and our despair at lockdowns and what is happening. She was quietly experiencing all of those emotions, and that loss and not saying very much but it was very real. [Female, self-advocate, 55-64 years old] Participants highlighted that while some people with PMID may not be able to fully communicate their needs it did not mean that they had not been emotionally and psychologically impacted by the pandemic. Some selfadvocates described how their family members' behaviour had changed as a result of stress brought on by lockdown restrictions. For example: --- He [my son] was actually showing a lot of signs of stress by then… …[My son] was an extremely easy going guy, never had any behavioural issues with him in the sense of being aggressive or anything like that. If the staff took him out in the car just to have his hour out or they took him out in the wheelchair for an hour, when they came back and they were trying to get him back indoors he was kicking them, he was punching, he was trying to close the car door. There were all sorts of reactions that were saying he didn't want to go back in, he was happy being outand-about and that was so unlike him. [Female, self-advocate, 65+ years old] --- Isolation Isolation is often an issue for families of people with more severe intellectual disabilities or PMID, with the pandemic exacerbating this. Families had many of the services and supports they regularly relied upon removed, leaving them feeling powerless and alone: It's an isolated group that we work with anyway but when services stopped for them a lot of folk withdrew their children before the schools finished and services began to stop. [Female, NGOW, 45-54 years old] It's like we said they are very isolated and particularly when they have young children, normally parents meet at activity clubs and school gates and our families don't have that. Also with family carers, one in particular I remember her son like a lot of people with PMID is non-verbal and she has a small circle of family. During the pandemic, she can speak to her son but her son can't speak back to her and so she is having no communication at all, that would be something if there was somebody to chat to. [Male, NGOW, 25-34 years old] --- Fear of COVID-19 While there was clearly a certain amount of fear among the general population regarding COVID-19, this was exacerbated by the increased vulnerability among those with learning disabilities, as exemplified by the following quotes: We were also scared --- The exhaustion of caring Exhaustion was a common subtheme expressed by both the self-advocates and NGOW. Family carers were left to cope without the support of day centres, family, friends and with the removal of social care packages. Many faced a disproportionate burden of care leading to increased stress and depression [1,13]. --- Many families had some support but most of them didn't have the support they really needed and they were already at exhaustion stage prior to the pandemic coming in. [Female, NGOW, 55-64 years old] It's like when you are in the middle of a crisis you keep going because you have to keep going and then when the crisis starts to subside you start to let out stuff that you have been holding in for a long time. I think we are coming to that phase with staff who are absolutely worn out and fatigued. [Female, selfadvocate, 35-44 years old] "They who shout the loudest" A recurring theme across the data was how many family carers had to fight for 'basic' needs to be met and to have necessary supports provided. There was a feeling among those participants who were self-advocates that they had a black mark against their name. They felt that they had to battle and argue for so long and with such force that service providers were likely to have 'marked their cards' , meaning that those in positions of authority disapproved of them. One NGO sector participant said: --- They had to fight really hard to have the support put in place and it was all taken away from them [at the start of the pandemic]. [Female, NGOW, 45-54 years old] Another member of staff emphasised how necessary it was for family carers to equip themselves with the necessary knowledge to, in a sense, campaign and 'fight the system' to have some of their needs met: The thing that we have always seen is that when parents are able to kick up a fuss, in the nicest possible way, and are informed with the right information then that is game changing… …You shouldn't have to feel like you are battling against the system every time you want something done that you are well within your rights to do. --- [Male, NGOW, 25-34 years old] It seemed that family carers who had the ability, strength and capacity to 'fight the system' would have their voices heard, and there was a sense that less vocal families might lose out. One self-advocate voiced her anger and the lengths that she had to go to, to have her son's essential care and support needs met: --- Nobody gives a toss until I start to create a fuss over one thing or another they all assume everything is ok… …Even to get him vaccinated I had to create hell… … I went around all of my friends and wound them up as well, gave them the same arguments. [Female, self-advocate, 65+ years old] I have always got the services that I wanted for [my son], I have always had to shout for them but how awful that you give services to people who bully you into it because that is essentially what I have done over the years. I think that is a corrupt system that lets me do that, but while the system works like that I will work the system to [my son]'s benefit. [Female, self-advocate, 65+ years old] There was also emphasis that families were not asking for any services above and beyond what they were entitled to. --- So, we were not asking for a Rolls Royce service, we were asking them to just do what they had promised when he got to adulthood. It was a long battle and it got quite horrible at times but anyway we got him in there. [Female, self-advocate, 65+ years old] "No one arrived to help out" Lack of trust in statutory services A further theme which emerged from the data was the criticism by some participants regarding statutory services. There was a sense that NGOs were willing to go the extra mile for families, in comparison to statutory services. This is illustrated below where a Female, NGOW, 55-64 years old, described a lack of response from statutory services to continued queries from family carers. --- I'm not trying to blow the [NGO] trumpet but lots of families have said if it hadn't been for our organisation they don't know what they would have done or where they would have been. [Female, NGOW, 45-54 years old] I think the third sector organisations have completely held up the community through the pandemic, absolutely and it's shocking. I was in a meeting a couple of weeks ago with other third sector organisations and yes we don't have all the red tape and we can respond flexibly but I mean there is no excuse for some of it. [Female, NGOW, 45-54 years old] Some participants mentioned the difficulty in communication with, for example, social services, during the pandemic: You just can't imagine how some of the families managed because no one came across the door, no one arrived to help out. They would make phone calls to relevant professionals in social work, GP practices and there was no answer, there was no one willing to come out which was unbelievable for all of us. [Female, NGOW, 55-64 years old] A lack of trust was identified between family carers and some statutory services or local authorities. Family carers reported a lack of trust in statutory service providers during the pandemic due to concerns that they were being forgotten, a sentiment echoed by other researchers [36]. Poor communication and ongoing problems with service provision may have further reduced trust and led to family carers turning to NGOs as sources of support and advice: --- The whole redesign of services has meant that families have lost trust in local authorities and services and there wasn't a lot of trust before. There was never a great partnership because of communication between statutory services, parents and the lack of support. That trust is almost all gone now and how that ever can be brought back I don't know. [Female, NGOW, 45-54 years old] Fundamentally there is a huge lack of trust between families and local authorities, and local authorities and families. I think local authorities think that families are going to abuse the system which they are not… …A lot of what the people are asking for is not pie in the sky stuff, it's all relatively low cost realistic stuff that makes sure people are happy and healthy. [Male, NGOW, 25-34 years old] The need to consult family carers regarding matters that affect them was identified as an issue, which statutory services neglected to do on a regular basis. This left family carers feeling unheard by statutory services which may have contributed to the erosion of trust. Conversely, NGOs took the time to listen to the concerns of family carers resulting in greater trust and provision of a service which was directly informed by their needs: --- For risk assessments, we were asking families what should be here, what are you thinking? What is an issue for you, what should we be doing to keep you safe? It's about making families feel included which they don't feel like through other statutory services. They don't have that real consultation where they are listened to. [Female, NGOW, 45-54 years old] --- Creating an online support programme One of the main foci of the focus group discussions was encouraging NGOW to think about what a successful support programme for family carers would comprise and might look like. Responses strongly emphasised the need to adopt a coproduction model -i.e. consulting family carers about what they would look for in a support programme: --- I think anything that you create needs to be created alongside involving experts by experience be that carers themselves, people with learning disabilities, they will come up with the best suggestions of what they think works and what doesn't. [Female, NGOW, 45-54 years old] I suppose we take it so much for granted but the word co-produce, I would say speak to families and ask them what they need, what would work for them, what time of day, what sort of venues, what sort of things would be useful. Some things are so basic but sometimes people develop all these things and they have never actually spoken to somebody about what would work, so that would be my thought. [Female, NGOW, 55-64 years old] A common topic discussed was the need for peer support, the extent to which peers have supported each other during the pandemic, and the potential for a support programme to connect people. It was felt that peer support for family carers would reduce feelings of isolation through sharing experiences and creating social connections to support emotional wellbeing. Family carers have expressed appreciation for support received through the use of technology which linked them with family, friends and certain services [20]. --- First thing that comes to mind if some sort of reduction in isolation. I think a lot of our families are very lonely so anything that can help that. [Female, NGOW, 45-54 years old] One thing I've always thought would be really good and I think it would be difficult to get up and running and also to facilitate it but some sort of buddy system to introduce families to others. [Male, NGOW, 25-34 years old] What people need is a network that they feel they can tap into when they need to. I am a big believer and I think it has helped a lot of people together, somewhere where people can just get together, have a natter and have a coffee. Those circles of support have been lifelines for a lot of people I think during the pandemic. For me if there was something that could really highlight stuff it's people really seeing the value in those, and they are not just nice to have things, they are actually critical. [Male, NGOW, 25-34 years old] Some participants discussed efforts that their organisations made, or that they had been involved in, during the pandemic, which were helpful in supporting family carers. Some NGOs had established online activities to engage those they work with, for example online quizzes, bingo, cookery demonstrations, yoga, mindfulness, wellbeing, music therapy, horticulture, exercise classes. --- We were able to develop our virtual activity programme. Every single member of staff pulled into the development of that programme and what we now have and have consistently had pretty much since the first couple of weeks within the pandemic is a rolling programme of music, drama, storytelling, arts and crafts, sporting activities, carer catchup, information workshops. [Female, NGOW, 45-54 years old] Others felt that an online programme might include a one-stop shop repository of relevant, honest, reliable and easy-accessible information. I think that information which is needed is where people can get support for a range of things really: care, support, leisure. [Female, NGOW, 45-54 years old] --- Let's say when I was thinking about [my son] going into independent living, I would have liked to have been able to type in something like -can people tell me what arguments to use when I am trying to get my head around how I am going to present my case? Because no person ever presents your case for you, you are the one who is going to have to do that. [Female, self-advocate, 65+ years] There was emphasis on the business and exhaustion of family carers and the fact that they would only be able to access support if they were relieved of their caring duties or if it was at the right time of day: --- Discussion This research adds to the body of evidence demonstrating the impact of the COVID-19 pandemic on the mental health and well-being of family carers through survey [2][3][4]. It extends this work by exploring this impact on family carers of those with PMID, by capturing the experiences of NGOW and self-advocates across the UK and Ireland. Our qualitative approach has added depth to the descriptions of reductions in mental health experienced by family carers. Our first research question sought to explore the challenges faced by family carers of people with PMID during the COVID-19 pandemic from the perspective of NGOW. Participants recounted how family carers experienced feelings of isolation, confusion, fear and exhaustion. These feelings will likely be echoed by many readers of this journal, however, carers of people with PMID were already experiencing significant strain prior to the pandemic [37,38] which was then exacerbated by restrictions. It is unknown what the long-term impact of the COVID-19 pandemic will be on family carers, however, efforts should be made to provide appropriate, tailored resources and targeted interventions to reduce any impacts. We would suggest this be achieved through co-production with key stakeholders including family carers and NGOs who possess significant lived experience. Co-production has been successfully used in the creation of a computerised cognitive behavioural therapy package for carers of people with dementia, where it was felt to be central to design of the programme [33]. Our findings highlight the ongoing need for family carers to fight for services and supports to which they are entitled. Family carers who have the time, resources and confidence to advocate for their children with disabilities find this struggle difficult [38]. Knowledgeable, self-advocates described how they had been forced to 'work the system' to have their son vaccinated during the COVID-19 pandemic. Family carers who lack the confidence, knowledge or time to fight for service provision for their children may not receive the services they are entitled to [39,40]. Improved, easily accessible information on the entitlements of families to services should be provided to make carers aware of what services are available. GPs, social workers and other professionals who work closely with family carers should better understand their ongoing needs and provide services which meet these needs. These findings should be considered in the design of future service provision, particularly in the context of ameliorating the impact of the COVID-19 pandemic. Participants identified a lack, or decrease in trust, with statutory services when compared to NGOs. Services during the COVID-19 pandemic were placed under considerable pressure meaning that many supports, initiatives and some health procedures were not provided [41,42]. Research has consistently reported that families experienced feelings of abandonment by statutory services during the pandemic [16,20] with NGOs, who had built relationships with families prior to and throughout the pandemic, stepping in to offer support. This lack of provision could be one reason why trust had been damaged. However, while this may have been most acutely felt during the COVID-19 pandemic the feeling that statutory services do not 'listen to' families is a common theme prior to the pandemic [40,43,44]. Participants described a lack of response to requests for support from statutory services which left families feeling isolated and uncared for. In contrast, NGOs were perceived as a 'lifeline' for families who offered support and fostered a sense of community. In addition to capturing NGOW views on the impact of the COVID-19 pandemic on family carers our second research question sought to understand what aspects of an online support programme might prove useful in reducing such impacts. Participants felt that including the voices and experiences of family carers in any such programme would be crucial as would some sort of mechanism for connecting people. Non-Governmental Organisation Workers described the efforts their organisations had gone to in providing activities and social events online that were delivered at times suitable for carers. Our recent systematic review of the evidencebase showed a lack of consultation or co-production in the development of online interventions for family carers of individuals with intellectual disabilities [27]. This may be due to therapeutic considerations , however, interventions which take the needs of family carers into consideration when designing such programmes may address a gap in provision and meet current and future unmet needs. Online programmes offer the potential for a low cost, accessible and adaptable approach to supporting family carers which are accessible from home. Researchers have employed co-production in developing an online mindfulness intervention to reduce stress for family carers of people with ID [5]. This research built on the existing intervention by co-producing a peer mentoring component to the existing mindfulness programme [5]. Other researchers have sought to consult with family carers about their preferences and desire for further supports in the Positive Parenting Program which they achieved through focus group interviews [45]. However, online platforms are not without disadvantage which should be taken into consideration in designing any new programme. For example, families may not possess adequate technology or access to a reliable internet connection [46,47], may experience technical difficulties [48] or may prefer face to face interactions. Our findings suggest that family carers trust NGOs, therefore, it would be advisable to build on established relationships with these organisations, who have provided ongoing support to families, in some cases for many years, when creating online programmes. --- Policy implications Policy makers and service providers need to respond to and address the long-term consequences of disrupted care during the COVID-19 pandemic. There is an existing deficit in service provision for many families who may struggled to access adequate supports, a challenge that remains to be addressed. It is important to meaningfully include family carers in the design and implementation of future support programmes to ensure they are acceptable and fit for purpose. There is therefore an opportunity for policy makers and service providers to model future changes to existing or new services in consultations with NGOs to ensure the views, experiences and needs of family carers are heard and addressed. --- Strengths and limitations To our knowledge, this is the first UK and Republic of Ireland wide study to explore the impact of the COVID-19 pandemic on family carers through the lens of NGOW. This is important as NGOs worked closely with family carers during this period and were an invaluable support for many. A further strength of this study is the use of qualitative methods which adds depth to a number of existing surveys. Research efforts in the early stages of the pandemic sought to gather data quickly, in a responsive fashion. While extremely useful, this data failed to capture the complete picture of individuals' experiences, a gap which this study addresses. The participants included in this study were aged from between 25-34 and 65 + years. This broad age range suggests that our participants may be representative of NGOW in the UK and Ireland. However, it should be noted that 100% of participants were Caucasian which is unlikely to reflect the diversity of NGOW. Participants included NGOW and self-advocates which adds a crucial perspective to study findings. Self-advocates as carers gave a personal account of how the pandemic impacted on their families and the stresses they themselves experienced. Whilst our study was conducted across the UK and Ireland, it is not possible to fully determine the transferability of the findings to other countries. Ultimately, this study explored the experiences of family carers viewed through the lens of NGOW. Therefore, the findings may not fully represent the experiences of family carers more widely. We are now preparing a follow-on paper using further data which will capture the direct experiences of family carers of people with PMID. --- Conclusions The reduction in services, coupled with ongoing historical problems with accessing supports, has increased the vulnerability of family carers of those with PMID. These families will continue to provide crucial care for their family members across the lifespan, sometimes to the detriment of their own health and well-being. Non-Governmental Organisations have provided ongoing support during the COVID-19 pandemic and are a trusted source of advice and information for carers. They are thus an important contributor to any efforts to design new, or adapt existing, services. Such services may be delivered in person or online with the caveat that family carers should also be involved. --- --- Abbreviations Authors' contributions ML formulated the research question and wrote the manuscript. TF collected, analysed and interpreted the data. MB, LM, MT, EMcC, ST and NH critically reviewed the manuscript and approved the final draft. --- --- --- Competing interests The authors report no competing interests. • fast, convenient online submission • thorough peer review by experienced researchers in your field --- • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research ? Choose BMC and benefit from: ? Choose BMC and benefit from: ---

Background: Family carers of people with profound and multiple intellectual disabilities (PMID) experienced a reduction in healthcare services due to the COVID-19 pandemic. Many subsequently turned to Non-Governmental Organisations who worked to support families. However, little research has sought to capture the experiences of family carers or identify effective interventions which might support them. To address these concerns we explored the views of Non-Governmental sector workers across the UK and Ireland who supported families people with PMID during the COVID-19 pandemic. We also sought to explore their views on the characteristics of online support programmes for family carers. Methods: This study employed a qualitative design using focus groups with participants (n = 24) from five Non-Governmental Organisations across the UK and Ireland. A focus group guide included questions on challenges, supports, coping and resources which helped during lockdown restrictions. Focus groups were held online, were audio recorded and transcribed verbatim. The resulting transcripts were pseudonymised and subjected to thematic analysis. Findings: Four themes were identified (i) 'mental and emotional health' , (ii) 'they who shout the loudest' (fighting for services), (iii) 'lack of trust in statutory services' and (iv) 'creating an online support programme' . Mental and emotional health emerged as the most prominent theme and included three subthemes named as 'isolation' , 'fear of COVID-19' and 'the exhaustion of caring' .The COVID-19 pandemic has increased the vulnerability of family carers who were already experiencing difficulties in accessing services and supports for their families. While Non-Governmental Organisations have been a crucial lifeline there is urgent need to design services, including online support programmes, in partnership with family carers which adequately address their needs.

INTRODUCTION Community-engaged research can provide researchers a unique opportunity to work more closely with their population of interest and therefore, the opportunity to obtain a more "emic" or insider perspective. Community-engaged research is a framework or approach for conducting research, not a methodology in and of itself. It is characterized by the principles that guide the research and the relationships between the communities and academic researchers. Community-engaged research requires partnership development, cooperation and negotiation, and commitment to addressing local health issues. The opportunity for community members and researchers to learn from each other to work towards a common goal can make interventions more relevant and meaningful for the target population and more effective for interventionists and other researchers. Community-engaged research in the area of HIV-prevention intervention development has led to the a number of population-specific prevention interventions . Research has shown that African American young men who have sex with men are at high risk for contracting HIV and other sexually transmitted infections . Nearly two-thirds of all YMSM aged 13-24 years with HIV infection in 2008 were AAYMSM . Specific to Los Angeles County, Africans Americans represent 9.4% of the county, but 20% of AIDS cases Further, AAYMSM in Los Angeles have much higher HIV prevalence than White YMSM , and have 4.4 times higher odds of HIV infection compared to White YMSM . It is therefore important to gain a rich and meaningful understanding of the different communities in which AAYMSM live and interact to provide researchers and service providers the opportunity to better understand the social and cultural structures that may serve as risk or protective factors for HIV-risk behaviors. This level of understanding is crucial to the development of culturally and developmentally appropriate HIV-prevention interventions. --- Minority Stress and Social Support AAYMSM occupy a unique position in the larger society; they are often stigmatized within the larger African American and gay communities and face racism and homophobia in the general society . While the effects of coping with these challenges are not well understood, a theoretical framework to explain the relationship between stigma, prejudice, discrimination and negative health outcomes, including involvement in HIV risk-related behaviors is the theory of minority stress ). Minority Stress Theory suggests that socially marginalized groups, such as the African American community and the gay community, experience mental and physical health problems resulting from negative social environments created by stigma, prejudice, and discrimination Social support may serve as an effective coping mechanism against life stress. Few researchers have explored the stress-buffering and positive effects of social support among African Americans, and those studies report inconsistent or contradicting and complex findings . Researchers have found that friend and family support can serve as a stress-buffer , whereas other researchers have found no stress-buffering effects . Interestingly, African Americans have been shown to seek out informal support for advice or help to reduce negative consequences of stress . Moreover, an ethnographic study on African American MSM in New York City reveals that AAMSM developed alternative familial networks as an alternative to their biological family . This seeking of support from informal networks is not unique for AAYMSM. Research among other youth who feel marginalized from more mainstream society indicates that these youth may form or join a subculture to provide them support and identity . Subcultures such as punks, Goths, "gamers" and "ravers" offer a place and community that is welcoming of youth of various ethnic and cultural backgrounds seeking others with similar interests and beliefs, and who, for a variety of reasons, may not fit into mainstream --- Introduction to the House and Ball Communities The House and Ball community is one such subculture. The roots of the House and Ball communities originate in Harlem in the 1920s. Annual Harlem Balls of the 1920s involved working-class, mostly African American men under the age of 30 performing in elaborate and ostentatious female attire as a popular form of entertainment . In addition to offering entertainment for spectators at these annual Balls, young men dressed in women's attire, or "in drag", to compete in costume competitions for cash prizes. Although Balls originated as a safe space for "female impersonators" to compete amongst one another in New York, this phenomenon has become more inclusive of AAYMSM of different sexual and gender identities. House and Ball communities are currently present in cities such as Los Angeles, Oakland, Atlanta, Chicago, Philadelphia, Baltimore, and Washington, DC . As a result of a 1990 documentary of the New York House and Ball scene, "Paris is Burning," the House and Ball scene gained some notoriety as a subculture for transgender and men who have sex with men of color. The limited research with these communities suggests that HIV is a major public health concern. From 1998 to 2000, research on an outbreak of tuberculosis among House members in Baltimore, Maryland and the New York City area found that 16 of the 26 House and Ball participants were living with HIV . This study highlights the importance of HIV research within these communities as well as the closely connected nature of its members, which has epidemiologic implications for the spread and transmission of disease. In a more recent study on the New York City House and Ball scene, found a seroprevalence rate of 20%, with 73% of those testing positive for HIV unaware of their HIV status. Little research exists on the House and Ball scene and the role it plays with respect to social support. One qualitative study describes the House and Ball communities in the San Francisco and Detroit areas as communities that are accepting of AAYMSM, as well as African Americans of other gender and sexual identities. Houses are identified as having a sense of family, friendship, and support. Sanchez, Finlayson, Murrill, Guilin and Dean also suggest that support is an integral component of the House and Ball communities that assuages the negative effects of stigma and life stress on risk-taking behaviors. In this paper, we describe a community-engaged research project with the Los Angeles House and Ball scene, in which the perspectives of the leaders of these communities are captured to better understand how these communities may protect and/or increase its members' risks for HIV infection. Data from this study will be used to inform the development of an intervention that is designed for and by the target population. --- METHODS This study utilized a number of different methods including; participant observations at Balls, meetings and other events; and semi-structured, qualitative interviews with House leaders. All methods were presented to the local House leaders in order to ensure that the methods were not intrusive to the community's activities and that the methods were appropriate for the target population. The methods and research activities were reviewed and approved by the Institutional Review Board at Children's Hospital Los Angeles. Prior to being enrolled in the study, all interview respondents completed an informed consent. --- Ethnographic Study Design, Sampling, and Measures Participant observations were completed under the direction of the ethnographer, in teams of 2-4 research staff. Between June 2008 and December 2009, participant observations were conducted for a total of 135 hours at 37 events. Research staff attended all scheduled Ball events as well as monthly Alliance meetings . Observation notes were taken and stored on an observation form which was developed to capture observations related to: the setting, number and types of participants, behavior related to social and sexual networks , familism within and between Houses , substance use, House standards and rules and presence and attitudes towards HIV prevention . These observations provided an opportunity to observe the different events and conduct informal interviews with participants including House parents and members. Staff members were encouraged to interact with participants and ask questions to clarify activities or to help identify members of the community. Between July 2008 and December 2009, 26 respondents were identified and recruited to participate in semi-structured qualitative interviews. Through our community engagement activities and participant observations, we identified and approached each of the local leaders to participate in this portion of the data collection; all leaders who we approached agreed to participate in the study. Follow-up phone calls and emails were made to confirm interview dates and times; at that time, three House leaders did not return our messages. Participants included at least one parent or leader from each of the fifteen Los Angeles Houses. The interview discussion guide was designed to gather in-depth information on the structural, social, and cultural characteristics of the Houses; challenges members experience in the House and Ball scene; perceived benefits of participation; House rules, activities, and communication; relationships within and outside the House; values, norms, and expectations related to HIV/STI risk behaviors; and receptiveness to interventions. Each interview lasted 1.5 to 2.5 hours and was digitally recorded, professionally transcribed and checked for accuracy. All interviews were conducted in the P3 project offices or at a convenient location . Respondents were provided a $45 incentive for completing the interview. --- Analysis The qualitative analysis for this manuscript utilized a "constant comparative" approach, an aspect of grounded theory that entails the simultaneous process of data collection, analysis and description. . In this process, data are analyzed for patterns and themes to discover the categories that are most salient, as well as any theoretical implications that may emerge. As the data are collected, they are immediately analyzed for patterns and themes, with the primary objective of discovering theory implicit in the data. Interview transcripts and observation notes/field notes were included in the analysis. Atlas.ti was used for coding and analysis of relationships between and within text segments. Interview transcripts and observation notes were imported into a single Atlas.ti database for analysis. Members of the research team reviewed an initial sample of interviews and field notes to identify key themes, which formed the basis of the project codebook. Codes focusing on a range of topics were identified and defined, based on the key constructs included in the discussion guide and observation form. The codebook was modified as needed, and once finalized, three members of the research team were responsible for coding the interviews. Inter-coder reliability was assessed through double coding a sample of approximately 15% of the interviews. Differences in coding were discussed and resolved by the team. The open coding process included refining codes based on the data. Codes related to leader expectations, challenges and risks experienced in the House and Ball scene, benefits to participation, and experiences with discrimination were included in the current analysis. This process led to the structure of the present study which: 1) describes the Los Angeles House and Ball communities based on participant observations at local events; 2) describes the perceived benefits and sources of sources of participating in the House and Ball community; and 3) the perceived risks that members experience from participating in the community. Throughout the paper, pseudonyms in the form of spices are used in lieu of House names to maintain confidentiality and protect respondents' identities. --- RESULTS --- Houses and Balls Data gathered from participant observations has provided a general description of the Los Angeles House and Ball communities. In general, we found that the House and Ball communities work in tandem to develop and support a community involving primarily African American and Latino individuals of diverse sexual and gender identities ranging from gay men to transgender women. Balls and Houses function as different entities, yet are dependent of each other in order for each community to sustain. Houses are different groups of individuals that compete against each other during Balls-underground events that reward individuals who win competitions focused around dance, athletics, and gender expression. It is important to note the gender and sexual fluidity of many of the House members and Ball participants. Observations consistently noted that it is not uncommon young men to wear women's attire and refer to each other with female pronouns while taking on stereotypically masculine gender roles during their normal, everyday lives. This is an integral aspect of the Ball scene given the few safe spaces AAYMSM and transgender youth typically are able to access. Although Balls have occurred in other parts of the United States since the 1920s, it was not until 1998 that the first Ball occurred in Los Angeles. House parents related that the history of the local Ball scene, revealing that the first House, originally named the Hoopla Divas, was created in 1996. From 1999 to the present, the Los Angeles Balls flourished, and during the study's data collection period, the number of Houses fluctuated between twelve and fifteen active Houses competing in the Ball scene. Most competitors of Balls are affiliated with a House, a group of young people who prepare to compete at Balls against other Houses. Houses are typically led by a House mother and/or father, and may have more than one mother and/or father in a single House-depending on House structure. Observations indicate that the number of members in Los Angeles Houses ranges from five to thirty-five. House structures vary and may include other roles besides father and mother, such as prince/ princess, godfather/godmother, emperor/empress, and ambassador. The roles and responsibilities of a House parent vary from House to House but typically involve: recruiting new members; maintaining the direction and status of the House; organizing Ball events and House meetings; grooming members for specific categories; acting as mentors for Ballroom success; and liaising with out of state chapters of the House. Los Angeles Balls are typically organized and sponsored by a single House in a range of different venues such as bars, warehouses, and community spaces. During the study period, we found that, Balls typically start at 2:00 am and can end as late as 8:00 AM, and are held once or twice per month during the "Ball season" . Balls are highlighted by competitions, which combine athletics, dance, and gender performance, where participants compete for cash prizes in front of a panel of judges comprised of House leaders. Balls are characterized by categories that focus on different aspects of a competitor's skills, endurance, athleticism, knowledge, confidence, and creativity. Observations and category lists prepared for each Ball indicate standards for each major category. For example, Fashion and Runway categories test a competitor's knowledge in current fashions and trends, as well as their ability to confidently walk down a runway in front of a panel and a room full of spectators. Realness categories focus on a person's ability to pass as a heterosexual person or a biological woman through appearance, demeanor, and attitude. Performance and Vogue categories test dance and athletic skills as participants battle each other on the runway by incorporating dance moves such as vogueing, pirouettes, flips, and the shablam, a move that requires a person to slam her or his body on the floor followed by an immediate lift off the ground. Face categories require competitors to have a well-groomed face to personify youth and beauty. Features such as clear skin, bone structure, trimmed hair, and straight, white teeth are lauded. Body and Sex Siren competitors are judged on their bodies, and competitors usually wear little clothing to emphasize muscular structure. Lastly, the Virgin category provides an opportunity for novice competitors who have not walked before to make themselves known in the Ball community. --- Study Sample The age of the House parents interviewed ranged from 22 to 42, with a mean age of 31. The majority are African American or Black , followed by Latino . Sixty-four percent were male and 36% are transgender . --- House and Ball Scene as a Supportive Environment Based on descriptions provided by the House parents, the House and Ball scene serves as a subculture that is able to provide a sanctuary where its members can freely express themselves creatively, artistically and sexually. House parents tended to view the Ball and House communities as a "safe haven" for its members. Parents often described instances of rejection from family or friends that their members experienced and felt that the House and Ball scene offered an attractive alternative to a "society who doesn't know how to deal with you or doesn't know how you feel". The Mother of Fennel described this safe haven as essentially being able to create "our very own family": Some of us come from backgrounds where we were abused or kicked out or not really appreciated or supported by our own family because of our journeys whatever they might be. So this was kind of an avenue to create that safe haven and to really be able to say, "I do belong to something. I am wanted." House parents frequently described scenarios that they or their members had faced dealing with discrimination based on their gender or sexuality. For those facing such challenges, the House and Ball scene offered an opportunity to be themselves, express their individuality and be creative. For AAMSM, finding a safe place to "support their norms" was particularly challenging, as Father Clove reported : White gay men have the opportunity to--economically had the opportunity to create geographies that are based upon their norms…[where] White gay men can shop, gym, bank, party, work all in the community that supports their norms. Black gay men oftentimes still live in the social minefields of the Black community where homophobia is huge. According to most parents, this support was something that was generally not afforded to them. Instead, parents reported that they and their children were often criticized, harassed, bullied or abused when expressing who they really were. The Father of the House of Cinnamon described his own abusive childhood, which he believed was related to his parents finding out that he was gay. He compared the House and Ball scene to the fantasy world of the X-Men comic books: For some reason America is not ready for X-Men. Now I use the term X-Men because we are a people who are outcasted because of our sexuality … Sometimes we feel like X-Men and I am Professor X rolling around on my wheelchair and there is this great school for creative kids with a gift of being gay. For those involved, the House and Ball scene is typically viewed as a type of refuge from the stigma and discrimination that they often face in the outside world. House parents, when describing the benefits of being involved in the House and Ball communities, generally cited: 1) the structure as it provides a sense of family and support; 2) a sense of acceptance; and 3) the validation and recognition that participants receive from others in the community. Though described as separate and unique systems, both the House structure and the Balls themselves were reported to provide these different benefits to members. --- House Structure: Family and Support The basic structure of the House, with a Mother and a Father naturally suggests a sense of family. This sense was referenced by each of the House parents, as they described their role as a Mother or Father. For some, being a parent was what they identified as their primary motivation for remaining in the House and Ball scene. When asked what made her want to remain in the community, one leader from the House of Coriander reported that "the love of my family … or the love of my House keeps me in it … I love them more than I love the Ball. It's the people, our family." This sense of love led several to also indicate that being a parent carried with it a certain level of responsibility or as one Mother put it, an "opportunity to contribute". At times, this opportunity was described as the ability to make a difference in a young person's life, and in some ways, repaying a debt to their own House Mother. This opportunity was described by the Mother of Wasabi as a sense of gratitude: I feel like a real mother because it's like gratitude. 'Oh my God, he had a good night, he won the trophy he wanted … or whatever. We all got together and became a family', not just for the Ball but to help each other out and that kind of stuff is rewarding. For many, it was this family structure that was the primary benefit of being a member of a House. The Father of the House of Cinnamon reported that when he first joined a House as an eighteen-year-old, he was immediately attracted to this family structure: I didn't quite understand it, but I knew at that point in time that I was engulfed in it. I saw the energy, the passion, the love, the sorrow. My father never told me he loved me. My mom barely was there …so coming into that type of family setting and people told you that 'Hey we care about you, we love you.' It made me feel like this is for me. Parents reported that the family structure provided House members support in the form of love, a place to stay when needed, and financial assistance. When necessary, Houses may replace biological families by providing support, guidance and a place to go when feeling alone or on holidays. Father Rosemary described his last Thanksgiving as including both biological family and his House children: "a lot of them [House members] won't really participate with their family, probably due to rejection and stuff. They prefer to come around the House families and have Thanksgiving and stuff." Parents acknowledged that this sense of togetherness was often set aside during the Balls themselves when House members often compete against each other for prizes. However, at the end of the night, parents and "kids" both agree that they are still a family. The Mother of Fennel summarized it as excitement, which she described as: …getting together, the competition, the enjoyment, getting around friends .. . being able to compete and then after the Ball is over everybody gets together and laughs .. .it kind of looks separate but in actuality it's like a tight knit family bond. Several parents reported being a member of different Houses before either founding their current House or rising to the rank of parent in their current House. In spite of this instability in membership, most reported feeling that their House was an integral part of their identity. I am always going to be a part of the House. When I am seventy years old, I am always going to be a [Star Anise]. It's not just a House, it's not just a family. It becomes a lifestyle…it helps me define myself. In addition to the House structure, several parents also mentioned that there are other informal family structures within the community which they termed as "personal" or "gay families". Members of a gay family may be from different Houses or may not be members of a House at all. Gay families within this community can be very complex, including mothers, fathers, aunts, uncles and siblings. Informal interviews conducted during observations indicate that these families are an additional source of support and may be more stable because their composition does not depend on competitions. Mother Ginseng explained her personal family as: I just, it's just like family that you actually weren't born into. You actually got to pick who is going to be your family and it just goes from having each other's back 100% and just knowing that if you are absolutely down and out and you just can't take it no more…or you need something it will pretty much be done or taken care of. --- Acceptance As previously discussed, House parents reported that many of the members of their Houses experienced discrimination in the form of racism and homophobia throughout their lives. This discrimination may have come from peers, neighbors, strangers, employers, or family members. Parents reported that some House members were forced out of their home when their family discovered their sexuality. For these members, discovering the House and Ball communities was seen as fortuitous as these communities were welcoming of many different types of people, making young gay, bisexual and transgender men and women feel accepted and at home. For some, this was the first community that was truly welcoming of their beliefs, talents, and gender/sexual identities. Becoming a part of this community was described as a "freeing experience", giving individuals a "sense of belonging" and providing a space to freely express themselves without being stigmatized. The Mother of the House of Cardamom described this as: It's like Cheers, it's like everybody knows your name…it's just a really, cool outlet where you can be appreciated by people who understand who you are or what it is that you do. This sense of belonging was the most common reason for people wanting to be a part of the Ball community. The Mother of the House of Saffron, a young transgender woman, related that she left home at eighteen, confused and fearful of the world, searching for some kind of acceptance. Finding the Ballroom scene helped her realize that there are "other people like me that were living this whole other life which to the 'normal world' was a mystery." This intrigued her, and she quickly joined a House where she was able to find the support and acceptance she had not found at home. Some of the older parents, those who had been a part of the original Los Angeles Houses, also spoke about the fact that the Ballroom community recognizes all types of people and has categories for different body types, talents, genders and styles. Some of them spoke about how this was one of the most unique aspects of the community, in that it was constantly evolving, adding new categories to recognize and including different types of people. In this way, the Ballroom scene has the ability to adapt to changes in fashion and participants -making it a more permanent fixture in the underground gay community. The Father of the House of Clove described this aspect as one of the reasons people gravitate to the scene, as someone may finally identify a place where "I can be me. I am in a place where I can fit in because they have different categories that I can fit into and a place that is actually going to award me." --- Validation and Recognition Similar to finding acceptance, many of the parents spoke about how the House and Ball communities can provide a sense of validation through Ball competitions and individual contributions to the House. This validation and recognition is most prevalent in the status system of the Ball scene, which recognizes those performers who have made a name for themselves in the community. Before the beginning of each Ball, individuals are called out to "walk" the runway during the "Stars, Statements, and Legends" 1 . Younger members who have just begun to consistently win and are known in the scene, are deemed "Stars"; those with more experience and have "memorable moments" at major competitions are called "Statements" and only those who have excelled in a particular category and have been involved in the scene for a number of years can be deemed "Legends". To achieve this status one must not only win, but also gain the respect of others in the community through their leadership and visibility. The Mother of the House of Ginseng, a young man who, at the age of nineteen had already achieved a high status as a "Butch Queen in Drags" explained: Everybody's objective is to become legendary and so until you become a legend or a person known in the Ballroom scene and respected, then it kind of makes you want to keep going and push for that status. Apart from gaining status in the community, some parents felt that individuals' contributions to the House were also a source of validation and recognition. In these instances, helping to plan a Ball or a group performance category at a Ball was a way to achieve recognition from parents and other leaders. In addition, "turning" a good performance, regardless of whether they won, was also an opportunity to attain validation within the House, as one leader explained: "it's always nice to hear your dad go, 'Miss Thing you did that … you turned that.' That always feels good when you have somebody that you look up to giving you validation." For some, this validation was something that they had strived to achieve for a number of years in their everyday "real world". Achieving this in the mainstream culture was not necessarily possible, given the challenges that many experienced in the form of homophobia 1 Stars, Statements, and Legends is held before every Ball to recognize significant figures in the House and Ball community. and discrimination. Father Cinnamon related that he was invited to join a House when he was in high school while working at McDonalds. He felt that joining a House and competing in the Ball scene helped him realize that he could achieve something in life, something his biological family had never provided for him. Well my story is unlike any others due to the fact my Ballroom scene started before I entered the Ballroom, and what I mean by that in terms of wanting to be accepted in society and looked upon as something greater than myself. So the Ballroom scene was something that was inside of me for a long time, and it developed at an early age…. As it began to develop, I began to like myself, because as a child I really didn't like myself… I was always told I could not accomplish anything. I was going to be a failure, that I would be gay…as I approached the Ballroom scene, it made me feel like I was important. It gave me some insight to a whole brand new society that I was willing to just jump into. The sense of recognition and validation was something that was seen as highly valuedprimarily because it was something that could not be achieved in the outside world. Similarly, validation achieved through expressing one's true self -for which many of the communities' members had been rejected or devalued -was perhaps even more valuable, because the House and Ball scene is the only venue where this recognition could be achieved. It's about being accepted in a society in which you can delve in and take it as far as you possibly can…You can express it though your sex. You can express it through the beauty of your face. It is our--it is our awards, our--it's like being at Kodak Theater. Being at a Ball for us, is like being--getting a Grammy that night. So, that's what it is for us being able to express yourself in a way in which nobody else would see it vastly or remotely acceptable in a society of today. --- Challenges Encountered in the House and Ball Scene While describing the many benefits of the House and Ball communities, parents also acknowledged that the Ball scene can have a number of challenges for AAYMSM and other sexual minority youth. Many of the challenges noted by House parents can increase members' risk for HIV and other sexually transmitted infections . Interestingly, these challenges were often directly related to some of the supportive aspects the parents also cited, such as the idea of a "safe haven" free from rejection and discrimination, and the creative outlet the Balls provide. In general, the most common negative aspects of the Ballroom scene identified by parents included: 1) members' struggle to maintain status in the Ballroom scene, which at times led to violence, sex work or drug use; 2) the danger of becoming too involved in the Ball community, inhibiting one's ability to effectively manage in the real world; and 3) the perception and stigma of the Ballroom scene within the larger gay community. --- Maintaining Ballroom Status As previously discussed, members typically strive to attain a high level of status in the scene, which in turn provides a greater level of recognition. While this level of status was viewed as a positive aspect, the path to achieving and maintaining this status had a number of challenges. Observations documented a number of altercations that occurred at Balls or after Balls that often resulted in members from different Houses engaging in a physical fight that could include weapons such as knives or mace. These fights usually occurred after a perceived unfair judging decision or other issue on the runway. Parents also reported that these fights occurred due to members "disrespecting" others by stating they are "better or that they can accomplish more. It's like a tug-a-war of respect." It was often the case that the manner in which these instances of disrespect were handled was to start a fight with the other party -generally as a means to defend one's status in the scene. The former Mother of Rosehips related how one of her kids handled a situation where he was continually "chopped"2 from his category because one judge did not want her son competing against his House member. And [House Member} went home … put all his knives in his bag and came back to the Ball and he just went-the Ball just went crazy and that was it. It was over because he was tired of them trying him when he knows that he went out and did that. .. it was like earning your respect. Achieving status in the Ballroom scene can also be extremely expensive as travel and designer clothes or "labels" are often required. As many parents related, Los Angeles will always be considered "late" to the cities on the East Coast -meaning that Los Angeles was late to the scene and will always be catching up to the trends of Houses and performers on the East Coast. Therefore, for a House member to achieve a higher status, he or she should also compete and win on the East Coast. In addition, some categories such as European/ American Runway, Foot and Eye, and Labels require walkers to have new and au courant designer clothes and accessories. Parents reported that being able to afford travel and clothing was extremely difficult for most members. Therefore, illegal and often risky activities are to obtain the necessary items. For example, parents reported that it was not unusual for kids to make Ball-related purchases through credit card fraud. Perhaps most relevant when considering HIV prevention, is the practice of turning to sex work or "escorting" for money to pay for travel and clothes. Parents agreed that escorting was a major problem in the Ball scene and had increased in recent years, attributing this increase mostly to the ease of soliciting sex online. In the past, escorting was thought to be limited to transgenders or "femme queens" as they needed money to pay for hormones and surgery. Parents reported a recent shift in this behavior as Father Ginseng reported: It's been over the last five to ten years that it has just exploded … These guys can make easy money. Most of these guys are pretty young… They--somebody gives you 50 dollars for this or 100 dollars for that, it's easy. It's-five or ten clients within a day and you make a lot of money but at what price or at what cost? Most parents acknowledged that members of their House were involved in escorting. Some parents said that they have spoken with their kids about it, warning them of the dangers. However, parents also reported that they could not necessarily change their behavior, they could only let their kids know they would help them if they needed it. Parents also felt that some of the pressures within the Ballroom scene contributed to substance use. For example, Balls typically start between 2:00 and 3:00 AM and can last until 8:00 or 9:00 in the morning. Some parents felt that needing enough energy to compete encouraged members to take different stimulants or club drugs, such as cocaine and ecstasy. Others reported that use of these substances was particularly common among those individuals who walk in some of the performance categories, as the drugs can numb the pain from performing "shablams" or other moves. Father Rosemary explained that these substances may also lessen the anxiety performers might have about hurting themselves: "Well they got to do a lot of extra performance stuff when they think they are going to go out and break their neck." Most of the parents had very strong opinions about drug use within the scene, with several of the original Los Angeles House members relating how drug use had torn apart one of the original Houses and created a huge riff in the community. The Mother of the House of Chile, who was a member of one of the original Los Angeles Houses, reported that her memories of that time keep her from using those drugs: Like I said, dealing with it in the past and watching the Cumin House makes me harder on it [I: So, that really did tear up the Cumin House?] Oh, yes. These people were--they were exceptional and it's like a lot of people just deteriorated, like some of the best or the most gorgeous Face kids, all that. When you look now, you are only seeing remainders of them. These parents voiced concern about the current amount of drug use in the scene and felt that it was a growing problem; several also they feared the increase in escorting was related to the increase in drug use also: I think that we are going to see--I think in five years we are going to see the effects of escorting… higher drug use because people have to get high to deal with what they are doing or to stomach what they have done. So, I think the drug epidemic is definitely going to go up .. .it's cancerous. It's going to deteriorate a little bit more. --- The "Normal" and "Fantasy" Worlds Several parents believed that one of the allures of the Ball scene was the "unreal" or "fantasy" nature of the community. Like some other subcultures, the Ballroom community allows its members to create an "alter ego" complete with a House name and persona. However, parents like the Mother of Saffron explained that an individual in the Ballroom scene must be able to differentiate between their two realities: I still do live in this world which is very hateful towards people like myself, and I have to know how to differentiate between reality and fantasy. People don't live in the Ballroom scene, they get caught up in it, into the hype. This "unreal" world serves as an escape for many who enter the scene. While this was considered a benefit of the scene, as it does offer a sense of belonging and acceptance, it could also lead to a number of risks as the Mother Rosehips warned: I mean, the Ballroom scene consists of a world, when you go in there you can be all that you want to be, and when you walk outside that door you are back homeless. You are back broke. Your life is the same as it was before you walked inside those doors. It's just a world of-a world of--a fairy tale world. The risks parents associated with the Ball scene were seen as even more problematic for young men who become too involved in the scene and drop out of school or cannot keep a job. About half of the Los Angeles Houses have rules that specifically state that all members must either be in school or have a job. This rule exists for a reason, as many parents felt it keeps their members more grounded in reality. The Mother of the House of Licorice reported that while there is not a strict rule in his House, he stresses the importance of having a real and meaningful life outside the Ballroom: A lot of people sometimes get caught up in the fantasy and forget that there are other things besides the Ball scene. That's why I try to remember to keep the kids in school or work. You got to do something outside of the Ball scene because most will sit and wait for the next Ball … and not try to do anything else with their life. Then when it comes, they would be like, 'I don't have any money. ' Several parents also linked this issue of being too wrapped up in the scene to escorting, believing that those involved may be doing so because they did not graduate from school and/or cannot find a job. They worried about what would happen to those members as they got older, and escorting is no longer a viable option "The problem is escorting is a youthful event. The older you get, the less work. So the thing is, what happens when you are no longer escorting … All that time is gone, you don't have a skill". --- Outsider Perspective of the House and Ball Scene While members of the House and Ball communities are continually striving to achieve status within the scene, some parents did acknowledge that this status was worth very little outside the scene. In general, parents and others feel that individuals outside the Ballroom community view the scene and those in it as "messy", meaning gossipy, unmanageable and/ or prone to start a fight. Several mentioned that even within the larger gay community, Ballroom kids are stigmatized for their membership and seen as "just a bunch of [gays] showing off, it's just for vanity and there's nothing else behind it". Some reported that they do not talk about the scene with outsiders, particularly romantic interests, and "hide my trophies" when people come to their homes. The Father of the House of Chile reported that his biological mother did not want him to be involved in the Ball scene because "she thought I was going to change and start doing the strange drugs … neglecting bills, school". Father Clove, who had previously worked in an HIV prevention organization, felt that these stereotypes had bigger implications, believing that service organizations had historically stayed away because of negative perceptions associated with the Ballroom scene. "The House and Ball community is a community of people who [are] riff raff, in their minds. Who engage in malfeasant behaviors, all they [do] is craft or booze or steal." This perception was confirmed through observations of House and Alliance meetings where members commented on the lack of consistent service providers' presence in the community over the last several years. --- DISCUSSION The data presented in this study provide a description of the House and Ball communities and how the structure and activities within this subculture offers both support and risk to its members. Parent descriptions indicate that the ability to be oneself was freeing and gave respondents a sense of respite from the outside world. These data indicate that members of the House and Ball communities are, like members of other subcultures , savvy cultural actors who use the cultural resources available to them in creative and interesting ways to make their lives more tolerable or exciting. Like other subcultures, the House and Ball scene provides a safe and supportive environment for those who may not feel that they fit in with the larger society. However, the very factors that make the House and Ball communities an appealing environment for its members may also present a number of risks for those who become enmeshed in its surroundings. The House and Ball communities provide a sense of support for AAYMSM and other sexual minorities, primarily through the House structure which provides members with a sense of family. African Americans have been shown to seek out informal support for advice or help to reduce negative consequences of stress . Similarly, there is a long tradition of ethnographic research that documents the "fictive kin" networks among African Americans as a means to extend one's social network, which can serve as additional social and economic resources to manage societal challenges including poverty, discrimination, and rejection . Similar alternative family structures or fictive kin networks have been identified also among other sexual minority groups in general and AAMSM specifically. This support can be crucial to the health and well-being of AAYMSM who have experienced rejection or discrimination from their family, peers, the larger African American community, and society in general. This rejection and discrimination can in turn contribute to the negative health outcomes of discrimination and homophobia including psychological distress, substance use and high-risk sex use . Related to the support AAYMSM typically receive from their House family, members of the House and Ball scene also gain recognition and validation through their performance and achievement in walking the many different categories at Balls. Parents expressed that what was most important about this validation and recognition was that members are able to gain this simply through expressing themselves, something that may have been repressed or seen as taboo in the world outside the Ballroom. Finding a space that embraces and celebrates the uniqueness of its members was seen as invaluable to House parents. Similar findings have been noted with other YMSM populations, noting the lack of "safe spaces" for YMSM and the need for more spaces where young men can be themselves, grow as individuals, and feel accepted in their community . However, as observed at a number of events, this need to gain or retain status within the community was often related to physical altercations with others; ultimately contributing to the negative perception of the Ballroom scene to those outside the community. Parents also mentioned that obtaining validation and recognition in the Ball scene can make it difficult to succeed in larger society. Parents described members who are admired and celebrated during a Ball, but still have to walk outside and deal with the reality of the outside world -a world that may have a very negative perception of the Ballroom scene. As parents described, discovering that this acceptance and recognition is only valid within a community that is not respected in the larger world, may lead members to become too involved in the Ballroom scene. Ultimately, parents felt that this blurring of realities between the "fairy tale" of the Ballroom and mainstream society may contribute to members being unable to find work or remain in school, or become involved in escorting. Finally, the results from this analysis indicate that parents feel that in order to obtain and maintain status in the Ball community, there is a great deal of pressure for members to "get their tens"3 and successfully compete at Balls. To do this, participants may engage in illegal or risky activities to travel to out-of-town Balls and/or purchase designer clothes for their chosen category. Sex work, or escorting, was identified as a particular problem within the Ballroom community, with parents estimating that up to 70% of Ballroom "kids" are involved in escorting. Several attributed this increase in escorting to the ease of the Internet, stating that members no longer have to look for "tricks" on the street. Rather, they can post a profile online, allowing for greater concealment of the sex exchange, and also making this activity appear to be less unseemly, a phenomenon described by others . Condom use is typically lower among sex workers, and sex work can increase one's risk for exposure to violence and substance use . Research among MSM sex workers indicates similar findings, particularly among MSM of color . There are several limitations to this study. The analyses rely on the perceptions of parents and leaders in the House and Ball community in one city. This analysis does not include the members' perceptions of the communities; future analyses should investigate whether members' perceptions are comparable to the House parents. These analyses are based on perceptions from one Ballroom community and may not be generalizable to communities in other cities; other cities' Ballroom communities include larger proportions of Latino and Latina members and future studies may want o investigate potential cultural or ethnic differences in these cities. For HIV prevention efforts, future studies should include comparisons to other cities to identify other mechanisms of support or risk within the scene. However, in spite of these limitations, we feel the data presented here are important and provide ideas for providers on leveraging the support systems available in the House and Ball scene so they can develop and/or advocate for new programs designed for House and Ballroom communities. Little is known about these communities, and this study provides an important foundation for understanding the social and structural characteristics of a Ballroom community, a community whose members have previously been identified as being at-risk for HIV infection . The findings presented here have a number of implications for practice. Families are often mentioned as integral to HIV prevention efforts . This study builds on what others have learned about the family structure of both sexual minorities and African Americans, in that families are not limited to biological members and can encompass a range of individuals. Service providers and interventionists should ensure that, when working with AAYMSM, the term family is broadly defined, and clients should be made to feel comfortable in including a House parent or "personal/gay family" member in HIV prevention work. This follows on the recommendations of the National Institute of Mental Health which created guidelines for measuring and defining the family . Non-biological family members can be powerful sources of support and should be considered when developing HIV prevention programs. House leaders unanimously identified sex work as one of the most pressing problems in their community. The ease of soliciting sex on the Internet has reportedly made this practice common for members of all genders and sexual orientations in the House and Ball scene. Providers working with these communities should be aware that this is becoming an increasingly common problem and should consider incorporating harm reduction programs into their service array. For example, education and prevention programs can provide those who are involved in sex work with information that may reduce HIV-risk behaviors. Empowerment programs can help to increase the self-esteem and mental health of those involved in sex exchange to increase their ability to advocate for their own safety and sexual health . Finally, the House and Ball scene offers many healthful aspects that can be leveraged as part of an intervention effort. In particular, the scene offers a unique, supportive environment for its members to freely express themselves. Prevention programs designed for this community should consider building on these strengths. For example, programs can train current leaders in HIV prevention so that information is disseminated through the House networks. The creative side of the House and Ball scene can also be used in prevention work. Designing programs that encourage performance and other activities may be more appealing for this target population, while also recognizing that it is possible to maintain creative expression outside the Ball scene. Additionally, the practice may provide opportunities for members to express larger social issues and identify new solutions for managing these experiences. This practice has been used by other social justice activists to overcome oppression . Finally, it is important for programs to emphasize the importance of maintaining a life outside the Ballroom scene. This can be accomplished through referrals or engaging guest speakers to provide outside perspectives on how to attain recognition and validation through other outlets. Using a community-engaged approach to accomplish this work was integral to the success of the project. Our relationships with the House leaders granted us access to formal and informal events within the communities, providing a more inside perspective of the communities. For others wanting to work with the House and Ball scene, it is important to include the leaders of the communities in the planning and implementation of the project. Giving the leaders a voice in developing projects will likely result in programs that are better grounded in the community's needs and are well received within the community.

This paper describes a community-engaged study with the Los Angeles House and Ball scene, in which the perspectives of the leaders of these communities are captured to better understand how the House and Ball communities may protect and/or increase its members' risks for HIV infection. Data were collected through in-depth interviews with House parents (N=26). This study identified key features of both support (e.g., family and support; acceptance; validation and recognition) and risk (e.g., members' struggle to maintain status in the Ballroom scene; sex work; substance use; danger of becoming too involved in the Ball community; perception and stigma of Ballroom scene within the larger gay community) within these communities. Findings are discussed in relation to framing how to leverage the supportive aspects of the House and Ball communities to design relevant HIV prevention interventions.

Introduction Technologies and applications evolve to create new eco-systems of heterogeneous and distributed services that are available for people anytime and anywhere. Nowadays, your environment may be plenty of services that support your daily life: services that track your activity through the mobile phone, that allow you to do an efficient use of your home heating and lighting, that allow you to interact with social networks, that provide you with the weather forecast or traffic status in real time, and so on. Although these services can be used individually, it is their composed usage what has the potential to create new value-added services for end users. In addition, in a world where end users play a more and more important role in the development of content, it makes sense to think about the possibility of end users creating new services by composing existing ones. By upgrading end users to prosumers and involving them in the process of service creation, both service consumers and service providers can benefit from a cheaper, faster, and better service provisioning [1]. Currently, there exist a myriad of end-user environments that face this composition challenge . However, only a few of them consider an aspect that end users demand on current software solutions: social support. Nowadays, millions of people use social networks such as Facebook or Twitter to share with others what is happening in their lives. Messages, images, videos or links are continuously spread through the Internet in order to make people feel that they are connected to others. In the same way, social networks allow people to share their relationships: who are their friends and how many they have, who are their relatives, whether or not they have a sentimental relationship, and so on. Even more, they also share relations with things they have or like. We can find applications that allow people to share books, products, car journeys, homes, etc. So, considering this scenario, why not to share also service compositions with other users through social networks? Indeed, we think that a social structure created specifically to support end users in the composition of services can introduce several benefits. First, it can be a valuable mechanism to facilitate end users to discover services instead of relying on typical internet discovery solutions that cannot scale to the increasing amount of available services . For instance, social relationships can be used to allows end users to browse services within a structure they perfectly know ; and they can be also exploited to make service recommendations based on friends' interests. Second, a social network can help end users to share knowledge with other end users to improve their skills in composing services. Finally, it can also be a collaboration space among end users and developers, allowing the combination of the innovation and creativity of end users with the expertise of developers. Considering the motivation presented above, the problem that this work tries to improve can be stated by the following two research questions: -How can we define a social structure that captures the intrinsic characteristics of the composition of services by end users? -How can this social structure be exploited to support end users in the composition of services? The main contributions of this work have been developed to answer the research questions presented above: -We propose a graph-based definition of a social structure that characterizes the activity of composing services by end users. -We analyze how the underlying connections that are defined in the social structure can be exploited to both help end users to discover services by browsing social structure's connections; and recommend services to end users during the composition activity. As proof of concept, we implement the proposed social network in the context of EUCalipTool, a mobile end-user environment for composing services [5]. This implementation is evaluated through an experiment with end users. The rest of the paper is organized as follows: Section 2 presents the related work. Section 3 introduces some foundations about the service composition by end users and an intuitive characterization of the proposed social network. Section 4 presents the definition of the social network with graph theory. Section 5 analyses the defined social structure to delimit how it can be used to browse and recommend services. Section 6 presents the implementation of the social version of EUCalipTool as a proof of concept. Section 7 introduces the evaluation of our work. A discussion is presented in Section 8. Section 9 finishes the paper with some conclusions and the analysis of further work. --- Related Work Some works face the challenge of supporting end users in the creation of service compositions with solutions that include some social issues. These social enduser composition approaches can be classified into: community oriented approaches, which introduce features to share the knowledge that is produced within a community; and social network oriented approaches, which are those that create social structures to connect users with common interests. Next, we introduce the related works of each type. Note that many other works allow end users to compose services. However, we have only considered those that include some social issue, single end-user oriented approaches have been omitted. --- Community-oriented approaches As introduced above, approaches in this category introduce some features to share knowledge within a community of users. These features generally consist of mechanisms to annotate, rank or rate compositions as well as recommendation capabilities. However, there is not a social structure underlying the proposed solutions. One of the most significant solutions in this category is IFTTT [9]. It is based on trigger-action programming to provide a complete software platform that allows end users to create what they call Applets , which connect apps, devices and services to trigger one or more automation rules. This platform recommends users with Applets that are created by the community in order to be used and rated, allowing also sharing them by sending their URL through different apps installed in the device . The main difference with our approach is that Applets are explicitly shared among one user and another sending their URL. There is not social structure that allows end users to follow others in order to browse the Applets they create. In addition, our approach uses the connections created in the social structure to recommend the composition of services done by others according to similar interests. Dlvr.it [10] is a web tool that allows end users to aggregate data feeds from different websites, sort and filter out those feeds, as well as mash them up into a single feed. It supports end users in the sharing of their feed aggregations through existing social networks such as Facebook or Twitter. In contrast to our approach, this platform is based on social networks of generalpurpose to share aggregation of data feeds. Our work proposes a social network specifically designed to support the activity of service creation which allows exploiting the generated connections to improve the discover of services and their recommendation. Zapier [11] is a web-environment that allows end user to create their own applications by composing conditions and actions that are already predefined by professional programmers of several vendors. It allows users to define interests on specific apps, and recommend Zaps of other users that include these apps. The recommended Zaps can be taken as a basis to create new ones. The main difference with our work is that a structure to socialize the creation of Zaps is not proposed. Thus, users cannot share Zaps among them. This work shares with our proposal the possibility of making recommendations based on interests. However, these recommendations seem to be supported by a typical content-based algorithm, i.e. they recommend the content that is similar to the content that a user has already consumed . Our approach goes a step beyond and recommend services that may be of interest to a user by considering the similarity among this user and others. This similarity is calculated by analyzing the connections that are implicitly created in the proposed social structure. Node-RED [12] is a visual wiring tool to create action composition in an Internet of Things context. It provides a graphical web environment where a user can define nodes that represent actions over IoT devices, and connect these nodes in order to create a sequence. It provides a flow library to share compositions with other users in JSON format. Their efforts to allow users to share composition among them are limited to creating a repository from which users can download compositions created by others. However, there is not a structure that facilitates to follow users and share composition with them. In the same way, there is not support to make recommendations based on the affinity among users. --- Social network approaches There are a few solutions that specifically focus their efforts on proposing a social network structure in order to improve the composition of services by end users. One of the exceptions is [13] which presents the foundations to support a social-awareness web service composition and focuses on exploiting the social data to make recommendations. However, authors only use the frequency that a user includes a service in their compositions in order to exploit social data. We propose a social structure with a richer set of relationships. Soriano et al. [14] proposed an inception idea for socializing the composition of services through a user-generated catalog of resources founded on the Web 2.0 vision for user co-production and harnessing of collective intelligence. However, a social structure is not proposed. Jiang et al [15] study the idea of social manufacturing and propose a preliminary solution to achieve that production service providers and prosumers collaborate for production. This work analyzes how specific production services can be shared in order to allow prosumers to use them and create new products. However, the creation of new production services is not considered from a social perspective. Other works use the concept of social network or a similar one in the area of service science, although they are not focused on end-user development. Similar to our idea, Tamburri et al. [16] introduces the main pillars to create a social network of developers and shows the potentials of this idea in practice, by reporting on its application to a real-life industrial scenario. Maamar et al. [17] introduce a social network of web services focusing on three types of relations: recommendation, similarity, and collaboration. Yu & Woodard [18] create a mashups API affiliation network by analyzing the APIs that were used to create each mashup. This study concludes that the web APIs are used in the creation of mashups by following a long-tail distribution. Chen et al. [19] construct a global social service network and provide generic quality criteria for social links which included dependency satisfaction rate, QoS preference, sociability preference, and preferential service connectivity in order to improve the quality of service management. Ren et al. [20] describe a service social network and five kinds of relationships, namely interactive transaction, co-community, physical distance, resource-related, and social similarity relationship. They discuss how these relationships can create a synergy effect, and develop a service selection model that support social collaboration between services. In the area of Social Internet of Things, we can find several works that focus their efforts on the idea of converging social networks with IoT. For instance, Kranz et al. [21] analyse the implications of a so-called 'social-technical network'; the concept of Blogject is presented in [22]; Atzori et al. [23] study the participation of smart objects in current social networks. Guinard et al. [24] propose the use of existing social networks such as Facebook to share data produced by services. Although all these works are not focused on the composition of services, they have done an excellent job in the analysis of social relationships between intelligent things, which has inspired us in the definition of this work. Following the path of these works, Meissa & Benharzallah [25] analyze social relations to determine the main challenges for the composition of services in Social IoT contexts. In the context of Cyber Physical Social Systems , Wang et al. [26] present an approach to dynamically create service composition. They intro-duce an algorithm based on QoS fluctuation computation and Skyline component computation that considers QoS constraints in order to satisfy users' requirements. Although not focused on service composition, social networks have been also considered in the area of end-user development. Reuter et al. [27] propose a EUD environment to analyze social network big data. This environment includes a social media API and a quality assessment service as well as a web application targeted at end users. Massa & Sapano [28] introduce FaceMashup, an end -user development environment supporting the manipulation of the Facebook graph and allowing end users to analyze their social data. --- Socializing the Composition of Services According to Boyd & Ellison [29], a social network allows individuals to construct a public or semi-public profile within a bounded system and articulates a list of connections that can be viewed and traversed. Considering this general description, we introduce some foundations on service composition by end users that helped us to intuitively characterize the social network that is proposed in this work. Characterization 1. Hung et al. [30] define a web service as an autonomous unit of application logic that provides either some business functionality or information to other applications through an Internet connection. In Service-Oriented Computing , developers use services as fundamental elements in their application-development processes. Service composition accelerates rapid application development, service reuse, and complex service consummation [31]. In this way, developers can solve complex business problems by combining and ordering available basic services to best suit their problem requirement. Thus, the social network should consider services of two types: Basic Services, which are executable logic units implemented by programming activities; and Composed Services, which are created through the composition of other services. In this sense, a service profile of the social network should differentiate between these two types of services. In addition, there should exist a connection among services that indicates that one Composed Service includes another service in its definition. Characterization 2. There exist several solutions to help developers in the composition of services [32]. However, the explosion of the number of web services and APIs exposed through the Web has accentuated the need for allowing end users to create their service compositions [1]. We can find a myriad of environments focused on allowing end users to compose services by their own . Thus, the social network should consider two types of users . End users should have the possibility of creating Composed Services while Developers should be authors of both Basic and Composed Services. In this sense, the social network should include a connection between a user and a service that indicates that this user is the author of the service. Regarding the user profiles, the social network should differentiate between End User and Developer. In addition, inspired by existing social networks, we introduce the possibility of creating a connection between two users in order to define the interest of a user in the composition activity of another one. Characterization 3. In the context of end-user development [33] it is well known that end users have many difficulties to create solutions from scratch. It is a good practice to provide predefined elements that can be taken as a basis to define new ones [34]. The composition of services is not an exception. Environments that support the composition of service by end users usually provide them with predefined compositions to facilitate the creation of new ones . In this sense, a Composed Service should be taken as a basis to create a new Composed Service. Thus, the social network should include a connection that indicates that one Composed Service has been created from another one. Characterization 4. Currently, the execution of services does not depends only on the computer where they are deployed but also on the context they are being executed . For instance, the rising of the Internet of Thing paradigm has introduced new services that depend on mobile devices or wearables and they must go with the user in order to provide their service wherever the user is [36]. Other services, however, are highly coupled with the physical environment where they are executed. For instance, smart buildings [37] provide services to control lighting, temperature or the doors of specific locations. Thus, a service can be characterized, among other contextual data, by the devices required for its execution, and the location they are coupled to . In this sense, a service profile should include this data. Characterization 5. Independently of how end users create a composition of services, from scratch or taking as a basis a predefined one, they need to find, select and include the services they need. According to the analysis done in the previous section, some approaches introduce mechanisms based on recommendations to help end users to discover the service they need. In the context of recommendation systems, one of the most used techniques is based on tags or keywords [38]. Thus, the social network should provide mechanisms to allow a tag-based recommendation of services in order to help users to find those that fit their interest. Thus, a service profile should include a set of semantic keywords that describe its execution logic and can be used to make recommendations. In the same way, a user profile should include a set of tags that describe service domains in which a user is interested in. Characterization 6. Finally, according to Boyd & Ellison [29] user profiles generally include data that can be used to characterize and "know" the user. In addition to the data introduced above , we consider that a name, one photo and a textual description is enough data to create a user profile for our purpose. In the same way, based on some existing service profiles [39][40] [41][42], in addition to the data introduced above , we think a service profile should include a name, a textual description, inputs required to execute a service, and outputs obtained after the execution. --- Social Network Definition In this section, we present a semi-formal description of the profiles and connections that define the social network that we propose to support end users in the composition of services. Social networks are usually represented as graphs . We used a directed, typed, constrained and attributed graph [43] [44]. In particular, we define the proposed social network with the following graph-based specification: -There exists a set of vertices V and a set of directed edges E. -Some constraints are defined over types of edges. For instance, a f ollower edge represents the interest of a user A on the service composition activity done by a user B. Thus, they can only be defined between two users. ∀ e ∈ E | type E = f ollower → type V ) = U ser & type v ) = U ser The rest of constrains have been omitted in order to not overload the paper. They are defined analogously and can be found in [45]. -A Composed Service is defined as a set of includes edges. ∀s ∈ V | type V = Service & subtype S = Composed → s = {e 1 , e 2 , ..e N } & ∀e ∈ s → e ∈ E & type E = includes -To represent the order in which services are included in a Composed Service we complement each Composed Service the relation < that defines the order among elements according to the edge creation time. Thus, e1 < e2 specifies that e1 was created before e2. The function definition returns the ordered set associated with each Composed Service. Thus, the definition of a Composed Service cs i is defined by as an ordered set: If we need to consider more complex compositions such as those done with, for instance, BPMN, a Composed Service would also include other elements such as conditions, loops or parallel executions. This could be done including additional types of vertex and edges that represent these elements and the relationships among them. However, this is out of the scope of this paper. def inition = {e 1 < e 2 < -There is a set of data vertices V D and a set of vertex-attribute edges E V A together with the source V A and target V A functions that indicate the source and target vertex of each vertex-attribute edges. -We include a set of types for vertex-attribute edges T V AE = {name, location, interest, description, input, ouput, semantics, device} and a function type V AE which indicates the type associated to a vertex-attribute edge. ∀ eae ∈ E V A | type V AE = input → type V ) = Service & target ∈ V D The rest of constraints have been omitted to not overload the paper. They can be found in [45]. In order to better understand the proposed social structure Figure 1 shows its definition in a UML Class Diagram. As we can see, there are Users and Services with their properties. There are two types of users: Developers and End Users. There are two types of Services: Basic and Composed. All users can be followers of other users and consumers of any type of service. Developers can be authors of any type of service. End users can be authors of Composed Services. A Composed Service includes services of any type, and there is a Previous association class that, given the services included in a Composed Service, indicates the service that is previous to another. A Composed Service can be defined from another Composed Service. --- Exploiting social network's connections to browse and discover services The main goal of the above-introduced social network is supporting end users in the composition of services. One of the most interesting features that it provides is the possibility of exploiting the set of connections that are created to improve the problem of service discovery. The proposed social structure can be used to allow end users to discover services by browsing profiles, but also to recommend end users with services during the composition process. Efficiently supporting end users in exploring services created by other users is a key requirement of the proposed social network. Exploring services is the process of navigating through available services and acquiring important knowledge of them [1] and it heavily relies on how the characteristics of services are represented, organized, and rendered. With the proposed social network, we provide end users with a tool that allows them to find services by browsing a set of users and services profiles. Social networks are currently one of the most used mobile apps [8] and the use of a social structure is familiar to end users, which may facilitate the task of finding services in comparison to current service repositories. Thus, considering the social structure presented above, a given user can: 1. Browse its f ollower connections. The user can access the list of its followed users and the data attributes that describe each of them . To describe this, we propose de function Followed Users : F U = {u j ∈ V | type V = U ser & ∃ e ∈ E | type E = f ollower & source = u i & target = u j } 2. Browse the author connections of each followed user. The user can access the services created by each followed user and the set of data attributes that describe each of them . To describe this, we propose the function Followed Created Services : F CS = {s ∈ V | type V = Service & ∃ e ∈ E | type E = author & source = u j & target = s & u j ∈ F U } 3. Browse the location, semantics, device, includes, and def inedF rom connections of each service created by a followed user. Thus, given a specific service profile S, a user can browse: The services that are associated with the same location as S. The services that share some semantic keyword with S. The services that depend on a same device as S. The services that include S in their definition. The services that have been defined taking S as a basis. A formal definition of these browsing options can be found in [45]. As representative example, we present the function Same Parent Services , that define the browsing option 3e: SP S = {s j ∈ V | typev = Service & ∃ e j ∈ E | type E = def inedF rom & source = s j & target = s i & ∃ e k ∈ E | type E = author & target = s j & source ∈ F U } Section 6.2 shows the implementation of these browsing options and their practical usage in the context of EUCalipTool. --- Recommending services in composition activities As commented above, end-user environments usually provide the possibility of composing services in two ways: from scratch and from a predefined composition. We focus on analyzing how social network's connections can be used to recommend services to end users in these two situations: 1. End users are creating a Composed Service from scratch. In this case, we want to provide end users with a list of services that can fit their needs before starting the composition of a new service. 2. End users want to compose a service from an existing Composed Service as a basis. In this case, we want to recommend the Composed Services that better fit end users' interests. In addition to these two situations, service recommendations can also be provided at any time during the composition process, independently of the way end users create a composition: 3. End users have a Composed Service partially defined. In this case, recommendations are focused on providing a list of the most suitable services to be included after the last service included in the Composed Service. In order to make recommendations in these three situations we propose the following functions: Composed Services . This function returns the services composed by a user u i . --- CS(u i ) = {s ∈ V |type V = Service & ∃ e ∈ E |type e = author & source = u i & target = s & = Composed} Inclusion Number . This function returns the number of times that a user u i has included a service s i when composing another service. IN =| {e i ∈ E | type E = inclusion & target = s i & ∃ e j ∈ E | type E = author&target = source & source = u i )} | Global Inclusion Number . This function returns the total amount of services included by a user u i . GIN = N k=1 IN Service Interest Level . This function returns the level of interest that a user u i has in a service s i . This level is calculated from the frequency that a service s i is included in the compositions done by a user u i . In case GIN returns zero, the fraction is not calculated to avoid problems of division by zero, and this function returns 0. SIL = IN GIN if GIN > 0, otherwise 0 Composed Service Interest Level . This function returns the level of interest that a user u i has in a Composed Service s i in terms of the interest that the services included in the Composed Service has for the user. CSIL = N k=1 SIL ∀ s k | ∃ e ∈ E & type E = includes & source = s i & target = s k Service User Similarity . This function returns the level of similarity that a user u i has with a user u j in relation to the interest in service s i . --- SU S(u i , u j , s i ) = SIL * ) SIL + 0 SIL + SIL * ) SIL + 0 SIL The SUS function should return the similarity level between two users from the division of the values that return their SIL functions. Instead of that, note that this function returns the result of adding two similar fractions in which the values of the SIL functions are divided but alternating the users between the numerator and the denominator. We do that to make the function symmetric, i.e. the function returns the same value independently of the order in which users are passed as arguments. Note also that to avoid problems of division by zero in case the SIL function returns this value, the numerator of each fraction is multiplied by 1-0 SIL , and the denominator is added with 0 SIL . Global User Similarity . This function returns the level of similarity that a user a u i has with a user u j globally, i.e. considering their interest of both in every service. --- GU S = N k=1 SU S Next, we introduce three algorithms that use these function to make recommendations in the three proposed situations. Note that the main goal of these algorithms is to provide contextualized recommendations, i.e. recommendations that fit the end-user situation in a specific moment. To do so, these algorithms try to predict the most probable actions that they can do in the context of the three different situations presented above. For instance, when end users are creating a Composed Service from scratch we want to predict the most probable services that they may want to add at a first place. In the area of prediction algorithms, the strategy that is usually followed to predict actions is analyzing a sequence of past actions done by a user in order to detect a pattern that matches with the current situation. The proposed algorithms try to do something similar, but instead of having a narrow view of a user's past actions, they have a broader vision when considering what the full list of followed users did before in each situation. For instance, if end users want to add a first service, we analyze what other users did in the same situation, and recommend the services they added first. Other solutions such as recommending the most used service in each situation could be used. However, we think that this solution does not consider properly the context of each situation. In the same sense, other dimensions such as service consumption are not considered by algorithms since they focus on the situations created when end users are composing a new service. The usage should be considered when recommending end users services to use. End users are creating a Composed Service from scratch. This algorithm recommends end users the first services used by their followed users in the Composed Services. This last recommendation is defined in Algorithm 1: given a specific user u i , we obtain all the Composed Services created by its followed users. For each of this Composed Services, we access the service included in the first place. Note that a composition is defined as a sequence of includes edges , so we access the target of the first edge to obtain the first service. Next, we calculate for it a Recommendation Level . RL is calculated from the product between the interest of the followed user in the service and the Global User Similarity between both users. Then, the list of services ordered by this level is returned. Algorithm 1. Recommendation of first services input u i ∈ V & type V = U ser f o r each u k ∈ F U f o r each cs k ∈ CS s k = target) RL k = SIL * GU S Add i n R e c L i s t end f o r end f o r S o r t R e c L i s t i n d e s c e n d i n g o r d e r o f RL k output R e c L i s t = { } End users want to compose a service by taking an existing Composed Service as basis. This algorithm recommends end users the Composed Services built by followed users with similar interests. This recommendation is defined in Algorithm 2: given a specific user u i , we obtain all the Composed Services created by its followed users. For each of these Composed Services, a Recommendation Level is calculated from the product between the Composed Service Interest Level of the followed user and the Global User Similarity between both users. Then, the list of Composed Services ordered by this level is returned. --- Algorithm 2. Recommendation of Composed Services input u i ∈ V & type V = U ser f o r each u k ∈ F U f o r each cs k ∈ CS RL k = CSIL * GU S Add i n R e c L i s t end f o r end f o r S o r t R e c L i s t i n d e s c e n d i n g o r d e r o f RL k output R e c L i s t = { } End users have a Composed Service partially defined. This algorithm recommends end users the services included in the Composed Services created by their followed users. This recommendation is defined in Algorithm 3. Given a specific user u i and a service s i that is the last service included in the composition that u i is currently creating: first, we access all the users followed by u i ; Next, we obtain all the includes edges that target s i and whose source is a Composed Service created by each followed user; for each edge, we obtain the service associated to the next includes edge within the service definition they belong; then, a Recommendation Level is calculated from the product between the interest of the followed user in the service and the Global User Similarity between both users. Finally, the list of Composed Services ordered by this level is returned. Note that this algorithm is applied after a first service is added, each time end users add a new service, independently of the position of the added service within the composition. Thus, when end users want to add the second service, the last added service refers to the first service; when users want to add the third service, the last added service refers to the second service; and so on. In this section, we present an implementation of the social network presented above. To achieve this, the first thing we need is an end-user environment to compose services and extend it to create such a social structure. For this purpose, we used EUCalipTool, an end-user mobile tool for composing services [5]. input u i ∈ V \& type V = User f o r each u k ∈ F U f o r each e k ∈ { e h ∈ E | type E = includes & target = s i & ∃ e j ∈ E | type E = author & target = source \& source = u k } snext=t a r g e t ) i f then RLnext = SIL * GU S Add i n R e c L i s t end i f end f o r end f o r S o --- Service Authoring Tool EUCalipTool proposed a mobile authoring environment which allowed end users to compose new services. The services that end users could compose were those registered into the platform by developers. To do so, developers were provided with a service registration form that allowed them to describe a service [5]. This form has been reused to provide developers with a user interface to create Basic Service profiles. EUCalipTool also proposed an editor to allow end users to create new services by composing those services that had been previously registered by developers. In this work, we have extended this editor in order to allow users from the social network to create services by composing both Basic Services and Composed Services . Figure 2 shows some snapshots of the first steps provided by the mobile authoring tool to create a new Composed Service. Such service can be created in two ways : from scratch or from an existing service. To create a Composed Service from scratch users must introduce a name and a description by using the proper form. To create a Composed Service from an existing one , users must select a service from the catalogue and customize it according to their needs. The catalogue of the non-social version of EUCalipTool provided end users with a list of: predefined examples and previous services composed by the own user. In this work, we go a step further providing a third option: Composed Services created by followed users that can be of interest to a given user. These Composed Services are provided by the Recommendation Algorithm 2 introduced in Section 5.2. We have also extended this tool in order to maintain the graph that represents the social network's structure and automatically create the edges that are inferred from the activity of composing services. Thus, note that the creation of a Composed Service through the mobile authoring tool results in the implicit definition in the social graph of an author edge between the current user and the newly created service. In addition, if the user selects an existing service to be used as a basis, a def inedF rom edge is implicitly created between both services. --- Fig. 2 Creation of a service composition with Recommendations A Composed Service is defined by using the metaphor of "adding an element" to a container. The Composed Service is the main container and users must include the services they want to compose to create the new service. The workflow metaphor is used to represent the sequence of the services included in a Composed Service. Graphically, it is represented by using the List layout . The order in which services are displayed represents the order in which services must be executed. Each service included in a Composed Service is connected graphically to the next one by a small inverted triangle. This aspect is inspired by the jigsaw metaphor, which defines pieces inserted into others to reinforce the notion of connection or combination of elements. Users just need to click the '+' button to access the list of available services . This list includes: the Basic Services created by developers, the services composed by the own end user; and the services used by followed users in the composition of services. The third option corresponds to the services that are recommended through the implementation of the Recommendation Algorithm 1 or 3 introduced in Section 5.2. Note that each time a service is added to a Composed Service an includes edge is implicitly created in the social graph between both services. In the same way, the ordered set that is associated to the Composed Service is dynamically updated. --- Service Browsing EUCalipTool has been extended with new user interfaces to browse services according to the possibilities presented in Section 5.1. If we were users of the proposed social network we would have the possibility of accessing the screen in Figure 3A, which shows a list of followed users . If we select a user from this list, we can access its profile , which includes the information proposed in Section 4. From a user profile, we can browse the services the user has created , which is shown in Figure 3C. In the same way, if we click on the plus button from the screen in Figure 3A we access the screen in Figure 3D that allows us to follow new users. Fig. 3 Browsing followed users and their associated services If we select a service from the list in Figure 3C we access its profile . This profile includes all the information proposed in Section 3.2. At the upper side of the screen, we can find a rounded icon next to the service's name. This button is indicating that it is a Composed Service. If we click it we access the authoring tool in order to create a new service taking as a basis the definition of this selected service . The rest of the information allows us to browse services from the different options proposed in Section 5.1. The location , the semantic tags and the device dependencies allow us to access the list of services that are related to this one according to these data. As a representative example, Figure 3F shows the list of services that have a dependency on the same location as the service in Figure 3E. In contrast to the list shown in Figure 3C, this one includes services of several followed users. Finally, note that a button labelled as "Similar Services" is located at the bottom side of the screen Figure 3E in order to access a menu that allows browsing other similar services. In particular, this menu presents two options to access the services that include this one and the services defined from this one --- Service Execution Although the execution of services is out of the scope of this paper, in this subsection we briefly introduce how a service is executed within the EUCalip-Tool platform. A detailed description of this issue can be found at [46]. Note, however, that we faced how an end user can execute their services. The execution of services created by others users of the social network requires a detailed analysis of all the security issues that may arise as well as how to solve the problem of executing services that depend on a device or a location that only have sense in the context of a specific user. This issue will be considered as further work. Figure 4 shows some snapshots of the screens that end users interact with when executing services. Figure 4A shows the list of services that are available for execution. Figure 4B shows an intermediate screen that informs about the execution process. Figure 4C shows a screen that requests end users some data required at runtime. Finally, Figure 4D shows the results of the execution, indicating the services that have been executed and the outcome provided by each of them. In this section, we present an experiment done with some students of our University, which were proposed to use the social network for one month. In particular, we asked 177 students to participate in the experiment. The experiment consisted of: completing a pre-test questionnaire to collect some data on students' knowledge and skills, using the proposed social network for a month and completing a post-test questionnaire to know the opinion of the participants about the social network. Next, we introduce the details about the experiment. Participants. From the 177 students that participated in the experiment, 94 were studying the forth year from the Degree of Computer Science. The other 83 students were studying the third year in the Degree of Audiovisual Communication. 124 were male while the rest were female. With some exception, most of the participants were between 20 and 25 years old. Apart from this demographic data, we used the pre-test questionnaire in order to know their technological habits. As expected due to the Degree they were studying, all of them had a high experience using smartphones and computers or laptops to browse the web, read e-mails, or use word processors or spreadsheets. In the same way, 96% of them used some social network daily. 48.7% of the participants also indicated that they play games assiduously. All of them had some experience in programming. As we can see, participants had a high technological background and skills. This was not a problem to perform the experiment since we wanted to evaluate the usefulness of the proposed social network. The usability of the composition environment provided by EU-CalipTool to end users with little technological background was evaluated in a previous work [46]. One of the most interesting data we wanted to know about participants was their experience on composing services or automating some set of actions. We though the evaluation of participants with this experience would be really interesting to estimate the acceptation of the proposed social network. Therefore, we included some questions in the pre-test questionnaire that asked if they had ever used a tool to automate tasks, including mobile apps like IFTTT with their applets, or home assistants like Alexa or Google Home with their routines. Around 31% of the participants answered affirmatively. Design. To present the experiment to students, we used the online teaching platform of our university as well as face to face classroom lessons. The experiment was done during a month and it was conducted as follow: 1. We asked students to complete the pre-test questionnaire to collect the data about participants presented above. This was a mandatory task that all of them must to do. 2. We arranged a classroom session to talk to students about the composition of services and present EUCalipTool. We prepared a case study to train students in the use of EUCalipTool. We also sent them a video tutorial of the tool in order to be further revised if they needed it. Afterwards, we asked them to use the tool freely at any time they wanted for a month. This was not a mandatory task. Not using it was an accepted option. We just proposed them to identify any situation in their daily life in which a set of tasks could be automated, and describe them with EUCalipTool. We reminded them that they had two options to do that, creating a Composed Service from scratch or reusing an existing one created by some followed friend. When composing services, we logged the selections done by students in order to know if the recommendations done by EUCalipTool were chosen. If students needed some service that was not available in the platform they could ask us for registering it. Once a week, we sent them an email remanding that they had EUCalipTool available to create service composition when they needed. We also introduced some remainder about EUCalipTool during the classroom lessons of the month that last the experiment. 3. Finally, those participants who used EUCalipTool to create some Composed Service were requested to complete a post-test questionnaire after the given month went by. We prepared a questionnaire with questions that focused on the experience of using the social network to compose services. It includes two type of questions: some that were used with a seven-part Likert scale from 1 to 7 points to evaluate them; and others that allow participants to introduce a free answer that justify some of the previous evaluations. In particular, we asked participants about: using a similar environment in a real scenario, using services composed by others when creating theirs, share their services with others; reasons for sending following requests; and the usefulness of the recommendation algorithms. Furthermore, we also included some questions to apply the Microsoft Product Reaction Cards [47] in order to evaluate the end-user satisfaction level. This method consists of providing participants with a list of words and asking them to choose the words that they would use to describe a product. The list includes positive words like 'Useful' and 'Engaging', but also negative words, such as 'Frustrating' and 'Ineffective'. Students were provided with an execution simulator that allowed participants to check the execution of services. This simulator is based on the user interface shown in Section 6.3. This simulator can be also used to execute the services of the followed users providing the possibility of testing these services before reusing them. Results. We analyzed the results of our experiment from two perspectives. On the one hand, we studied the participation obtained during the month that the experiment lasted. On the other hand, we analyzed the feedback provided by students through the post-test questionnaire. Regarding the participation, close to 57.6% of the students created, at least, one Composed Service. From these students , 44.1% created only one Composed Service; around 30% created two; 18.6% created three; and finally, only a 6.8% created four or more Composed Services. In absolute values, 193 Composed Services where created. Considering the total amount of potential users , we obtained, in average, a little more of 1 Composed Service per user. In order to know if these results can be considered good or bad from a participation point of view, we compared them with a well-established platform of task automation. In particular, we considered the data presented in [48] about the IFTTT platform, which had, in 2018, 14 million registered consumers and 75 million of applets since launched in 2010. In average, this means that each user created around 5 applets during 8 years. It is true that the comparison between a tool validated in an academic environment like ours and a platform used by millions of real users can be a little difficult due to the multiple contextual variables that need to be considered. For example, IFTTT users were not asked to use the tool in contrast to our students who were explicitly requested to use only EUCalipTool. However, we think that, on average values, it is a good point of reference to evaluate the participation of students in the experiment. Thus, considering that students were not force to create compositions, and that those created responded to a real situation identified by them, we consider that 1 Composed Service per user in a month reflects a good participation level if we compare this value with the 5 applets per user created with IFTTT during 8 years. An additional interesting data that we studied was the participation of those students with some experience in the automation of tasks. As commented above, 31% of the students indicated to have ever used the capabilities of tools such as IFTTT, Alexa o Google Home to compose an automation of tasks. From these students, close to 85% created two or more compositions, which illustrates a considerable interest in the use of EUCalipTool by those who were previously interested in automating some type of behaviour. This same data can be read from another perspective: 53.5% of those participants who created two or more Composed Services had some experience in this activity, and 46.5% of them used a tool for composing behaviour for the first time. Considering all the students that created some Composed Service , the platform provided by EUCalipTool was engaging enough to get 33 students who had never created such a composition before to create two or more Composed Services after using the platform. From the 193 Composed Services that students created, 41% of them were created from scratch and the other 59% were created by taking an existing service as a basis. This result reinforce the idea of using a social network as a valuable mechanism to create a repository of predefined elements for end users as it is recommended in [34]. The usefulness of taking a Composed Service as a basis to create a new one can be also appreciated in the right side of Figure 5. This figure shows how the total amount of Composed Services evolved along the month that the experiment lasted. We have identified four main periods, which are depicted by dashed red lines and explained next. We can see how the possibility of taking an existing service as a basis encouraged end users to create new ones. 1. Period 1 corresponds to the first week after we presented the experiment to students. As we can see, there was little activity in the creation of Composed Service by students. In average, 1.28 Composed Services per day were created. 2. Period 2 corresponds to the second week. After some reminders about the possibility of using EUCalipTool, students started to use the platform more intensively. In average, 6.28 Composed Service per day were created. An interesting data is that 92% of the Composed Services that were created during the first and second weeks of the experiment were created from scratch. In this sense, we concluded that students identified a scenario were some automation of tasks can be defined, searched for one that can support this scenario, didn't find any, and created a new one from scratch. 3. Period 3 corresponds to the next 10 days of the experiment. During these days the amount of created Composed Services increases significantly. In average, 11.72 Composed Service per day were created. In this case, however, 86% of them were created by taking an existing one as a basis. We concluded that students followed identifying scenarios were some automation of tasks can be defined. However, after the first two weeks of the experiment, there was a more extensive catalogue of existing Composed Services that can be used by students to create a new one that supports their need. 4. Period 4 corresponds to the last 5 days of the experiment. In this case, the amount of created Composed Services per day decreased. On average, 2.2 Composed Service per day were created. We concluded that students were not able to identify more scenarios that can be supported by a Composed Service. As far as the situations for which Composed Services were created, we detected that 35% of them were generally based on a smart home; 25% of them were based on smart cities; another 25% were based on the integration of mobile devices with social networks; and finally, a 15% supported some sport activity. Regarding the recommendations done by EUCalipTool, we suffered a wellknown problem in recommendation systems called the cold start problem during periods 1 and 2. This problem appears when a system tries to make recommendations and its knowledge base is empty. In our case, the problem appeared because students had created too few f ollower connections for the recommendation algorithms to provide valuable recommendations. This problem may be improved by initially considering the actions done by all the users, not only the followed ones. However, an exhaustive analysis of how solving this problem is needed. Note also that this evaluation focused on studying the usefulness of the proposed algorithms. An individual analysis of the functions that are used by these algorithms is out of the scope of this paper. In [45], we studied the correctness and completeness of these functions through a set of JUnit tests. Also, we did an experiment to evaluate the performance of their implementation and we obtained an acceptable execution time to achieve a good user experience when using EUCalipTool. Considering the Composed Services created in periods 3 and 4 , around 68% of those that were created by taking another as a basis were selected from the options recommended by EUCalipTool . The first services recommended by EUCalipTool were selected almost 55% of times, and the recommendations for the next services were selected 73% of times. These results are reinforced by the answer obtained in the post-test questionnaire, which was completed by the 102 students that created one or more Composed Service. Around 70% found Algorithms 2 and 3 useful . Algorithm 1 was found useful by a 61% of the participants. If we consider only those participants that had some experience on using tools for automating behaviour the results were the following: algorithm 1 was found useful by 69% of them, algorithm 2 by 76%, and algorithm 3 by 73%. These results indicate that the proposed algorithms help end users in the composition of services. However, they must be improved to obtain greater acceptance from users. A way these recommendations can be improved is by asking end users to describe the Composed Service they want to create in natural language, before starting to create it. We can match this description with the semantic keywords associated to the service recommended by algorithms in order to highlight those that better fit the Composed Service end users want to create. To do so, natural language processing techniques should be integrated with EUCalipTool. Following with the post-test questionnaire , around 80% of them stated that they would use a tool like EUCalipTool if it was available in a market such as Google Play or Apple Store. Regarding the questions about using the services created by others, around 74% of the participants would feel comfortable both sharing their services with others as and reusing services created by their followed users. Around a 69% of the participants indicated that the social network helped them to discover services . If we focus only on those participants that had some experience on using tools for automating behaviour , the results were the following: around 98% of them scored between 5 and 7 the question that stated that they would use a tool like EUCalipTool if it was available in a market. A 80% of the participants would feel comfortable sharing their services with others, and close to a 88% of the participants would feel comfortable reusing services created by their followed users. Finally, around 82% of the participants scored between 5 and 7 the question that indicates that the social network helped them to discover services. These results allow us to consider that, from a global perspective, the capabilities implemented in EUCalipTool from the proposed social network helped participants in the activity of creating services. An interesting issue commented by some participants was the possibility of sharing only some Composed Service to do only specific actions with them. Currently, an end user can access all the Composed Services created by their followed users to both take it as a basis to create a new one and include it in the definition of another service. The proposed idea consists in allowing end users to indicate if a Composed Service is shared or not , and if shared, which actions can do their followers with it. We found this idea really interesting and we are currently working on an extension of the social network to include this possibility. Participants also detected a problem when they reused a Composed Service that was published by a followed user and this service depended on a specific location or device. The actions that depend on these locations or devices cannot be directly reused and needed to be adapted. To improve this problem we are working on defining abstract services that can be instantiated when they are included by a specific end user . Regarding the connections among end users, a total of 218 following connections were created by the 102 students that create some composition. According to Figure 7, a 9.8% of these students created one following connection; around a 32.3% created two; around a 42.1% created three; and finally, around a 17.7% created four or more following connections. In average value, we obtained around 2 following connections per user, which represent a 1.9% of the active users . If we compare these results with a well-established social platform like Twitter, we can consider them quite good. According to [49], Twitter had in 2016 around 95 millions of active users and the average user had 707 followers, which represent 0.0007% of the active users. This percentage is quite lower than the one obtained in our experiment. The right side of Figure 7 shows the reasons that participants indicated to send a following request to another user. Two main answers were given: the user was a friend of mine , and the user had services I was interested in . Note that, initially, EUCAlipTool only allowed a user to know the services other user had created if it was a followed user. However, during the experiment, some students asked us to know the services that each user had created before to decide whether or not to send it a following request. Thus, we updated the EUCalipTool platform in order to show a preliminary list of the service that each user has created. In this sense, the use of a social platform seems to be an interesting mechanism to create trustworthiness and encourage end users to reuse services created by others to define their ones. As far as the end-user satisfaction level with the whole platform, which was evaluated through the use of Microsoft Reaction Cards, Figure 8 shows the words that got 10 or more occurrences. As we can see, "easy to use", "friendly" and "intuitive" were the participant's most selected keywords. These keywords show that subjects were pleased with the functions provided by the social platform and the way of using them. Conclusions. We can state that sharing services and browsing those published by others through a social structure is natural to most end users since they are accustomed to using social networks. In addition, recommendations done by analysing the social connections help end users in the activity of composing services. Thus, we can consider that the proposed social network is usefulness for end user to compose services. Of course, this is not a closed research work and some issues identified in the evaluation experiment need to Fig. 7 Participants and the number of connections they created and reasons to create a connection Fig. 8 Satisfaction evaluation with Microsoft React Cards be considered as further work. However, we feel reinforced in the idea that a social structure that support end users in the composition of services can be a valuable mechanism to help them to become into prosumers of services. Finally, as a representative example of the social network constructed in the experiment, Figure 9 shows a partial view of its social graph. It includes some fictitious Developers created by us . These developers provided several Basic Services that students could compose to create the composed ones . The graph also includes two End users that represent two students named Luis and Miguel. Luis created a Composed Service named waking up. Miguel followed Luis and created his waking up service by taking as a base the one created by Luis. In addition, Miguel created a Composed Service to support dinner at home. Threats of validity. This experiment was performed in an academic environment. We cannot assure the same results if the experiment is done in a real environment. The main reasons to do this statement are: 1. Students may have felt comfortable using and reusing services created by others because they all belong to the same group . Although unconsciously, a confidence relationship may exist among them. We cannot ensure the same behaviour when users have the possibility of reusing a service created by a totally strange user. 2. Our main goal was to have a set of users that can use our social network in a scenario close to a real one. However, note that a real scenario implies having users with a real need of creating a Composed Service. We cannot discard that some of them may have created services just for satisfying professor's desires. 8 Beyond a matter of sharing services among end users. Discussion In this work, we focused on improving the area of end-user development with a social platform to help end users to create their services and share them with others. Segal [34] recommends that end users can access a library of predefined components to use them as starting point. However, the same author demonstrated in [50] that these repositories of components are very difficult to maintain by developers in real contexts. The proposed social platform allows end users to access the services created by developers and other end users, and start to compose new services by using these "predefined examples" as a basis. In this sense, the social network becomes a continuously increasing repository of components that is maintained and improved by both end users and developers. Also, the proposed social network allows end users to find services by browsing a set user and services profiles, which are notions that are familiar for them. Social networks are currently one of the most used apps [8] and, as we validated in the previous section, the use of a social structure is familiar to end users, making the task of finding services easier than current service repositories. However, the idea of a social network in which end users and developers live together can be generalized to other domains rather than service creation. Even more, it can allow broader collaboration between end users and developers than only a shared creation of software artefacts. Regarding this last issue, it is interesting to remind that End-User Development focuses on allowing users without programming skills become into producers of software artefacts. From a Software Engineering perspective, EUD mainly focuses on the implementation stage. Other approaches advocate for the participation of end users in other phases of the development process. We can consider, among others, areas such as End-User Software Engineering [51], Participatory Design [52], Meta-Design [53], and so on. In this sense, we think that a social platform for end users and developers can be a valuable mechanism to allow them to collaborate not only in the implementation of software artefacts but also in other software development activities that focus on improving the quality of the developed artefacts. For instance, a user can publish an initial product description to allow other users of the social network to participate in the elicitation and management of requirements; or a software product can be published to be tested and debugged by others. This type of social platforms can support software development by and for the crowd. Note that current crowd computing research is investigating the ability to aggregate and employ human time and talent to develop specific tasks through digital media. We think that social platforms can be a valuable tool to do so in the context of software development. In any case, it requires an exhaustive analysis in order to properly define the social network that is needed for each purpose, determining user profiles, artefacts that can be shared, and the actions that each user profile can done with these artefacts. Next, we present a general description of the main concepts we think that are required to define a social network for developers and end users target at some software activity. Note that this is only an initial effort of generalizing our work. Further analysis and research are needed, but they are out of the scope of this paper. Figure 10 shows the concepts that we propose to characterize a social network target at software development activities. Each User of the social network is either a Professional, which is a user with skills in some development phase, or an End user, which lacks from these skills although have knowledge of the domain. Users are linked to each other through a Connection, which may be characterized as symmetric or asymmetric if we consider current trends in social networks. With a Symmetric connection, if a user A is connected to a user B it means that user B is also connected to user A . With an Asymmetric connection, the fact that a user A is connected to a user B doesn't mean that the user B is also connected to the user A . --- Fig. 10 Preliminary generalization effort Users can create Publications in order to share some Artefact with other users. We will need to precisely specify the artefact or artefacts that can be shared. Each user and artefact must have a profile made up of a set of properties which should be defined according to the software domain and development activities that must be supported by the network. These properties will be a valuable mechanism to browse artifacts. Each publication can have a Visibility in order to indicate which users can access it . Considering current social networks and some conclusions obtained from the experiment presented above, we can though of three types of visibility: Public, which indicates every user in the network can access the published artefact; Friendly, which indicates that the published artefact can only be accessed by connected users; and Private, which indicates that the artefact can only be accessed by the author. Each publication is associated with one or more phases of the development process which should be associated with the list of Actions that can be done with the published artifacts in each phase. For instance, if we define a social network that supports the phases of implementation and testing of a software product, we should define the actions that users can do with artefacts published in these phases. Finally, note that the artefacts published in each phase may be defined with tools specifically created for this purpose. Thus, the proposed social platform should be integrated with these tools in order to provide the proper working flow. Considering the social network proposed in this paper, the artefact that is shared among users of the social network is a Service. Each Service is published in the context of the implementation phase and has friendly visibility. Users are connected through asymmetric connections. The tool that supports the implementation of services is the EUCAlipTool authoring environment which is integrated with the new developed social platform. The actions that users can do with the published artefacts are two: taking it as a basis to create a new one and including it in the creation of another. If our social network would have supported the testing of services, we would have to consider which artefacts may be published for the testing phase , the actions that end users may do with them , and the integration of the social network with some execution and testing tool. In order to have another representative example, let's consider, for instance, the domain of Web Augmentation [54]. In this area, end users usually add, alter, or remove features of a web application interface by creating scripts that run on their web browser. A social network focused on the creation of these scripts would help end users to access scripts that may solve their needs, and which would have been created by professional scripters or other end users like them. This social network may be characterized, for instance, as follows: the artefact to be shared is a Web Augmentation Script, which is published for the implementation and testing phases. The actions that each user can do with a published artefact are two: the installation of the script in order to be tested, and the adaptation of a script in order to create a new one. The social network should be tightly integrated with some web browser and authoring tool in order to allow the deployment and edition of scripts. If additionally, we also want to consider the analysis phase, as approaches like CrowdMock [55] do, the artefact to be published can be, for instance, user histories that describe the new requirements to be supported by a Web Augmentation Script. The actions that other users can do with these histories are: to adapt or extend them in order to create new histories, to be associated to some existing script that solves it, or to be supported with the creation and publication of a new script. --- Conclusions and further work In this work, we have presented a social network to support end users in the composition of services. We want to encourage end users to become into producers of services and contribute to improving the research of end-user service composition. We have provided end users with a solution not only to build service compositions in an intuitive way but also to share the acquired knowledge among them. We have characterized the proposed social network and have defined it in a semi-formal way by using graph theory. We have also analysed how social connections can be exploited to facilitate end users to discover services through browsing these connections, and recommend services to end users during the composition activity. As proof of concept, we have extended EUCalipTool with social support. This tool supports end users in the composition of services by using mobile devices. We have extended it with social capabilities. This social version of EUCalipTool has been evaluated with students of our university for a whole month. The results of this experiment reinforced the idea that a social structure can be useful for end users in the activity of creating services on their own. However, there are still some challenges that need to be faced as further work. For instance, instead of using a simulator to test services, it should be interesting to allow end users to directly execute the services created by others. This also needs to face the security aspect of the social network by providing the possibility of defining grants over the shared Composed Services or integrating security frameworks with social support such as Anahita or Elgg to properly manage personal data. Furthermore, we are studying the possibility of including natural language descriptions done by end users in order to improve the recommendation process. Another interesting issue that we want to face is to include the usage dimension in the recommendations done by our social network. The main idea is to recommend end users with services to be consumed depending on the consumption done by their followed users in the same context . Furthermore, we are investigating how to apply a previous research work focused on considerate computing [56] studied how to achieve a considerate interaction with users when executing services. We plan to apply this work to the presented architecture to create social notifications in a considerate way. The main idea is to propose a conceptual framework that allows us to characterize both: the notifications that a user can receive from a social network like the proposed in this work, and the communication resources that must be used to deliver each notification to end users depending on their current situation. Finally, note that, although implemented in the context of EUCalipTool, the definition of the social structure can be applied to other end-user environments for composing services to provide them with social support. In the same way, we have presented a preliminary generalization effort to describe the main concepts that characterizes a social network target at any software development activity and any domain.

Nowadays, end users are surrounded by plenty of services that are somehow supporting their daily routines and activities. Involving end users into the process of service creation can allow end users to benefit from a cheaper, faster, and better service provisioning. Even though we can already find tools that face this challenge, they consider end users as isolate individuals. In this paper, we investigate how social networks can be used to improve the composition of services by end users. To do so, we propose a graph-based definition of a social structure, and analyse how social connections can be exploited to both facilitate end users to discover services through browsing these connections, and recommend services to end users during the composition activity. As proof of concept, we implement and evaluate the proposed social network in the context of EUCalipTool, a mobile end-user environment for composing services.

INTRODUCTION The increasing availability of miniaturized computing and networking equipment has fueled a multitude of research initiatives to delve into the issues and opportunities, of creating smart everyday objects, i.e. things that, on top of their usual function of office, kitchen, or decorative tools, are endowed with the capacity for autonomous rational action. A promising application for smart objects in domestic settings is supporting the wellbeing, independent living and social connectedness of ageing people. However, despite the claimed direct and indirect benefits that would result from a fast and widespread adoption, older people have not adopted these "smart object" technologies and in some cases actively resist, suggesting problems in their conception and or design [18,20,27]. As Taylor et al point out, perhaps resistance is because the technologies are designed to be smart, but do not make us feel smart [51]. Our work is aimed at designing technologies that engage older people by building on their individual affective attachment to habituated objects [6] and leveraging, from a participatory design perspective, the creative process through which people adapt their homes and tools to suit their individual needs and goals. This paper contributes a novel analytic framework based on an analysis of related research on appropriation and habituated objects, and with respect to technical advances on smart appliances and tangible/embodied interaction. We benchmark the framework against recent results from the design and evaluation of a smart communicating device conceived to enable connectedness over the traditional family routine of making tea: the Messaging Kettle. We show that characteristic phenomena of technology habituation cannot be easily understood only in terms of place-making, but require a more comprehensive model to cope with affection and the symbolic value of objects. --- RELATED RESEARCH Studies focused on design for domestic settings have investigated interesting properties of Internet enabled objects. Examples include almost any tool that could be found in the home, enriched of functionalities that may or may not be related to the original use of the tool: a knife and chopping board that know what food they are being used with [28]; a kettle that recognizes activity patterns of its users, such as when it has been picked up, if it is being refilled, etc. [3]; a coffee mug [4] augmented with a number of sensors for temperature, movements, position ; a tablecloth that holds and makes visible a memory of objects that were placed on it [19], and much more. As it has been observed, a large majority of such projects address isolated tools and their functionalities, rather than exploring how such objects could become part of everyday routines within the complex home ecology [53]. Paste the appropriate copyright/license statement here. ACM now supports three different publication options: • ACM copyright: ACM holds the copyright on the work. This is the historical approach. • License: The author retain copyright, but ACM receives an exclusive publication license. • Open Access: The author wish to pay for the work to be open access. The additional fee must be paid to ACM. This text field is large enough to hold the appropriate release statement assuming it is single-spaced in Times New Roman 8-point font. Please do not change or modify the size of this text box. Each submission will be assigned a DOI string to be included here. In fact, the goal of such smart objects is often focused on sensing and inferring either their own specific state or the surrounding context and act upon it. Additionally, such works are mostly innovation driven and empirically grounded; they explored interesting and challenging aspects of technology development and human-computer interaction, but made little attempt to explain why a certain design can succeed, whereas others may fail. The idea of smart home, on the other hand, has been largely sketched in terms of an assistive environment , capable of sensing and recording the activities of its inhabitants, and infer their needs or possible dangers, and often insisting on the need of a disappearing and unobtrusive design to encourage the acceptance of technology assisted living [27]. Such approach has drawn some criticism, not only based on privacy and ethical concerns, such as the risk of social stigma for the user, resulting from the lack of independence and disabilities [18], but more radically on the legitimacy of systematically anticipating and presuming how to care for the needs of users, rather than designing tools to empower them and their capacities [40]. Such a change of perspective is particularly urgent when designing for ageing people. In fact, while the specificity of needs and goals of older users is generally acknowledged, they are just as often stereotypically linked to loneliness and social isolation, illness, incapacity to use or learn to use technology [14]. Furthermore, the supposed homogeneity of people in old age, considered as a user group, has been challenged . Although evidence suggests a shared view of what 'ageing well' means among seniors, e.g. maintaining physical and financial independence, being socially active and actively engaged in their communities and families [36], it has been observed that the individual histories and experiences, including the scars that a long life invariably imposes, result for each person in a unique mix, and hence in a unique implementation of actual strategies for ageing well [36]. On the contrary, smart environments, as they are often conceived, offer little opportunity for active appropriation of, and engagement with, the technology: the older person is the monitored subject. Taking an opposite perspective, technology enhanced objects can be designed so as to reveal, amplify and inspire the capacities of people [39]. As Taylor and colleagues have observed, smartness does not belong to things, but rather to the way in which people appropriate and adapt those things: people shape and adapt their homes to their everyday use, for example using space and surfaces as a shared sketchpad to communicate with each other [51]. In fact, despite their potential, many innovations have failed to engage their potential users. To name a few examples, Pierce and colleagues [34] have pointed out the potential harm of energy feedback systems, including the risk of achieving opposite results from those intended by the designers. Rothensee [43] shows that potential users reacted with lack of enthusiasm when asked to evaluate the usefulness, ease of use and intention to use a smart fridge. An attitude of distrust and suspicion that may result in a firm rejection when the evaluation focuses on potential privacy violation [42]. The reasons behind such disappointing results are certainly complex and difficult to generalize; however, often smartness has been considered independently from the cultural and emotional attachments that characterize the home. As Leonardi and colleagues observe [29] the home is often the center of the emotional universe for seniors, technologies are more welcome in certain rooms, such as the kitchen, whereas the bedroom is often regarded as intimate, and the place of symbolic objects. Such differences, however, are far from universal; as pointed out by Rode [38] differences in infrastructure, size, number and use of rooms, number of occupants etc. call for more crosscultural ethnographies aimed at making sense of how different people conceive and shape their house. As an overall philosophy our research and investigation into technologies for the home focuses on engagement rather than monitoring. People thrive when socially engaged and older people may need more opportunities to socially engage. --- Technology acceptance and appropriation A large body of research exists that models user's attitude towards technological innovation and the successful adoption of new devices. The widely adopted Technology Acceptance Model [12] and its numerous improvements and extensions approach user acceptance in a workplace context, where perceived usefulness plays the pivotal role as a predictor of acceptance [21,43]. By contrast, the home setting, calls for a hedonic, rather than utilitarian dimension of interaction, in which ease of use and perceived enjoyment are key factors leading to acceptance and intended adoption [21,23]. Venkatesh [56] further deconstructs the perceived ease of use in a number of determinants, including control , computer playfulness , and emotion . Yet, it has been observed that TAMs, focusing on a managerial, top-down introduction of new technologies, fail to capture the user's interpretation of the role, value and purpose of a new tool [47] and the way a tool is integrated into one's activities and competencies. Such integration represents a continuation of design in use, as articulated e.g. by Suchman [49,50] and Ehn [15]: people actively figure out new uses for tools, and adapt their environment and practices to accommodate and exploit such uses in a continuing process of appropriation. In the home, this process has been characterized as "Everyday Design" by Wakkari and Maestri [57,58]: they observe how routines are often supported by finding new uses for existing tools, evolving and changing those tools, and discovering new affordances for them. Carroll [9] proposes for appropriation a model that separates technology-as-designed from technology-in-use, describing it as the process through which a new technology is adopted and transformed by its users, as opposite to nonappropriation and dis-appropriation that are observed when a technology fails to engage the users' interest, or, respectively, is abandoned even after an initial success, evidencing the dynamic and always evolving nature of the appropriation process. Riemer and Johnston [35] have characterized such adaptation in terms of place-making, i.e. an active process through which new technologies are first inspected by potential users and evaluated against existing skills, practices, and social norms; then the affordances of the new tool are explored, new skills may be acquired, and existing practices could be adapted accordingly; finally the new tool is placed amongst existing ones, and becomes part of the toolkit and of the social identity of the user. During this process the new technology moves from conscious attention to taken for granted , and from being perceived as a set of properties to being a means to achieve a specific goal. The factors that influence non-appropriation and disappropriation, such as cost, safety, security, usability, social status, etc. have been further studied in the context of mobile technology and young users [7,8]. Technology appropriation by older users has however received far less attention. While some of the factors that facilitate or inhibit appropriation in younger people may well apply, other important factors are likely to be peculiar of a senior audience. It has been observed, for example, how young users appreciate the ubiquity of mobile phones, and the possibility to access their favorite services anytime, anywhere [7] while the same possibility may be seen as invasive by some seniors, as it is not necessarily relevant to their own goals and values [6]. --- Habituated objects Robertson and colleagues underline the diversity of experiences, personal history, attitudes and needs, in contrast to the perceived homogeneity of ageing people that can be found in common sense and scientific literature, and that such heterogeneity of needs and attitudes also shaped the process of appropriation of new technologies, including the ones that are not specifically age-related [36]. Nansen and colleagues observed older people's experience of a natural user interface in relation to habits, habitus and habitats. They propose the concept of reciprocal habituation to underline the mutual adaptation that technologies and people act on each other through design, adoption and appropriation Hence, on the one hand, technology adapts to people's values and needs, practices and skills, houses and workplaces; on the other hand people acquire new skills, reshape their practices, redesign their houses to accommodate technology. When one or both parts fail to meet such accommodation, the whole process may fail, as is sometimes the case of new technologies that are not perceived as relevant or useful by older users [32]. On the other hand, habituated objects, which are already part of peoples' routines, may offer opportunities, through well considered technological enhancement, to enhance and extend pleasurable routines and socialization, leveraging the emotional value of the object or extending the associated rituals [6]. As observed by Forlizzi [16], products can evoke social behavior as they are embedded in a Product Ecology that includes social and cultural practices. The usual dimensions of interaction design, functionality and aesthetics, combined with the social, emotional and symbolic values of products, shape the activities and interactions that can take place in a given environment, including the social interactions [16]. Focusing on older users, Vaisutis and colleagues have investigated how some objects are invested with an emotional and social meaning by seniors [53]; they found that a special significance was often attributed to certain objects, for their capacity to afford independence, comfort, to communicate prestige or preserve tradition, to represent a social relation with a loved person, or to foster creativity and relax. They further observe that many of such objects are routinely used for social interaction, and could potentially be enhanced with technology to support communication and independent living. However they also stress that the strong emotional attachment that ageing people have to those objects, besides what they represent, means that such objects need to be treated very carefully in design [53]. Similarly, Leonardi and colleagues have investigated the meaning of the domestic space to older people, and the relation between objects and activities; they identify functional objects , symbolic objects and leisure objects , and recognized how different objects are found in different areas of the home, according to their intimate or emotional valence [29]. --- MESSAGING KETTLE It has been noted how elderly people, often marginalized in the adoption of new technologies, found a powerful motivation in keeping in touch with relatives, especially with grandchildren, for acquiring and learning to use a computer or mobile phone [33,52]. Family separation is becoming more and more common, as the world becomes smaller and the job market breaks through geographical boundaries, and new household models emerge that struggle to keep alive their family rituals. Yet, as we have shown above, existing messaging technologies, including mobile phones, social media, video conferencing, are not always easily integrated in the older users' practices, due to a lack of attention to such practices in their design. The use of messaging technologies by extended families has been the focus of extensive research. Rowan and Mynatt [44] describe a Digital Family Portrait capable of sensing and sharing the activities and whereabouts of an old person living alone to her distant adult children. The focus being on making routines visible, showing that life is going on as usual at Granma's, not much support is given to explicit messaging. Lindley [30] studied technology supported lightweight communication between households using a tablet-like device capable of supporting drawing, handwriting, taking photos and of sending such simple notes to preregistered contacts. Thanks to the convenience, asynchronicity and glanceability of the device, the study reports a great engagement and on the seniors' part, that notably were the most active in sending messages to their children and grandchildren. Tee and colleagues explored how extended families use technologies to share and communicate [52]. They highlight several themes that describe common challenges in family communication, including fitting communication in people's busy lifestyle; inadequacy of hardware/software to cope with certain media formats or lack of specific skills; concerns arising from the sense of obligation and the burden of keeping in touch. They also underline how a larger majority of participants reported a desire for more communication with their extended families, but are either too busy or believe the others are too busy [52]. Other projects have explored the affordances of various house tools and devices, how they were used to support communication and how they could be extended to support remote communication or share a ritual/routine over a distance: among the many examples, the fridge door [51], a cup [10], and a candle burner [1]. Finally, intimate communication at a distance has been proposed by Mueller and colleagues [31] in the form of a hug at a distance; we considered the most limited form of sending hugs, bits [25,26] or pokes , to be too limiting in terms of engagement and semantic expression. These limited forms seem to survive in social media, because these services support additional expression through other means. However none of the very limited forms of phatic expression appear to have gone to long term trials or to have survived beyond the concept stage. With such scenarios of separation between older people and their adult children in mind, our research on the Messaging Kettle aims at supporting communication over a distance by augmenting the functionalities of a traditional kettle with messaging and networking capabilities. The communication is asynchronous in order to be tolerant to time zone differences of families living in different parts of the world, but should instead unroll over a longer period of time, while yet preserving some of the features of the original tradition, i.e. gathering around tea and sharing family matters. The Messaging Kettle is different from previous research in several aspects. First the object, a kettle, is one that is used in particular domestic routines. Second the communication method indicates in real-time when the routine activity is occurring and leaves a trace that it occurred. Third, compared to many of the examples given above, communication is of a slightly higher yet limited bandwidth, allowing brief asynchronous messaging that allows engagement and semantic expression. The overall setup and a prototype implementation are briefly illustrated below. A traditional kettle is augmented by means of external devices that provide sensing and messaging capabilities. A Kettle Mate embeds a temperature sensor and is capable of inferring when the kettle has been turned on and of providing voice input/output. A smart Tea Box includes a dedicated tablet with touch and pen input for creating text based messages and provides the computing capabilities and Internet connectivity. Two Messaging Kettles are coupled, as in a presidential hotline, to display each other's state, and to forward voice and text messages. When one Messaging Kettle is turned on, the companion device, possibly located several time-zones away, starts to glow in order to make visible the remote activity. The user has the possibility to attach a voice or text message to the activity, hence sending a greeting to the distant person. By combining pragmatic elements, hedonistic motives and emotional value, the Messaging Kettle represents a useful case study to shed light on technology habituation and on how to support and encourage intimate communication by means of smart familiar objects. In particular, our research is aimed at showing that focusing on engagement older users can be encourage to embrace smart objects, and stay connected with their extended family, that supporting the specific ritual of making tea can shape the conversation in a peculiar way , and that externally augmenting an existing habituated object can introduce new functionalities to enhance an existing practice without compromising the affective value of existing objects. A long term evaluation is currently being carried on, and results gathered so far are presented further below. The evaluation is articulated in several phases. After building the initial prototype, with the goal of testing and refining the concept, we demonstrated it to several potential users. We hosted two "morning teas" during which the functionalities, intended use, and design alternatives were discussed. --- Appropriation Inspection [35] Properties are compared to existing skills, expected affordances, existing practices, projects and social norms. --- Attractors Cheap, Convenient, Control, Usable, Fashionable, Familiar, Our Stuff [7] External support, Playfulness, Usefulness [56] Place-making [35] New skills are acquired, new affordances are discovered, too is placed among other tools, finds its place into practices and socia identity. Performance [35] The new tool withdraws from attention, has its own place among tools and practices, is a mean to a goal and is part o social identity. Habituated object [6] Tool is adapted to people's values, needs, practices, skills, and houses [32]; has a role as functional, social, symbolic or leisure tool [16,29]. Potentially holds a sentimental value, and may not be replaced [16,53]. Appropriation criteria [7] Social Management, Leisure Use Safety/Security, Information Management, Lifestyle Organizer, Critical Mass Significance [53] Object affords independence, provides comfort, communicates prestige/tradition, represents social relationship, fosters creativity/relax. Non-adoption [9] Tool fails to engage potential users, can't match existing skills/practices, uses are deemed not relevant Dis-appropriation [9] Despite initial adoption and integration, falls into disuse --- Repellents Costly, Inconvenient, Controlled, Frivolous, Unfashionable, Unfamiliar, Their Stuff [7] Computer anxiety [56] Dis-appropriation criteria Hidden Cost, Unusable due to poor health, Reception , Unusable, Un-learnable [7] Figure 2. A framework for technology habituation; derived from [5][6][7]9,16,29,32,35,53,56] Further habituation criteria • gradual habituation • place prototype in situ • time management • accommodate asymmetry The first morning tea was held with 5 people in their 50s-60s all of whom had an older parent in their 80s. The second one was held with 6 people in their late 70s accompanied by two adult children in their 40s and 50s. The discussion during morning teas focused, for the first one, on the challenges and experiences of staying connected with their older parents or adult children. The second morning tea focused on experiences of difficulties with technology and how this limits the opportunities for communication with their adult children. The morning teas are methodologically grounded in Future Technology Workshops [54], and aim at uncovering the gap and possible interactions between current and envisioned technologies, leveraging the domain expertise of participants, and understand how future activities will be reshaped by technology. Following the morning teas, two in-home demonstrations were performed with an older relative of one of the research team and a friend of hers . Morning teas and in-home demonstrations have been documented in [5]. After a further phase of development and several improvements, a more stable prototype is now being deployed. One installation is currently running between the homes of one of the research team and one older relative . Four more prototypes will be gradually deployed in the next several months, with modalities that will be decided and adjusted based on the current progress. The prototypes are being rolled out slowly in order to: gradually learn about the details of habituation; refine them based upon feedback; and to minimize participant inconvenience and management issues associated with deploying a research prototype in real homes. The need of longitudinal studies, particularly when observing technologies that fit into the daily activities and domestic routines, has been repeatedly advocated . The approach described above aims at uncovering issues of longer term habituation involving the concrete prototype and its use experience in one particular place and social relationship at a time. Our goal is to elicit reflection and facilitate the projection of possible uses and misuses by personally adopting the new technology and presenting it in its situated use, in a serendipitous, rather than artificial, context. As argued by Sacks [46], novel technologies are introduced into existing practices , and it is from such process that real uses emerge and design insights can be figured out. For this to happen, long term use and continuous access to the technology is crucial to move past superficial insights about novelty and usability [22], in order to uncover the affective and symbolic dimension of the augmented habituated object. The researchers keep track of such insights by means of a diary, snippets from which are reported below where appropriate. For clarity, quotes from the morning teas will be annotated as [MT1] and [MT2] respectively, quotes gathered during the in home demos will be indicated by [H] while snippets from the research diary will be indicated by [D]. --- A FRAMEWORK FOR TECHNOLOGY HABITUATION To better understand and structure the analysis of technology habituation phenomena, we propose the framework detailed in Figure 2. It combines the related literature presented above to our own findings and relative determinant factors and serves as a guide to structure the exploration and understanding of the process that leads to technology habituation. Related research has been integrated in the framework when the proposed models addressed specifically the appropriation or habituation phenomena or when they discussed variables or determinants that are consistent with related work on appropriation . Other models, most notably the TAMs focusing on a utilitarian perspective and workplace domain, seem less capable of capturing the social, emotional and symbolic value of habituated objects and the process that leads to habituation. In this section we structure our findings from the ongoing evaluation of the Messaging Kettle. Our aim is to show that many observations (namely those ones that relate to the affective and symbolic dimensions, cannot be easily understood solely in terms of technology acceptance or appropriation, and rather require a more comprehensive model. --- Figure 3 One of the first sketches received by the researcher was a note on the weather, initially only black colour was used. --- Steps of Habituation --- Inspection Inspection involves getting an overall impression of the device and deciding whether or not one will try it out. This might be in a shop or in a friends house. In our case inspection happened at morning teas and also when the kettle was taken and demonstrated in the home of two octogenarians. During inspection the properties of the new tool/object are explored and compared to existing skills, expected affordances, existing practices and social norms [35].The alternative modalities and the visual feedback offered by the messaging kettle were seen as complementing each other. Participants in morning teas described problems with poor sight or hearing and felt the complementary modalities of the glowing lights, scribble pad and voice message would be helpful. Other everyday technologies, such as phone calls were deemed somehow less reliable: [MT1] "my mother hates her mobile because she has an hearing aid ... and typing on the keypad... it's all too small for her", which results in frustrating attempts to communicate or have news [MT1] "when she takes off her hearing aid, she forgets to put it on again, she can't hear the phone and you can't get any answer", or also, [MT1] "there are times when mum forgets to hang-up the normal phone and the mobile is not charged and I cannot get through". With the kettle deployed in the homes of researcher and mother, several friends have seen the kettle in use and this leads them to make projections on how it would work in their relationships. Seeing a prototype situated in a home and seeing how it is used, is very evocative for others and enables them to project all sorts of scenarios if it were deployed in their own relationships. Projections ranged from positive to negative. One single mother really wanted one to keep in touch with her own mother who lives in another city in the same country and they have asked to participate in the trial. Others had relationships in which they would find it irritating and onerous to deal with the messages with certain family members. One felt that some people in relationships with power imbalances would insist on it as a surveillance device for their partners. Quickly the question is raised of how many family members should be able to connect and what if one person wants the connection but the other doesn't. --- The emergence of adoption and Place Making At some point a decision is framed in which a person decides whether or not to try out a new technology. It may be an independent decision, one made at the behest of or inspired by friends and relatives etc. The framing and eventual decision are subject to many contextual factors, including whether one wants to engage or not with another person or persons through a particular technology platform Offering the messaging kettle was a delicate affair, not wanting to foist it on a relative and offering as much to take it away so it wouldn't clutter the space as to leave it there. It was important to us that participation was voluntary. It was accepted tentatively. [H] "Well I suppose we better try it out hadn't we." It is the sort of comment that is very hard to interpret even if you know the family. Would or wouldn't she like it? Was or wasn't she willing? It wasn't clear. She seemed willing but there were probably some reservations. We could only wait to see if it got used. Inspection doesn't necessarily result in clear decisions about adoption. It is an emergent affair in many cases and not something than can necessarily be resolved in design workshops or demonstrations. --- Place-Making Place-making involves acquiring new skills, discovering new affordances and making a place in an ecology of devices and practices [35]. The In-home demonstrations revealed issues regarding the availability of space and power sockets in the kitchen, and how the new device could be made space for [5]. One participant proposed a different form factor for the Tea Box, that could be hang on the wall, to take less space. The first skill to develop during the long term trial was to make sure it connected between Europe and Australia. --- [D] 'First we needed to work out if it is working. At first I got no messages from Mum which provoked all sorts of scenarios in my mind. Did she feel obliged? Is it somewhat of a tyranny? Has she just switched it off as she can't be bothered with it; I didn't want to hassle her so waited a few days before phoning. Then we worked out on the phone that the connection at my [the researcher's] end wasn't working." She had been sending messages and getting no response. Although configuration has been made minimal, the initial setup needed working out and this is almost always the case. Both parties learned what signs of life in the messaging kettles meant that they were connected. Both parties also learned how to reboot their messaging kettles when they didn't seem to work. Once connected, learning the features was quite gradual evolving over two weeks with each once or twice daily message (see Figure 3 --- ) [D] "My Mum is only using black pen at the moment. Not sure if it is because she likes the contrast for her vision, or if she hasn't figured out the colours…. Today she is writing in colour. … Yesterday she went to her Macular Disiease society meeting and today she has drawn her first drawing, a seeing-eye dog… Surprise, today my sister showed up in a message on the teabox -she is visiting my Mum. She has tried out the audio button and managed to leave a message. I still haven't had one from my Mum though. …Now we have figured out how to both write on the same page". --- Performance Once performance is enabled the new tool withdraws from attention and has its own place among tools and practices. Once habituated the tool is adapted to people values and needs and has a role as a functional, symbolic or leisure tool [35]. Place-making, performance and habituation are somewhat overlapping phases. Although place making has begun and some performance has been enabled with the tool being used on a daily basis, there are still aspects of performance being worked out such as sound. The researcher is reluctant to push the sound feature too quickly, as it seems to be the least reliable aspect of the prototype. It seems better to simply enjoy the tea-box messaging and lava lamp glow of the kettle-mate when the remote kettle is boiling. The sound can be worked out later… it might need fixing. It seems important not to rush and focus on the technology when the tea-box messaging and kettle glow is an enjoyable way to communicate. While performance is being worked out, some aspects of habituation are being experienced. --- Habituation When a tool is habituated it fits people's values, needs, practices, skills, and environments [6,32]; the tool has a particular role as a functional, symbolic or leisure object [29]; potentially it will be given a sentimental value, and can be considered hardly replaceable [53]. Although performance is still being enabled some aspects of habituation are occurring. The kettle-mate has begun to represent the extended presence of the older person. --- [D] "I think the aesthetic is lovely -much more lava lamp than smoke alarm. I find it both exciting and calming. I love it when my son says "Mum, grandmas kettle is boiling and I look across and see the lavalamp like glow." Presence is felt though shared experience over time. [D] "I used to calculate if my Mum was up or not. Now I often see the kettle glow between 7:30pm and 9pm at night, when she is having her morning tea. It's the only time I see it, in part because of my own routines. Now I think of this evening time as the time that she is around and we are both available to each other. But I am not sure if she sees the glowing often in a habitual way yet". We acknowledge that the reporting of experience is researcher focused due to the greater availability of the researcher's diary. [D] "As for what my Mum thinks, I ask her each time I call . I haven't asked her to diarise her feelings, as it seems using the new technology is enough to ask. Each time it breaks I ask if she has had enough. But she is persistent, saying I think we should keep going to make sure it works. Then she tells me of the messages she wanted to write except that she's has a problem with the tablet after going away for a week when it lost all its charge. She actually seems to really enjoy it. I am equally aware that at some point she could find it burdensome though, only because you never know. Those are the range of experienced that I can imagine, knowing her well". --- Determinant Factors --- Attractors The design builds upon familiar and relatively ubiquitous practices, such as leaving a note in the kitchen and recording a voice message. It was expected that the leisure use act positively towards place-making. Attractors are principles that can facilitate or encourage the adoption of a technology, as described in [7,9,56]. Participants from the 50 and 60 y.o. workshop were generally enthusiastic about the messaging kettle, as they intended it as specifically designed to keep in touch with their older parents [5]. All participants immediately pointed out the capability of the messaging kettle to enable a calm form of virtual presence that allows each user to be aware of some activities that are going on at the other end.[MT1] "[…] it would be great if you want to keep an eye on a parent […]". "so… it actually has two purposes: communication and seeing that they actually are still active, that there's nothing wrong". While the risk of a domestic accident is always present, not only in the elderly people's life, knowing that daily routines are going on as usual is reassuring. One participant mentioned being in contact with a neighbor: [MT1] "we have a neighbor, if the blind doesn't go up in the morning they go and check on her". --- Repellents Several critical aspects were considered in the design. The cost of the service represents a potential repellent, as is the availability of necessary infrastructure as well as the perceived control over the system, including privacy issues. Ease of use and learnability have been considered in terms of heuristics, such as visibility of controls and system status, error prevention, minimalistic design. Repellents play against adoption, and can result in the technology being refused [7,9,56]. Some older participants expressed a preference for more traditional forms of voice and face to face communication, saying [MT2] "The new way is to text etc. but I like to talk on the phone". "Messages can be received the wrong way if you can't see or hear the person". Some participant reported that their relatives were very active in various respects, but not particularly into technology: [MT1] "I first tried to get her to have an iPad, originally she got a laptop, she doesn't have internet access, she tried with email and photos, and things like that, but it was too much for her […] after a couple of months she gave her laptop away". On the other hand, the simplicity of the interaction with the messaging kettle was regarded as a clear advantage: [MT1] "just a button to press to record something, and then you get it on the other end". One participant commented that such simplicity was likely to be an attraction for an older user:[MT1] "it may be psychologically good for elderly people [as they] think they're not very clever... they may be very impressed with their expertise if they can do that". Issues with potential costs [7] were observed during the deployment as well:[D] "she likes to turn off her computer and her home wifi all starts up when she turns on her computer". Also in this phase it becomes clear if the functionalities are well suited to the user's skills and capabilities [7]: [D] "[she] is only using black pen at the moment. Not sure if it is because she likes the contrast for her vision, or if she hasn't figured out the colours". --- Appropriation Criteria Participants elaborated on the motivation that could influence their own or their relatives' appropriation of the messaging kettle. With appropriation criteria we focus on aspects that could impact a regular daily use, and the technology entering the people's routines [7,9]. The possibility of 'keeping an eye' on old parents has been discussed above, but pure and simple pleasure of keeping in touch on daily basis was frequently mentioned, especially for [MT1] "just those little things when you don't need to have a big conversation". Participants felt that just recording a short message when using the kettle would be enjoyable but not onerous. Also, the possibility to receive a message from grandchildren was highlighted as most desirable: [MT1] "and kids... they can draw something for grandparents... and its fantastic... to get something from them". Being parents themselves, and having sometimes adult children living and working abroad: [MT1] "[useful to] other than old people, my youngest daughter lives in --... so she could get up and leave a message Hi mum... Hello", some of the participants highlighted the possibility to receive such tiny messages: [MT1] "they could send a quick message or something and vice versa […] rather than rely on phone calls". This is seen as a way of preserving a peace of mind, [MT1] "because when you are a long way away from your family you know they're safe, you know they're all right, they're well, but sometimes you just wonder if you haven't heard from them". --- Dis-appropriation criteria On the contrary, dis-appropriation may occur when the usefulness or convenience of a technology is no longer valued or relevant [7,9]. Participants almost invariably stressed that the messaging kettle was: [MT1] "useful -as long as it was simple -as long as there was no setup". Yet, participants observed that the cost, and availability of infrastructure, could be an issue, as it was in the case of landline and mobile telephone:[MT1] "because where she is that still costs her money, because she doesn't have one of those unlimited plans", also: "with the old parents... they don't have internet access… mum, she's got internet access... but they had to change their device to be able to have that connectivity…". The aspects of time management were often mentioned as critical and potentially a deal breaker. Having to connect people across time zones the Messaging Kettle could become an annoyance if it would call for attention at night [5]. Older participants often expressed a desire to communicate more, but also a difficulty to find the proper time to connect: [MT2] "They're always busy, I'd love to speak that one now but it's the wrong time.". --- Significance The Messaging Kettle builds upon the significance of the family tradition of tea making to frame and contextualize the interaction. The corresponding messaging capabilities are designed to provide a one-to-one link with a loved person, hence amplifying the intimate and personal dimension. After some weeks of constant use The Messaging Kettle is became part a daily routine [7]: [D] "the kettle is now linked to my Mum. When I am in the kitchen I think of her. I think of her much more because of the kettle"; the choice of implementing it as one-to-one messaging system resulted in a certain unobtrusiveness, and the device is starting to symbolize the relationship [53] that exists between its two users: [D] "not feel like I have to open a computer and deal with emails from everywhere else". --- It is also representative of the other person's routines: [D] "I can envisage where she is. If it is a text it could be from anywhere. But it is her handwriting in her kitchen". The aspect of using handwritten notes was particularly relevant: [D] "Usually most of what I see these days is typed", perhaps because of its association with tradition and intimate communication: [D] "I usually only see […] handwriting on birthday cards these days". Eventually, the scribbling functionality started to be used more creatively [53] to send humorous messages: [D] "My mum sent a pinch and a punch for the first day of the month" and sketches "Today my Mum has drawn a dog! A seeing eye golden labrador that was at her Macular Disease meeting" --- DISCUSSION AND CONCLUSION The analytical model presented in Figure 2 deconstructs appropriation into phases, from inspection to performance. When users encounter a new appliance or device, they first consider whether they would want to use it and whether it would fit for them. In terms of details of appropriation theory this would be described as first inspecting its properties to match them against existing practices and skills, and frame them into existing activities, practices and social norms. We can reasonably expect the Messaging Kettle to undergo a similar inspection. It can be described based on its properties as a messaging device, a note-taking pad, a kitchen related sensor, etc. Attractors and repellents are considered in this phase, and can lead to intended adoption, and eventually place-making, or to non-adoption. As discussed above, in the example of the Messaging Kettle, cost, usability and familiarity can be expected to play a critical part. . During a subsequent phase of place-making, people acquire new skills and discover new affordances, i.e. invent new uses, for the tool. In the case of the Messaging Kettle, people may learn how to operate its controls, recover from errors, etc. As the Kettle Mate is designed to sit on the kitchen bench close to the kettle, a place for it must be organized, with access to a power plug, in a position accessible and convenient to clean. Addressing such practical concerns is critical for the new tool appliance to disappear from attention, and become a taken for granted accessory of the kitchen activities. In this phase, and continuing throughput the lifecycle of the tool, appropriation and dis-appropriation criteria play a relevant role. Leisure use, safety and privacy, hidden costs and usability issues may determine the success or failure of the place-making activity. The subsequent performance, in our intention as designers, can see the Messaging Kettle become part of the daily routine of tea making, and checking messages or leaving a message for the loved one become a part of the ritual of tea making, i.e. just like the kettle is the tool for making tea, the messaging kettle becomes the tool for making tea and sharing it with a distant person. A crucial role to facilitate the appropriation process relies on the practical and affective value of the already existing kettle, and the surrounding practices and rituals. From its status as a habituated object, we expect the kettle to facilitate the appropriation of the new functionalities of the messaging kettle, and the corresponding place making for the Kettle Mate and tea box The early experience of using the kettle confirms that all of these stages occur from inspection to place-making through to performance and habituation. It highlights the extent to which adoption and habituation can be gradual and evolving, its success dependent upon many contextual factors, in particular the relationships within which the technology is embedded, but also the habits, habitus and habitats as articulated [32] by Nansen et al. The model also indicates the role of gradually habituating prototype technologies as a design method that places a prototype in situ and thereby allows people who encounter the prototype to project how it would play out in their own and others relationships. This method relates of course to many related methods of designing "in the wild" [41], RAID "reflective agile iterative design" [22] and technology probe methods [24]. But usually these methods elicit the experience only of the participants. Habituation, which acknowledges relationships and place-making, also acknowledges all of the people who encounter the prototype and then consider how it might relate to their own lives. This has been a strong element of the experience of prototyping in the Messaging Kettle project. Gradual habituation may well be an important approach to designing technologies for older people, who do not want to spend their lives configuring technologies, and may like to see how they are used by others before adopting them. So far this projecting phenomena has largely been witnessed on the side of the adult child , although there have been two known demonstrations by the older adult, which have led to projections about use by those seeing the prototype. In this paper we have discussed a conceptual model of the process of appropriation that we apply to discuss the design and evaluation of the Messaging Kettle. Our work is grounded on related research on placemaking and habituated objects, and highlights the main phases of appropriation: inspection, place-making and performance, in relation to the main variables that facilitate or obstruct the process of appropriation. We further link the models of appropriation available in the literature to the concepts of habituated objects and reciprocal habituation, and the main variable underpinning the significance of habituated objects for elderly people. The Messaging Kettle represents an example of design for appropriation that builds on affective value and existing function of an already habituated object. By choosing to augment the functionalities of the existing kettle by means of a Kettle Mate, we aimed at exploiting the existing tool as a host for new functionalities, leveraging the affective attachment and the existing place-making, and framing the use to the existing rituals.

Smart everyday objects could support the wellbeing, independent living and social connectedness of ageing people, but their successful adoption depends upon them fitting with their skills, values and goals. Many technologies fail in this respect. Our work is aimed at designs that engage older people by building on their individual affective attachment to habituated objects and leveraging, from a participatory design perspective, the creative process through which people continuously adapt their homes and tools to their own lifestyle. We contribute a novel analytic framework based on an analysis of related research on appropriation and habituated objects. It identifies steps in appropriation from inspection to performance and habituation. We test this framework with the preliminary testing of an augmented habituated object, a messaging kettle. While only used in one home so far, its daily use has provoked many thoughts, scenarios and projections about use by friends, both practical, utopian and dystopian.

Introduction When the COVID-19 pandemic first began in early 2020, the focus of public health measures and healthcare systems was to protect seniors and those with compromised immune systems. This focus was both prudent and understandable given age was, and still is, the strongest risk factor for COVID-19 mortality [1]. While early in the pandemic, cases were highest amongst older adults over the age of 65 years old, cases in young adults soon outpaced those of any other age group [2]. In the summer of 2020, the World Health Organization issued warnings citing young adults as the main "spreaders of the virus", as the COVID-19 incidence rate among the young adult age group was rapidly increasing around the world [3,4]. In the United States, it was reported young adults between 20-49 years old accounted for more than 70% of the spread in 2020 [5]. In addition, this age group will likely receive their vaccinations last as countries vaccinate their population based on age, yet they have the highest vaccine hesitancy [6]. Young adults also face unique challenges in the COVID-19 pandemic due to public health orders restricting social interactions at a time of crucial social development and independence, working in high-risk, low-pay essential service jobs, employment in occupations without medical benefits, and financial insecurity [7,8]. While others may have the option to safely stay at home due to the flexibility of their occupations, many young adults are essential workers and do not have this same option. In addition to the low-paying jobs that do not provide sick leave, young adults in such public-facing occupations are at a higher risk of COVID-19 exposure and infection. However, studies have argued the lower perceived risk and worry regarding COVID-19 in young adults may have resulted in lower compliance with public health orders such as restrictions on social gatherings and mask wearing compared to older adults [9][10][11]. Moreover, mass media stories and public health messages have resorted to blaming and shaming for individual decisions on health behaviours despite evidence indicating this as ineffective [5,12,13]. Despite the high number of cases and concern of adherence to public health messages, few jurisdictions have implemented public health messaging targeting this age group. There is limited understanding of young adult perceptions on COVID-19 and effective public health messaging towards youth for communicable diseases [14,15]. Therefore, the current study sought to explore the experiences and perceptions of young adults aged 18-40 in British Columbia , Canada relating to the COVID-19 pandemic to best inform public health messaging. --- Method Participants were recruited via existing research networks, social media , and snowball recruitment from October to November 2020. Participants were eligible to participate if they were between 18 and 40 years of age, had resided in B.C. since March 1, 2020, had access to a device capable of an online conferencing software either by computer, tablet or phone and were able to provide informed consent. The inclusion criteria that participants had to reside in B.C. since March 1, 2020, ensured the participant resided in B.C. since the first provincial lockdown on March 17 th , 2020. Over 200 potential participants showed interest to participate, but seven did not meet the inclusion criteria. We strived for representation across gender, geography, employment status and education. Following approval from the Simon Fraser University Research Ethics Boards, 50 participants were recruited to participate in the study. Participants gave verbal or written informed consent prior to enrollment in the study and were scheduled into groups of 4-6 to participate in a single, semi-structured, open-ended focus group audio-recorded via Zoom video conferencing. The focus group discussion methodology was chosen for its strength in providing access to in-depth understanding of shared social meanings or norms. Through focus group discussions, we aimed to explore the natural interactions participants might have in their daily lives and how these interactions may influence their perception of existing health messaging. Each focus group was attended by one moderator and one notetaker from the research team. The focus groups were conducted in English and lasted between 60-90 minutes, depending on the participants' engagement in the discussion. Each participant was identified in the transcript by the focus group they participated in and the gender of the participant. The research team facilitated each focus group with a semi-structured question guide that inquired about participant experiences and perceptions of current public health messaging on COVID-19. All audio recordings were transcribed verbatim. --- Focus group discussion analysis Qualitative analysis of the focus group transcripts was performed by two experienced qualitative researchers. Using NVivo qualitative thematic analysis software 12 TM , findings were compiled on the verbatim transcription of the transcripts following the six phases of thematic analysis as outlined by Braun and Clarke [16,17]. The phases summarized in Table 1 were completed Themes were inductively identified from the raw data without any predetermined classification. The researchers first generated rough codebooks to guide the coding process and the codebooks were refined as the coding process went on. This coding process was interactive and reflective with movement between phases and two experienced qualitative data analysts of the research team working in collaboration, minimizing bias in the interpretation of the interview data. Discrepancies among codes that emerged throughout were reviewed and merged with other suitable themes after group discussions until consensus was reached. We closely followed the consolidated criteria for reporting qualitative research checklist to guide and report our findings [18]. --- Results --- Participant demographics A total of 50 participants within the age range of 18-40 participated in the focus groups. Almost half of the participants were women and the other half men with --- Searching for themes Excerpts from each code were re-read. Themes were generated by organizing initial codes across all eight focus group transcripts. --- Reviewing themes Researchers collectively reviewed the themes to reach a consensus through a series of group discussions. Extracts from each code were reviewed to check whether they are represented by the themes. --- Defining and describing themes Researchers defined and named the themes to ensure accurate representation of central ideas and concepts within the data. Ongoing analysis was performed to refine the specifics of each theme and checked that each theme captured unique features in relation to the overall research question. 6. Producing the report Produced a written report of the results https://doi.org/10.1371/journal.pone.0258121.t001 three non-binary participants Table 2. All participants had completed a high school education and were either currently enrolled in a post-secondary education program or had already completed one, and 82% of participants resided in an urban area. Qualitative thematic analysis revealed four domains of themes within the data: risk of disease contraction, perceived sociocultural impact of COVID-19, role of institutions, and effective public health messaging Table 3. Each domain has a number of subthemes that fall within the overarching theme. Each theme was reviewed in detail for additional elucidation and potential guidance on next steps regarding messaging and recommendations to government officials. I: Risk of disease contraction. While most participants believed they have been practicing COVID-19 safety measures as promoted by public health, such as keep small social circles and avoid unessential meetings, participants acknowledged that compliance to safety measures lies on a spectrum. One participant explained: "For me I think it's very concerning that a lot of people have a different mindset or different information that they're getting. The fact that it's dividing a lot of people within what is right and wrong is very concerning." Participants acknowledged the added anxiety from constantly assessing the risk of their daily behaviours. As individual perceptions of safety vary, participants expressed their concerns when their threshold of safety was different from others. Power dynamics were explored when participants had to weigh the risk according to their own safety standards and reject social invitations, or bear sacrifices in their lives. In particular, majority of participants felt they unconsciously judge others based on their compliance to public health orders. "I think that my friends are taking it seriously but I don't feel like they are meeting the standards that I would take, so that is a little bit awkward." II: Impact of COVID-19. Participants recognized the depth of impact COVID-19 has had on epidemiological, individual, and societal levels. Epidemiology. Participants gave uniform answers on knowledge of the virus origin, transmission, health consequences and treatments. They discussed the virus both from a immunological and geographic perspectives where one stated: "I guess some more information we could add is like the origins, where it started in China and it travelled across the world due to transportation not being shut down quickly enough." Participants also believed the virus was transmitted through small particles that were either airborne or through droplets. A few participants worried about how asymptomatic individuals could spread the disease without knowing they had it because of the long incubation period in which individuals do not develop symptoms. Many participants spoke on their knowledge of the virus: "It's in the same family of diseases as SARS and MERS, it mainly attacks the respiratory system, and it's spread through droplets predominantly." . "It is extremely contagious and requires people to protect themselves with facemasks or other like PPE." All participants recognized there are no existing treatments or cures for COVID-19 and acknowledged the difficulties faced by public health in finding a treatment or vaccine. However, participants agreed vaccines are deemed necessary for a safe return to normal lives, yet many still worried about the rapid development of vaccines. One participant explained: "I just hope that when the vaccine is actually released that it's gone through rigorous testing and its actually proven to be safe because I feel that the timeline right now like for vaccines to be developed is rushed." Individual. Participants expressed their concerns for long-term physical and mental health consequences, fear of contracting the virus as well as the stigma around contracting the virus. Lifestyle adjustments such as making decisions on safety, financial burdens, as well as the importance of family were also common topics within the impact of COVID-19 theme. One participant explained: "I don't drive, and I take public transit. So I just wonder how many active cases there are, and think, like, what are the consequences of doing this? What is the likelihood that I'll get sick? And what are the consequences of not doing this, like what are consequences to my mental health, my motivation?" Another participant described the sudden loss of financial income: "So, the entire month of April [2020], I guess about six weeks, actually, my husband and I were both without income. And he's a physiotherapist and he had recently opened his own clinic, so that sucked a lot. Anyway, yeah, so went from pretty decent income to zero overnight." Participants also mentioned their frustrations with the federal financial supports, explaining the inability to fit into the support criteria. While many also reported a loss of financial income during COVID-19, most participants discussed how not being able to take care of, see, and connect with their family members being the most devastating part of the pandemic. One participant explained: "I think I, just echoing everybody, it's definitely been the, the family interaction and how that's changed." "I'm not as concerned about myself, but like, I have elderly grandparents and my parents are older, and so, like, I know for me my biggest fear is, like, contracting it and not knowing and then giving it to other people and getting them sick." . Societal. There was an overall concern of the impact that COVID-19 will cause on the economy focusing on small businesses. Many believed more aid from the government should be available to support both the individual and other small outlets. One participant expressed: "I think, for me what, what really, like stresses me out with COVID-19 is how it's impacting our small businesses and how so many people are unable to get work or have lost their jobs. And just, just on a, purely, like, the economic fallout of it has been very stressful to me." III: Role of institutions. This theme focused on the three major institutions that participants identified to have clear responsibilities in the COVID-19 outbreak response: the government, the media, and the individual. Government. Participants believed it is the government's responsibility to mandate and enforce health orders set out by public health officials. More importantly, participants believed there should be more positively framed messaging coming from the government that encompass more emotional appeals to adhere to restrictions. One participant explained: "Instead of the government just kind of keep rolling the same ball, let's pick it up and throw it through the hoop. Like, have something, something different, and I think people would be a little bit more responsive to being like, okay, if we just buckle down and we, we work on this, you know, we can get through this. I think if the government had a different approach doing that way, I think we'll definitely be a lot more responsive." Media. Participants recognized media as a very important medium to deliver information. It was noted by participants in the focus groups they remembered seeing reports about people in their age group violating COVID-19 health orders. However, participants also believed that, at times, exaggerated and biased wording was being used by the media to describe the behaviour of young adults. When discussing how young adults were being portrayed on the news, a participant mentioned: "They are disregarding the fact that young adults are also the bulk of the working population who don't get the choice to go home and work in their home office are these same people. So yes, you know, the, the, there's accuracy in that it's more and more young people are driving infections, but there are more and more young people that have to be out. Have to resume life because they don't have the savings or the luxury to alternate their work schedule." The lack of balanced reporting of both positive and negative behaviours of young adults by the media left the participants unsatisfied. On the other hand, there were also concerns expressed on how both the timing and the content of the information being shared on mainstream media and the messages being shared by the government failed to align with one another. While they appreciated that the media must report stories as they happen, these challenges and conflicts ultimately raised questions regarding the motivations, values, and ultimately, the credibility of the news media as a whole. Individual. Participants acknowledged the responsibility of the individual to keep both themselves and others safe, and that only a collective individual effort will stop the pandemic. By staying at home when ill, maintaining physical distancing, and wearing a mask, participants believed practicing non-pharmaceutical interventions are a way to show willingness to protect others. "I think in theory, nobody would have to take responsibility for themselves if every single person took responsibility for others. However, obviously that's not happening, it's not the case." Participants believed each institution shared unique responsibilities to contribute to the control of spread, enforcement of guidelines and punishment for those who do not follow guidelines and the rebuild when the pandemic is declared over. Collectively, they believed that each group owned a shared responsibility, between the government, the media, and the individual, in stopping the spread of COVID-19 and, ultimately, the plan for a safe return to a quality of life which includes both economic productivity, meaningful social engagement, and improved health and well-being. IV: Effective public health messaging. The theme of effective public health messaging refers to the suggestions made to ultimately improve the effectiveness of the public health messaging surrounding the COVID-19 pandemic, vaccination, and return to meaningful daily interactions. During the focus groups, participants emphasized the importance of information seeking and receiving and identifying the current critiques of public health messaging. Information seeking and receiving. While participants acknowledged social media platforms such as Instagram, Facebook, and TikTok are often not trustworthy, they are the most popular medium young adults use for news, as well as information seeking and receiving on a daily basis. One participant explained: "Instagram is the one place I am going to go every single day regardless of whether or not I have the energy to look at the news and I do find the government is really slow to advertise messaging on that platform." Participants also claimed that after receiving information from social media they often cross reference the information by looking for credible sources through government websites, scientific articles, news networks and health authorities. Suggestions for effective messaging. Most participants voiced their concerns regarding the uncertainties in public health messaging and highlighted the quick changes in public health orders. For example, one participant mentioned: "Recently with public health saying to limit ourselves to one household and six safe people I think there's a lot of ambiguity around what it means to be safe." Many participants suggested utilizing multiple social media platforms to deliver public health messages and claimed messages would be stronger if presented with emotion and in a positive and creative manner. Responses regarding such suggestions included: "It needs to be geared towards the, the base positive feelings that all human beings have, which is the need to connect, which is the need to belong, the need to feel cared about, the need to feel valued." While statistical projections should still be provided, participants longed for messages that relay empathy in recognizing people's hardships. Participants also desired for one unifying message that instilled hope for a return to normal life while also motivating the public to adhere to restrictions. Finally, participants wished for contrary projections to the traditional negative bias of young adult behaviour with that of positive role models and the effects to be had if everyone committed to safe practices as a way of communicating a light at the end of a very long and dark tunnel that is the COVID-19 global pandemic. --- Discussion The current qualitative study examined young adults' perceptions of the COVID-19 pandemic and the effectiveness of current public health messaging. Overall, our participants, contrary to mass media coverage, were highly concerned about the spread of COVID-19 and perceived it as their responsibility to protect themselves and their loved ones. Similarly, recent research found younger adults have a higher perception of risk vulnerability with respect to COVID-19 [14,19]. Young adults in our study highlighted the difficulties they faced and resiliency they demonstrated in dealing with COVID-19. Furthermore, they expressed concerns and confusion around current public health messaging and provided valuable insights for future effective messaging on COVID-19 as we enter the next phase of the pandemic-vaccination. --- The power of positive messaging A heightened level of stress and anxiety was noted to be felt consistently through discussion among participants. Similarly, young adults in other studies were also found to be moderately less optimistic about the pandemic's outlook and perceive higher levels of risk vulnerability than older adults, as well as anxiety and depression [14,[19][20][21]. Previous research also suggests young adults are more easily stressed if health threats are perceived, leading to long-term negative mental health consequences such as anxiety and depression [22][23][24][25]. Young adults also have experienced far greater psychological distress symptoms during the pandemic than older generations [26]. Stress may be further exacerbated by lower savings, less occupation stability, and taking on responsibility as a caregiver [27]. While understanding public cooperation is required, our participants identified that repetitive messaging of health orders and restrictions with no positive projection for the future could easily lead to information and compliance fatigue and therefore threaten individual trust for the government. While older adults may have more perceived emotional regulation strategies such as mature social networks that allow them to focus on emotions to reduce adverse effects, young adults may require active and ongoing exposure to positive messaging in order to generate positive attitudes and emotions [19,28]. Most existing public health messaging focuses on the collective good using inclusive language, providing ideas to social safety, or emphasizing the risk of contracting COVID-19, which may elicit fear [29]. There is a reluctance to incorporate positive messaging because of potentially creating perceptions of false security, though available evidence does not support this concern and suggests compensation is not discernible at a population level [30]. Our data suggest not providing positive outlooks risk alienating and disengaging this age group linking to long-term pandemic response success. Instilling hope during a public crisis is a powerful tool in combination with fear [29,31]. An instillation of hope, besides mathematical modeling, should include the acknowledgment of the public's sacrifices and summarization of the public's response since the start of the pandemic to flatten the curve [32]. For example, Angela Merkel, Chancellor of Germany, used science and a clear explanation of the disease reproduction number and modeling to discuss an exit out of lockdown. While our participants agree it is often difficult to change behaviours of those who hold on tightly to their beliefs, an installation of hope through positively framed messaging may bring people together to facilitate collective change. --- The power of being heard Almost all young adults who participated in focus group discussions were aware of media coverage on events caused by people in their age group that resulted in community spread. Yet, many raised questions on the accuracy and intentions of the reports as the biased wordings in many media messages framed young adults in a shameful way. Shaming is an inefficient method for behaviour change as it entrenches polarization, and discourages compliance [29]. On the other hand, messaging that shows an understanding of people's situations, sitting in someone else's shoes, effectively elicits compassion lead to corporations among the public [29]. During the time of the focus group, the province had been in a state of emergency for approximately eight months, with changing public health orders restricting social gatherings, traveling, and indoor activities. The province had been in full lockdown since early spring of 2020. While the provincial restrictions had relaxed for the summer of 2020 following the province's restart plan, the declaration of a second wave in October amended further strict regulations on gatherings. Furthermore, the provincial health officer issued two-week regional lockdown measures, including restricting all social gatherings of any size. While public services such as schools remained open, local small businesses such as restaurants and dance studios suspended operations according to changing health orders. A provincial mask mandate was issued shortly after the focus groups were conducted. Young adults are more likely to rely on public transportation, work in essential service positions, and live in houses with multiple roommates that reduce cost but also increase exposure to the virus. To alleviate these unique group of stressors, providing an official platform for young adults to share their obstacles and lived experiences had been utilized in other nations. For example, in New Zealand Jacinda Ardern has live-streamed her conversations with regular New Zealanders to share their stories and advice in "Conversations through COVID". It is notable that numerous participants reached out to us following completion the study to express their gratitude for an opportunity to communicate their thoughts and feelings. --- Two-way communication The varying interpretations and confusion around BC's public health orders was a reoccurring topic of conversation amongst our participants. Inconsistency between media reports and public health reports was a persistent issue. For example, participants had questions understanding the change in definition of a "close contact" and were interested in knowing what they can do instead of what they cannot do. Following the change in health orders from "safe six" , to "core bubble" , some participants were unaware of the change and for those who were aware, many questioned why inperson interactions such as at schools were still allowed. Public health research also found messages that came out first tend to be believed the most strongly [33]. Delays and uncertainties in understanding public health orders suggest improvements need to be made to increase the reach and effectiveness of public health messaging-regardless of the content of the message. Our results indicate the mediums young adults use to actively seek for information is often different from the mediums they interact with the most. Therefore, the optimum strategy would be providing credible information on mediums that young adults interact with the most. Indeed, effective approaches to young adults in public health messaging have been complex in past health initiatives for communicable and chronic diseases including sexual health promotion, and the H1N1 outbreak [23,34,35]. However, tailored messages on a medium that is convenient, one that is highly accessible and widely used was found to be most effective in engaging youth [23,[36][37][38]. Social media was suggested as a frequently used medium and a widely accessible platform where young adults are comfortable sharing their concerns. Efforts have been made both locally and internationally to share credible information on social media, where users are perceived to be engaged in two-way communication [39]. The extended social networks that stem from social media can amplify positive behaviours, especially when central figures such as influencers spread positive health behaviours like handwashing and physical distancing. Previous health interventions had leveraged social media to deliver tailored messaging as young adults are mainly digital health users with high digital literacy [40]. Our participants had specifically mentioned following individuals that are active in science communication, such as Dr. Samantha Yammine, on social media. Similarly, other health care professionals have been active in delivering credible information on social media, such as Dr. Christian Drosten, a leading virologist who launched a podcast to explain the science behind the virus and the latest research [41]. --- Strength and limitations Strengths of this study include providing a unique young adult perspective by investigating the lived experiences of young adults using qualitative analytic strategies resulting in concrete themes for consideration when developing future public health messaging. Moreover, this paper suggests innovative ways to connect with this age group such as facilitating two-way communication using social media. All participants had some degree of post-secondary education which limited our ability to gain perspectives from individuals who are less educated or with limited language ability and technological access. Participants may also be influenced by social desirability factors, and withheld information if they failed to comply to public health orders. --- Conclusion Our findings suggest messages that reach young adults should 1) have a positive framing, 2) reflects lived experiences of this age demographic, and 3) delivered on an accessible platform Contrary to popular belief, young adults perceive a high level of concern and stress regarding the pandemic to keep themselves and their loved ones safe. While young adults may have questions and concerns towards public health messaging, a traditional broadcasting one-directional communication strategy makes voicing concerns difficult. Respectfully, we urge stakeholders including government officials and media outlets to head these recommendations, reporting and creating messaging that answers young adults' concerns. Tailored messaging is needed, desperately. --- The data used in this study pertains to transcripts from interviews, some with personal experiences. Therefore, we are not able to share the transcripts Participants were not informed or requested to allow for the data to be shared publicly. The Simon Fraser University Office of Research Ethics can be contacted at 778-782-3447 .

The COVID-19 global incidence rate among young adults (age 19-40) drastically increased since summer 2020, and young adults were often portrayed by popular media as the "main spreader" of the pandemic. However, young adults faced unique challenges during the pandemic due to working in high-risk, low-paying essential service occupations, as well as having higher levels of financial insecurity and mental burden. This qualitative study aims to examine the attitudes and perceptions of health orders of young adults to better inform public health messaging to reach this demographic and increase compliance to public health orders. A total of 50 young adults residing in British Columbia, Canada, were recruited to participate in focus group in groups of four to six. Focus group discussions were conducted via teleconferencing. Thematic analysis revealed four major themes: 1) risks of contracting the disease, 2) the perceived impact of COVID-19, 3) responsibility of institutions, 4) and effective public health messaging. Contrary to existing literature, our findings suggest young adults feel highly responsible for protecting themselves and others. They face a higher risk of depression and anxiety compared to other age groups, especially when they take on multiple social roles such as caregivers and parents. Our findings suggest young adults face confusion due to inconsistent messaging and are not reached due to the ineffectiveness of existing strategies. We recommend using evidence-based strategies proven to promote behaviour change to address the barriers identified by young adults through tailoring public health messages, specifically by using positive messaging, messaging that considers the context of the intended audiences, and utilizing digital platforms to facilitate two-way communication.

Introduction The police interview serves as a context wherein the central activity involves the reconstruction of preceding events, and a significant tool for the police to initiate and engage in this activity is through the use of questions [1,4,50]. The distribution of access to prior events under investigation frequently exhibits discrepancies among participants, since victim, suspect and witness are likely to have first-hand information to the events under investigation [cf. 26,33,46,47], whereas the police and other institutional representatives often have to rely on concrete evidence for the reconstruction of them. As a result, the police's understanding of the alleged crime depend upon testimonies provided by witnesses, which in turn make the interview situation a crucial link in a police investigation. Details crucial for the investigation may reside in the interviewee's memory, making elicitation as well as recollection of memories recurrent and frequent interactional practices in PIs. Asserting an absence of memory or stating other hindrances in recalling preceding events has been reported to be a potential strategy in the interview situation [28], which recurrently has been studied with a primary focus on cognitive aspects [52,53]. Nevertheless, scarce empirical focus has been directed towards how knowledge of previous events is embedded in talk about memories and remembering, and how this is interactionally managed in high-stake PIs [51]. This is the focus of the current paper. Significant contributions that explore language use in PIs have emerged from the domain of Forensic Linguistics, which brings together research that focuses on the "investigation and elucidation of language evidence in a legal context" . One early study, still frequently referred to, is Loftus & Palmer's classical study from 1974 [27] in which they show how witnesses reported different speed estimates of a car in a car crash dependent on which verb was used in the question asking about the car's forward motion. For instance, the verb smash generated a higher speed estimate compared to the verbs collide, bump, contact or hit, in the question 'About how fast were the cars going when they smashed into each other?'. Over past decades, studies investigating interactional features, employing either a sociolinguistic or conversation analytical framework, have expanded our knowledge on both language use and how social interaction is organized in PIs. For example, research has focused on topic management and turn-taking as tools for pressuring the suspect [29], how suspects' rights are explained and exercised [32,51], how questions constitute a linguistic tool for the police in the PI [1,4,50], and how the use of guidelines may facilitate or hinder fair questioning and equal treatment [22,37]. However, studies have primarily been focusing on English speaking contexts . As stated by Haworth [17], conditions and results are contingent upon cultural as well as social factors, emphasizing the need for research drawing on data from other social and cultural contexts, such as Swedish high-stake interviews, which still only have been examined in a few studies [3,23]. To increase our understanding of interactional processes in this specific context, the present paper employs 1 3 "I don't Remember that": Negotiating Memories and Epistemic… Conversation Analysis [41] to explore interactional environments in Swedish high-stake police interviews where the interviewee's ability to recall, as well as the police's challenge of the interviewee's claimed incapability to recall is managed in interaction. Within the realm of Conversation Analysis, focus has been directed toward examining how participants handle epistemic dimensions in social interaction, as these have been found to be omnipresent [19,46]. In conversation, speakers display a sensitivity towards who knows what as well as who is entitled or ought to know what, and their utterances reflect this orientation, as shown in previous research . This research sheds light on a fundamental aspect of human interaction, at the same time as it contributes to explore how norms are established and renegotiated in different institutions, as patterns of how knowledge is managed vary depending on contextual factors such as participants' roles and mandate [20]. Increased knowledge of such helps to expand our comprehension of fundamental organisations of our society. Previous conversation analytical research on epistemics form the theoretical foundation for this study, which explores how participants of social interaction handle epistemic elements in the specific environment that the police interview constitute. The overarching objective of this study is to enhance the understanding of interactional procedures utilized in high-stake interviews conducted in Sweden, and in particular, to explore sequential environments in which claims of not recalling are displayed by the interviewee and challenged by the police, as they manage epistemic claims related to the event under investigation. Data used for this study contains anonymized audio recordings recorded during a preliminary murder investigation . The investigation is conducted 20 + years after the suspected crime was committed due to new forensic evidence that may link the suspect to the scene of the crime. A majority of the witnesses in the data, as well as the suspect, have been interviewed at different times in the past. Since more than two decades have passed since the event, the time frame is recurrently constructed as an omnipresent factor by both the police and the interviewees in the interviews. The study focuses on how this time frame is incorporated in the police's questions, and in turn, what interactional space this creates for the interviewees as well as how it becomes a potential resource for the interviewees when constructing an "I don't remember/recall"-response. An important result from the study is how a shift in focus of attention from the actual events under investigation towards the ability to remember or recall is facilitated by means of the question formulations. To expand our understanding of the described interactional procedure, more studies including data from other cases and contexts are needed, as this study explores one specific case. --- Access to Previous Events-Dimensions of Epistemics in Social Interaction In the police interview, the primary institutional goal for the police is to gather information from the person they interview in order to be able to clear up the case under investigation. This means that epistemic dimensions permeate most facets of the police interview, as the police works to identify who knows what and to make them talk about what they know. Depending on the role of the interviewee in relation to the alleged criminal activity, they may align with or diverge from this objective. At times, the interviewee might resist cooperation, possibly owing to their own involvement. In different scenarios, they might intend to provide information; however, factors like trauma, stress, or threats can render this endeavor challenging. In any event, knowledge on the asked about event is central, and in this section, studies targeting how knowledge are managed in interaction are summarized, as they form a backdrop for the reported study. A prominent study, recurrently referred to in work on epistemics, is Labov & Fanshel's study [26] on interaction in a psychotherapy session. Their work demonstrates "what is being done" interactionally in the analysed session, with particular attention paid to how propositions are embedded and launched into interaction. Labov & Fanshel propose a model of how speakers relate to knowledge in terms of events, and these are classified based on the speakers' knowledge about them, such as A-events: known to A and not to B, B-events: known to B and not to A, AB-events: known to both A and B, and D-events: known to be disputable. Speakers design their utterances, as well as interpret others', depending on how they perceive speakers' relation to these events. For instance, Labov & Fanshel suggest that "[i]f A makes a statement about B-events, then it is heard as a request for confirmation" . This rule emphasises how participants are continuously oriented to what type of knowledge their co-speakers have, as a statement may be treated as e.g., either a request for information or a request for confirmation, depending on the perceived access and distribution of speakers' knowledge. As shown by Cerović [5], B-events can be used as interactional vehicles by the police in police interviews as either confirmation seeking questions, or as accusations, depending on the local interactional environment. If the IE is cooperative, B-events in confirmation seeking question are charged with lower epistemic claims than in sequences labelled as 'hostile', where amplified epistemic claims in B-events are treated as accusations. Pomerantz [33] has also described speakers' access to events in terms of Type 1 and Type 2 knowables, which distinguish between who has primary and direct access to an event , or what a speaker has learned second hand. Depending on the degree of access to the event, speakers may be subject to expectations displaying different degrees of obligations and responsibilities. For instance, an eyewitness may be expected to be able to account for an actual event, whereas information provided via a telling constitute a Type 2 knowable, consequently providing reduced degree of obligation to account for the original event. How access to information may be employed for eliciting further information is described by Pomerantz in a later study [35], in which she examines how speakers propose a possible response by offering a candidate answer, embedded in the original question. This is argued to function as an information seeking strategy, and it also reflects epistemic dimensions as it displays the speaker's own knowledge -and access, of the asked about event. As shown by Pomerantz, this strategy is deployed in interaction recurrently, and as found in Sect. 4.2 below, it is also employed by the police in the PI for information elicitation . 1 3 "I don't Remember that": Negotiating Memories and Epistemic… Stivers, Mondada and Steensig [46] offer a description of the epistemic layers in interaction by proposing three dimensions of knowledge: epistemic access, epistemic primacy and epistemic responsibility. They argue that speakers display orientation towards who knows what , as well as the relative rights and authorities to know what , and lastly who is expected to know what . Speakers exhibit sensitivity to these dimensions and shape their utterances accordingly, so that information likely to be known by the recipient is not framed as news [16], and claims are reduced if the recipient is expected to have epistemic primacy to certain information [21]. Yet a different terminology is used by Heritage [19], introducing the notion of an epistemic gradient, where speakers are placed based on their degree of access to a certain epistemic content . This way, speaker and recipient may be constructed as either K + or K-, and this is displayed in the turn design. Heritage argues that epistemic stance is encoded into the utterance through various linguistic resources, and function to create space for speakers to e.g., elicit information or confirmations, depending on how and what type of relation to certain knowledge is constructed in any given utterance. In contrast to Labov & Fanshel, Heritage states that two speakers rarely ever share the same access to an event , as their experience and status always varies. As this bulk of research shows, speakers monitor and adjust the design of their utterances reflecting their own and their co-participants' perceived knowledge distribution. However, the majority of examples from the cited research are derived from interactions among individuals acquainted with each other. In contrast, the police interview offers a setting in which much is at stake, and the entire enterprise is about detecting the actual 'knowable'. In this context, claims of not knowing can be deployed for serving multiple purposes. As stated by Keevallik [24], a 'no knowledge'-response can be a resource for providing a response in the response slot, while still rejecting the terms set up in the previous turn. In non-institutional settings this can be an item used for resisting the agenda formulated in a first pair part of an adjacency pair. This has been found in research on couples therapy, where claims of not remembering can function as a resistance or avoidance of the therapeutical agenda [31]. However, in an institutional setting where one party has the mandate to strongly control the interactional agenda, it may be more difficult to claim lack of knowledge, which has been found in studies on regulated settings such as courtrooms [8,25]. Claiming to have no knowledge can also function as an interactive resource when framed as 'forgetfulness' [16]. Goodwin describes how this resource provides an opportunity for speakers to rearrange the participation framework, when for instance switching from the discursive identity of listener to a story, to that of a co-producer of the same story. This may also invoke a co-participant in an 'identity relationship' such as a couple, marking shared and common knowledge. This way, the construction of knowing and unknowing recipients provide opportunities for speakers that may not primarily be connected to what is actually known and unknown information. In the analysis, Goodwin also shows how syntactic and intonational resources form pivotal features for shaping different activities, such as requests. For example, a wh-question may construct a recipient as knowing, which in turn may open for the knowing recipient to take a telling in a new direction. This can be pre-empted through the use of a request for confirmation, which instead forms a limited space for the knowing recipients to contribute to. In conclusion, the display of access to knowledge may serve various interactional purposes, as well as creating opportunities for how actions and sequential trajectories are created momentto-moment in social interaction. To 'gather' or 'elicit' information convey the notion of a neutral effort, but in police interviews, there are several factors that come into play and make it a complex one. Cognitive psychologists have predominantly focused on studying the retrieval and management of knowledge stored in memories during police interviews, and although many of the studies target how memories are formulated [e.g., 40,42,54,55], language is often treated as a product rather than a social activity. One such example is how transcripts are often used as the preliminary source of data, instead of the video-or audio recordings, which include dimensions of social behaviours such as gaze, intonation, and prosody [2, p. 1441]. As demonstrated by Goodwin [15,16] multimodal cues like tracking the gaze of co-participants can clarify instances of self-repair, which in a transcript might resemble a hesitation. On the other hand, studies with discourse analytic foci have revealed that memories, akin to other cognitive processes, must be regarded and comprehended as resources within interaction, shaped by the social as well as the interactional context [10,30]. Goodman and Walker [14] studied 'memory talk' related to instances of alleged intimate partner violence, i.e., how memories related to these events was talked about in interviews with male perpetrators. They reports how the memory talk functions to manage accountability and identity, as claims of not remembering were employed when talking about their own actions, potentially disclosing the exercised violence. A few studies have investigated the use of "I don't know" in the police interview setting, using transcripts as data. This includes Earhart et al. [9], who studied how ground rules formulated for "I don't know"-answers are formulated in interviews with allegedly abused children. As children might be inclined to provide responses to adults' questions, pre-interview instructions regarding this are sometimes used. Their study challenges prior experimental findings, which suggested that children were more inclined to employ "I don't know" responses when instructions permitted such usage. They also report that children rejected questions with "I don't know"answers in almost 30% of the cases, and these rejected questions were often followed by follow-up questionsto which children provided answers. In conclusion, previous studies highlight that speakers consistently attend to epistemic dimensions, rendering epistemics a fundamental facet of social interaction. Claiming to have no knowledge and/or to not remember have also be shown to operate as interactive resources, serving different purposes for speakers as they may, for instance, configurate the participation framework. --- Data and Methodology Data collection was conducted for the project Interactional Patterns in Swedish Police Interviews, which received funding from the Swedish Research Council [2021-03044], spanning from 2022 to 2024. As a precursor to this project, a pilot 1 3 "I don't Remember that": Negotiating Memories and Epistemic… study involved interviewing eight police officers regarding their roles as interviewers and the training they had received. Findings indicated that training methods varied based on rank and specialization, and it was revealed that not all police interviews are being recorded in Sweden. Typically, the police do not record interviews during investigations of routine crimes; however, more serious offenses like murder or crimes against children are audio-and/or videorecorded. The dataset comprises 51 anonymized audio-recorded interviews involving suspects and witnesses linked to a single case. The dataset underwent an internal confidentiality review by the Police authority and all personal information were removed before the interviews were released to the project researchers in a secure digital environment. The police have been offered to take part of the findings. Police interviews raise ethical concerns due to their potential depiction of traumatic situations and their potential to expose individuals in vulnerable circumstances. However, in order to increase the knowledge on interview techniques within the police profession, real high-stake interviews must be examined. The conversation analytical perspective specifically centers on the patterns of social interaction as significant social achievements, thus shifting the focus away from the participants as individuals. The police and the interviewee jointly shape the interaction, and it is this dynamic interaction-rather than the individuals-that constitutes the analytical focus. How the interaction is shaped and managed is viewed as a reflection of how this type of situation, as well as the particular institution, is "talked into being" [20, p. 290], rather than how specific individuals conduct a specific interview. The interviews collectively comprise around 27 hours of recorded material, and they include interviews with several different participant categories, such as suspect, witnesses and forensic pathologists. In total, the police interview 37 individuals, and there are four police officers alternating in pairs conducting the interviews. By the time of the interviews, between 20 and 25 years had passed from the crime under investigation. In the transcripts and in the article, 25 years is used as a time reference. The data was analyzed employing Conversation Analysis , which provides tools for investigating how participants in interaction are socially oriented towards each other and how norms and institutions are managed and maintained through social actions [41]. CA originates from ethnomethodology, which fundamentally seeks to explore how participants achieve intersubjectivity in interaction through social actions. Garfinkel [12] stated that mutual understanding is not just something that happens by itself, it is accomplished through language and other social actions, and it is thus visible, hence, possible to study. Since language is viewed as a tool for socialization and the primary resource for conducting social actions, CA focuses on the fundamental organization of talk. Initially, interaction in mundane settings were the primary target, but during the last decades attention has shifted towards interaction in institutional contexts. For instance, research has focused on explicating the management of interactional practices and examining the constitution of core activities in specific contexts [20]. In the realm of Forensic Linguistics, the use of CA is growing, given its capability to closely analyze how participants establish and negotiate social actions, such as complaining, defending, accusing, explaining. Through close analysis of the micro level of interaction and the participants' expectations and orientations, it can be revealed how norms are enacted and established, hence uncovering how institutions are "talked into being" [20, p. 290]. A recurrent theme found in many of these studies is how the participants organize and accomplish questioning and answering. These social actions compose two components of an adjacency pair that form a basic foundation in many interactions conducted in institutional contexts, through which the interaction is shaped and formed. By studying how questions and answers are managed, social and institutional norms become visible, as well as how they form conditions that the participants have to comply with. This does not mean that institutional norms are fixed, rather that they create frames that the participants orient towards, at the same time as they are being renegotiated in and through interaction [20]. The examples featured in this paper have been chosen from a collection of instances extracted from data where the interviewee asserts an inability to remember/recall, and this assertion is questioned by the police. The showcased examples were specifically selected due to their potential to illustrate a recurring characteristic. They serve to demonstrate how this recurring feature is enacted by the participants in this particular case. Further studies that utilize diverse datasets are encouraged in order to comprehensively grasp the phenomenon described. In the transcripts, 'P' refers to the Police officer, 'S' to the Suspect and 'W' to the Witness. In the analysis, however, 'IE' is predominantly employed when referring to either the suspect or any witness, since the analytic focus is on the participants' discursive roles and how they mutually manage the interview situation as an interactional project. In the analysis, their institutional roles are subordinate to their discursive ones , but for the sake of transparency the transcripts reflect this information. --- The Case The case being investigated in the collected data involves a murder, with the alleged incident occurring over two decades prior to the investigation. Due to advancements in forensic technology, the police reopened the case, conducting a new investigation 1 3 "I don't Remember that": Negotiating Memories and Epistemic… that involved new interviews. Through new analyses of historical evidence and the integration of witness testimonies, the police linked an individual to the crime scene. This person was subsequently acquitted in court. --- Remarks Regarding the Translation The focus of analysis is instances in which the interviewee states "I don't recall" or "I don't remember". In Swedish, the verbs used are komma ihåg and minnas . As per to The Contemporary Dictionary of the Swedish Academy, komma ihåg means "being able to retrieve [something] from memory", and minnas "bringing stored mental concepts to the forefront" . Similarly to English, these verbs can be used interchangeably. The distinction is akin to the one in English: for instance, you are more inclined to remember rather than recall fond memories from your childhood, while you might both remember and recall where you parked your car in the parking garage. In everyday Swedish, they are interchangeable, and both are employed within the dataset. --- Analysis As stated above, the number of years passed since the crime was committed is treated as an aggravating circumstance by both the police and the interviewee throughout the data. In this initial section of analysis, instances of how this is achieved in interaction are provided, demonstrating the construction of time as an omnipresent element within the data, serving as a potential obstacle to recollection. This section establishes a foundation for the subsequent one, wherein instances are examined in which interviewees formulate responses asserting their inability to recall, prompting various forms of challenge from the police. --- Making Time a Relevant Factor When Prompting Recalling Based on the recorded interactions, it becomes evident that nearly all interviewees have either received a written summons or a telephone call from the police, announcing that they are called for police questioning, which they must attend. Only the suspect and a crucial witness have been apprehended by the police without prior notification before the interview. At the outset of most recordings, the reason to why the interviewee has been summoned is reiterated for the benefit of the tape [cf. 48], prior to the police commencing the question-and-answer phase of the interview. In this initial phase leading up to the first question, it is common that the interviewing police officer provides a starting point for the interviewee. This establishes the groundwork for the initial question, and frequently, this phase incorporates a mention of the time frame. This is exemplified in the subsequent instances : Example 1 1 3 "I don't Remember that": Negotiating Memories and Epistemic… --- Example 2 In Example 1, the police officer produces several statements prior to initiating the first question line 10. In line 5, the police officer introduces the time frame by "it's been twenty-five years now". The witness responds promptly with an affirmative "yes", accompanied by laughter. This is followed by "long time ago, yes", also produced with laughter in the voice, indicating that this is a fact that is not entirely neutral [cf. 13]. These turns, produced by P and IE, contribute to setting a scene for the events to be discussed. Thus far, the laughter stands as the sole indicator of the time frame being viewed as a potential obstacle. However, the police officer's second question in line 12 alludes to this perception as they inquire "if you could try to recall what happened that night". This formulation imbues the question with the presupposition that recollection might present challenges. 'Try' implies that an effort is needed, which displays a stance from the police officer that acknowledges potential difficulties with the request. From a conversation analytical perspective, this illustrates how a participant in interaction can formulate a request so that it enables what is referred to as a dispreferred answer [34], reducing the effort it may take to reject the request in the subsequent answer. IE might be unable to provide an informative response, due to difficulties with retrieving the requested information -despite efforts of trying. By employing this question structure, the emphasis shifts to the potential challenge of remembering, thereby placing the recounting of events in a secondary position. This orientation aligns with the established acknowledgement of potential challenges in turns 5-7. Moving on to Example 2, the initial question addressed to IE is presented in lines 12-13. This question also involves a verb phrase indicating the potential complexities related to the desired action . There is no immediate uptake from IE, instead a lengthy pause of 2.0 s is found in line 14. This causes the police officer to take the turn in line 15 with "I'm aware it's a long time ago". This could be interpreted as the police officer treating the silence in line 14 as a sign of the prior request being difficult, hindering IE to launch a response, hence causing the silence in line 14. In contrast to Example 1, the IE is in this example, prompted to present the narrative "as accurately as possible". Retelling events that lie more than twenty years back in time may be difficult, but to also do it accurately suggests further implications, particularly in this context where the truth is a central, yet often an implicit, factor. Example 3 exemplifies a comparable procedure to the one identified in Example 2: --- Example 3 Turning to Example 3, the police officer starts by stating that the witness has undergone previous questioning. This directs attention to a specific time period, dating back over two decades. Subsequently, in line 4-5, the police officer asserts, "I do understand if it might be difficult to recall and so on", resonating the initial question in Example 1. A similar formulation is produced after the pause in line 1 3 "I don't Remember that": Negotiating Memories and Epistemic… 14, Example 2, potentially due to the police officer interpreting the lack of uptake in the possible speaker exchange slot as a sign of the witness' difficulty regarding producing a response. In examples 1 and 3 the time frame is established as a potential challenge for recollection. This positioning occurs within a preliminary sequence [38] that leads up to the question, forming the first component of an adjacency pair. While not delving into the second components of the adjacency pairs, which harbour insights into the interviewee's interpretation of the initial queries, one can still deduce that the context surrounding the launch of the first question configures the act of recollection as a potentially intricate endeavor, at the same time as it pushes what happened further into the question. Through the design of the question, the police officer situates IE in a position of possible limited access to the relevant event, while also acknowledging the plausibility of being unable to recollect. This situates IE in a position categorized as K- [19], where the prospective knowable [33] remains unspecified in terms of being a Type 1 or Type 2 knowable. Examples 1-3 also showcase how the police imbue the questions with diminished responsibility, thereby projecting low epistemic assertions. As previously mentioned in connection to the description of the data, the police possess evidence linking the suspect to the crime scene. The police can potentially employ this information strategically across various interviews. However, the methods for such strategic deployment, as well as potential inconsistencies regarding e.g., provided details across different interviewee categories, extend beyond the scope of this study. What is evident in this study, however, is how the police set the frame for the interviewees by means of how the initial question is designed : --- Example 4 In Example 4, the initiation of the interview's inaugural question is depicted in line 1-2. This sequence contains a clarification of what type of interview is being conducted and a brief instruction of how to respond to the questions . This contrasts this interview from the previous instances. The initial query is followed 1 3 "I don't Remember that": Negotiating Memories and Epistemic… by two more questions before P leaves the floor to IE. This opening phase diverges moderately from those presented in Examples 1-3 in terms of question order and the reference to the elapsed years. Nevertheless, the mentioning of the time frame occurs here, potentially constructing its relevance. Yet, the way in which it might operate as an influential factor remains unarticulated by the police. All three questions are also formulated with 'could', implying a limited level of epistemic access and, consequently, a diminished expectation of response capacity [19]. Examples 1-4 illustrate how the police employ the initial phase of the interview to underscore the significance of the considerable time elapsed since the investigated events occurred. Furthermore, numerous initial questions are formulated with phrases like "if you could try", coupled with reference to the time frame. This construction contributes to cultivating a diminished level of anticipation concerning the interviewee's capability to furnish responses pertaining to the subject event. This may constitute a tactic employed by the police to build rapport [see e.g., 53] and foster a sense of ease for the interviewee. However, this introduction also serves to situate the interviewee as someone potentially deficient in, or at least possessing limited access to, the event. In essence, the police create a context where the interviewee is invited to adopt a K-position, given that the knowable are positioned behind a temporal barrier-a challenge that the interviewee is prompted to surmount. The interviewee is summoned to the interview with with the anticipation that they are 'A-eventers' [26], possessing Type 1-knowable [33]. This implies expectations associated with the interviewee's role [cf. 56], and the instances demonstrate how, in the interview's preliminary phase, the police diminish this expectation. Additionally, the data unveils an apparent anticipation to remember, as depicted in Example 5, where the police officer commences by specifying the years during which the interviewee had been previously questioned: --- Example 5 In line 9, IE's statement "I don't recall that", offered in response to preceding utterances, conveys their perspective on the information presented in prior turns. That the sotto voce , indicated by the circles in the transcript, might convey a sense of hesitation is confirmed in uptake by P in line 10-11, as P states that "it's been many years, it's twenty-five years since it happened". Thus, the police officer invokes the time frame as an explanatory factor and a potential mitigating circumstance for IE's stated inability to recall. While the connection is not overtly stated, the sequential arrangement illustrates how "twenty-five years" is regarded as a rationale for the failure to recall, as it is placed directly adjacent to the turn where non-recollection is expressed. Subsequently, a "yes" is offered in line 12, affirming P's interpretation. The statement "I 1 3 "I don't Remember that": Negotiating Memories and Epistemic… don't recall that" serves to establish a K-position concerning an A-event, a stance validated by P as an admissible perspective. The final example in this section demonstrates IE's anticipations concerning their own performance, as they evaluate their own performance in relation to what they were able to recollect: --- Example 6 In Example 6, IE clearly demonstrates an anticipation of furnishing information related to an event queried by the police. The possible deficiency in performance is articulated in connection to memory, fashioning the potential limitation as a trait one might possess, rather than, for instance, being characterized as forgetfulness. Furthermore, as highlighted by Speer [43], self-deprecation creates an opening for co-participants to counter the critique; in line 8-9 + 10 the police officer offers a favorable assessment of IE's performance. To this point, the examples reveal that both police officers and interviewees are treating the time frame as a probable obstacle for evoking memories. As discussed earlier, this might function to facilitate rapport-building, however, simultaneously, it establishes a frame that diminishes expectations with regard to the pivotal task at hand: recollecting past events. While doing so, the police underscore the significance of recalling as a high-stakes activity central within the context of the interview, while concurrently projecting it being a task afflicted with potential difficulties. --- Treating "I Don't Recall/Remember" as an Insufficient Response This section focuses on interactional contexts where the interviewee's use of "I don't recall/remember" is treated as an inadequate response by the police. Examples illustrate the initiation and framing of IDRs, its subsequent challenge by the police, and the participants' departure from the sequence. In the first example, the police are inquiring about a specific car and potential access to its keys. The police have evidence tying the suspect to this car in connection with the time of the crime. In Example 7, an interview with a car owner unfolds, starting with the police officer introducing a question about the car keys in line 1: 1 3 Example 7 1 3 "I don't Remember that": Negotiating Memories and Epistemic… The turn in line 1 introduces the subject of car keys, and in lines 3-5, P presents what Pomerantz [35] refers to as a candidate answer. In this case, the police officer proposes a potential circumstance that would explain how someone other than the owner of the car was driving it. As argued by Goodwin [16], this creates a defined slot for the recipient, as the question embeds a candidate answer into the question . In line 6, IE responds in sotto voce "I don't remember". Through recycling IE's IDR-answer in line 7, P creates an opportunity to halt and direct attention to the given answer. While the police officer doesn't overtly contest IE's IDR response, the repetition with an interrogative tone renegotiates the IDR as a final response to the preceding question. Instead, the police officer produces a first pair part, projecting a second pair part that forces IE to reside on the topic, indicating that the provided response is treated as dispreferred [34]. This in turn, forces IE to respond again to the initial question, which is done in line 8 as they respond "no" with what can be referred to as a smiley voice, i.e., with voice qualities bordering laughter, such as respiration. This answer is treated in a similar manner, as P repeats this answer as well . In line 10, IE briefly develops the response by adding "it's not possible", which serves as an assessment of the requested recalling. This resonates with Edwards & Potter's [11] exploration of subject-side and object-side assessments, seemingly interchangeable but serving different interactional functions as the former formulates "a disposition of the speaker toward that object" , where the latter is used more like a description. The IDR in line 7 displays the speaker's stance towards the proposed facts in lines 3-4, that they cannot recall if this was the case. The "it's not possible" forms a more objective viewpoint, and it indicates that IE treats P's repetition as a request for an account. IE does not remember because this is objectively impossible. This is picked up by the police officer in line 11, who confirms that "it's a long time ago". This way P affiliates with IE's stance, invoking the time frame as a motive for not being able to remember. However, the police does not fully affiliate with the impossible part, as they add "it's not easy to remember". Not easy displays an understanding of the difficulties with recalling circumstances related to the keys, but as seen in what follows, it also projects that the subject is not exhausted, and IE's effort is not considered complete on this matter. Simultaneous to the talk in lines 13 + 15, the police officer displays an image of the car to IE. During this, they shift the attention away from car keys and emphasize the car itself. From what can be gathered from the audio recording, the witness looks at the picture and responds, "yeah that's my car", confirming what has already been established earlier in the interview. To this, the police officer responds "that you do remember", with a word order that stresses the initial object , contributing to emphasizing the police's orientation towards IE's ability to recall: they may not remember the keys-but at least they remember the car. This produces a new basis in the interaction, as the police continue to ask questions about the car. This sequence does not demonstrate an explicit challenge of the IDR, however, by repeating IE's IDR-answer, the police create an opportunity to halt and engage attention to this specific answer. As becomes visible when examined in close detail, the police treat the IDR as a dispreferred response, which is mildly challenged through the recycling and the repetitions that forces the witness to re-confirm the response. As we will see in the next example, the police have not abandoned the issue of the keys. The following example, collected from the same interview takes place about nine minutes after Example 7 has ended. Here, the police officer has, after some digressions, returned to the car previously discussed: 1 3 "I don't Remember that": Negotiating Memories and Epistemic… --- Example 8 In line 1-4, P produces an indirect question in the form of a statement, which is a reformulation of the information given by the suspect. The suspect has said that s/he has borrowed the car, "now and then" and "on and off" . IE responds to this in line 5, with an initial and cut off "jag ve-" , which is altered through a repair to "jag minns inte det" . In Swedish, the negation is placed after the verb in normal word order, if the subject is produced before the verb. The verb swap is an interesting detail, which opens for questions regarding how speakers in the data, as well as more generally, make interactional distinctions between know and remember [cf. 49]. In this example, this turn is followed by a lengthy pause of 3.0 seconds, after which IE returns to the previous statement and develops this response. IE initially produces a non-negated phrase "yes cause I rememb-" + "maybe", followed by "I really don't remember". In line 8, IE opens for a yes-response to the police's question: "maybe he has", before moving out of the turn with "well I don't remember much of that time". In the turn produced in lines 7-8, IE hesitates and admits to the possibility of the car being used by the suspect, at the same time as IE states that they do not remember. As the speaker moves through the turn, focus shifts from the lending of the car, which is treated as the 'knowable' by the police, to a position where IE states not being able to recall much at all from that time. This way, the 'knowable' is pushed further away, and the distance allows the speaker to gain a K-position regarding the factual events. In line 9, the police officer frames the upcoming disaffiliative turn with a minimal response, a common strategy used by speakers as shown by Stivers [45], before producing a turn that challenges IE:s prior statement as it displays noticeable distrust: "but surely you ought to remember that". This turn projects a high degree of epistemic responsibility back onto IE, treating what is requested as a Type 1-knowable to IE [33]. This is a rich example of when an IDR is not treated as an acceptable response to the question at hand and is explicitly challenged by the police. The 'ought to' does not only project a high degree of responsibility, it also projects obligation [19]. For that reason, the police officer borders blaming IE for stating that they do not recall. The challenge manifested through the IDR results in IE revising a prior response into a different one . This is accomplished by an explicit change of state, described by Heritage [18] as when speakers exhibit a change in stance due to e.g., new information. In this case, the change of state comes as a result of the police officer's challenge, displayed with an initial "aah" in line 11, followed by an alteration of the previous response. The initial "aah yes now that you mention it" creates an opportunity for IE to reverse the prior stance without performing a repair [39] of a prior turn. Rather, the change of state, which results in a yes-response to the initial question , is framed as being triggered by P's preceding turn. 1 3 "I don't Remember that": Negotiating Memories and Epistemic… The affirmative answer is also somewhat vague, as IE admits to the suspect having lent the car once, whereas the question implied that the suspect repeatedly borrowed the car. This way, IE provides an affiliative response, as is projected by the question formulation [36]. After a brief pause in line 14, IE repeats the statement in line 15: "now that you mention it I'm thinking it's true". Again, IE ties the change of state to the information provided by the police, and while doing so, they use thinking it's true, which still contains a degree of uncertainty, as think is a rather neutral activity exposing a low degree of epistemic responsibility, compared to e.g., believe, which is a closely related verb [cf. 19]. P produces minimal responses in line 16, before IE in line 18-19, further mitigates the response: "but I don't remember like when he used it and the like no". In sum, IE initially produces a negative response to the question regarding lending the suspect the car. When this is being challenged by P, IE immediately changes the response, but the information provided is rather vague, as they only admit to lending the car, not narrowing it down to any specific date or time. In this example the police officer's challenge did not lead to a hostile interactional environment [cf. 5], but the data contains such, as will be illustrated in the next example. In the interviews with the suspect, the police recurrently direct attention towards inconsistencies both regarding information provided by the suspect, as well as in relation to evidence collected elsewhere. In the following example, which is collected from the second interview with the suspect, prior to line 1, the suspect claims to have been in contact with a man, and the police officer questions when this meeting occured. --- 3 Example 9 1 3 "I don't Remember that": Negotiating Memories and Epistemic… Just like in Example 8, IE produces a "I don't know" response before stating that s/he does not recall. The turn in lines 3-4 is produced with hesitations, a cut off, a pause and multiple prolonged sounds. P produces a minimal response in proximal position after which IE adds that "it's very hard", referring to the action of recalling. This accomplishes a shift in focus, as IE targets the difficulties with recalling. The police officer, however, continues with "you have also the-" which is cut off, but projects that they are in possession of further details related to the question that IE is not responding to. It is likely, that "the" in line 7 projects a time or date in Swedish. By claiming to have more facts in addition to the minimal non-affiliative response, P makes clear that they have not exhausted the subject. IE interrupts P in line 8-9, stating that s/he "can't afterwards fabricate things". What this turn accomplishes, is that it introduces a new verb and a potentially new activity: fabricate. This is immediately confronted by P with a "no" , which functions as a rejection to IE:s allegation of P trying to elicit a fabricated response, i.e., one that is not truthful. This allegation put forward by IE, also implies that s/he is not a person that would fabricate an answer, i.e., implying that s/he leaves truthful and honest responses to the police -when this is possible. In turn, this constructs the police as someone potentially forcing the interviewee to lie. In line 11, S adds that "you are asking me to do something that is impossible" [cf. 11], connecting to the prior turn expressing that the recalling is difficult. What IE ultimately says, is that fabricating would be the only possible way to provide something that would be treated as sufficient answer to the question, since the endeavour demanded by the police cannot be accomplished. This leads to the police referring to information provided by IE in an interview two weeks prior, something that is confirmed by IE. Then, IE takes one step back and admits to the claims made by the police , but through introducing the concept of details. This echoes the example above with the car keys, which ended with IE admitting to having lent the car, but not having access to details such as date or time. The police officer continues to produce minimal silent responses. That these demonstrate non-affiliation with the suspects' responses becomes evident when the police pursue the topic in line 22. Throughout this sequence, the police show that indicating difficulty in recalling is not regarded as a satisfactory answer to the query. However, the suspect succeeds in shifting the focus from the inquiry about their contact with a particular person to the daunting challenge of recalling-an activity presented as impossible. The final example of this section is extracted from a witness interview, with a witness who may possess knowledge of the suspects' whereabouts and potential activities on the day of the crime. This example has been slightly altered in order to reveal minimal details of the witness' whereabouts. The witness' driving style, referred to in line 1, has been mentioned in a different interview. --- Example 10 In this last example, we initially find the police referring to a different witness' testimony, in which that witness accounts for this witness' driving style, as they have been going in the same car together the day that the crime was committed. The police officer treats this information as evidence for IE having experienced something traumatic that day, related to the crime under investigation. IE claims to have no knowledge of the crime, and in line 5 IE disputes the assertion provided by the police. The police officer reconfirms the information given to them, and states that the description regarding the witness' driving style was the experience of the other witness'. From this turn, IE choses to reaffirm that this was indeed the other person's experience, and this is immediately put in contrast to IE's memory, as IE states not remembering this . This functions as a semi-admittance of the other witness' observation, however, it also challenges the 1 3 "I don't Remember that": Negotiating Memories and Epistemic… grade of epistemic access. Experiencing someone else's driving style may on the one hand provide an A-event, and this experience is something that provides epistemic access to an event. However, the actual driving can only be performed by one person, in this case IE, and by refusing to even remembering this activity, IE disqualifies the other witness. As the potential driver, IE is also an 'A-eventer', and maybe even more so, as being the one whoactually drove, knows why they drove in a specific fashion. This challenge functions as an acceleration in the discreditation of the other witness, as the irrationality referred to was based on the driving style-and IE does not even admit to remembering driving a car. To this, the police officer responds with multiple no:s in line 11, before IE engages the police officer personally, by saying that P would not remember having driven a car that day either, implicating that P would give the same answer if roles were reversed. This functions to illustrate that the request made by the police officer is an impossible one, and it resonates with previous examples in this section, in which speakers frame their incapability to recall as an impossible task. This receives no uptake from P, who moves on to the next topic. This section has examined sequences in which the interviewee responds with "I don't recall/remember" , and the police subsequently challenge this in a third position. Examples have illustrated two distinct trajectories that unfold after the police officer's challenge: The challenge prompts a pause in the interaction, forcing IE to provide confirmations resulting in partial revision of the previous answer The challenge is treated as a confrontation requesting an impossible task: remembering or recalling is claimed to be an impossible achievement In summary, the police aim to elicit information concerning specific segments within a broader chain of events. In doing so, they direct their focus toward distinct 'knowables', treating IE as an individual with not only epistemic access to these elements but also accountability. However, IE dismisses this position, asserting an inability to recall. The police counter this assertion through diverse methods that effectively halts the interaction. This juncture serves as an opportunity for the interviewee to enter a renegotiation of the requested action in the initial question. IE accomplishes a shifts from the primary event-the subject of the question-and reconsiders the nature of the request: is it a reasonable inquiry? By shifting the question towards details from two decades ago, IE effectively constructs the requested knowable as unattainable. Thus, focus pivots to the activity of recalling, rather than engaging in accounts of events in connection to the committed crime. --- Concluding Remarks This study presents findings that explore interactional contexts within police interviews where the interviewee offers the response "I don't recall/remember" to a question, which in turn is challenged by the police in the third position. Results show that recalling/remembering are made central activities throughout, as both the police and the interviewee are oriented towards potential difficulties with recalling/remembering due to the time span of twenty-some years between the crime and the interviews. Consistently, the interviewees are prompted to make efforts in recalling/remembering, while the time frame is explicitly presented as a potential impediment to the requested task. This might establish rapport with the interviewee by mitigating the perceived stakes, while concurrently highlighting the substantial nature of this task. In this process, the focus is subtly shifted from the actual events under investigation to the endeavour of attempting to recall/remember. As shown in the introductory analytical section , interviewees manifest an awareness of the expectations regarding their capability to recall, along with the anticipation that they can provide accounts of the events in question. The analysis also suggests that the interactional work performed by the police officers while downplaying the stakes contributes to projecting a candidate understanding , framing the circumstances as difficult. This contradicts expectations typically ascribed to the interviewee, who is invited with the premise of occupying a knowledgeable position, endowed with access to the knowable aspects-preferably Type 1 knowledge. Stating challenges with remembering/recalling is observed across the dataset, however, interactional environments in which IDRs are challenged by the police , are found in close connection to what has been described as 'hostile environments' by Cerović [5]. In these environments, IE is ascribed with high degree of epistemic access, and the police's challenge is treated as accusatory. In these instances, the question does not merely serve as a neutral inquiry for information; instead, it implies a specific sequence of events. In these contexts, the questions appear designed to elicit confirmations regarding particular events, resulting in IDR responses from the interviewees. Upon being challenged by the police, the interviewees shift focus away from the event, to the challenges they encounter when attempting to recall the specific incident. Interactionally, the challenge is employed as a device to redirect attention away from the potential knowable that the police aim to elicit towards the seemingly challenging task of recalling. This operation aligns with Goodwin's description of forgetfulness as an interactive resource [16], as forgetfulness appears to serve as a vehicle for shifting focus, yet facilitating the interviewees' with a resource for providing responses in the response slot, resonating with Keevallik's argument [24]. This study contributes to the understanding of fine-grained details of interactions within high-stakes police interviews conducted in Swedish. The study illustrates a connection between question formulation and response in terms of orientation towards the time frame, which may act as an obstacle to memory retrieval. The findings also contribute to the ongoing exploration of epistemic dimensions, shedding light on how they are realized in high-stakes interactions. However, the study also generates new inquiries, including an exploration of how claims of not being able to remember or recall are systematically deployed in different sequential settings within police interviews. This is particularly relevant in hostile environments where, for example, presented evidence may be resisted by the ---

This paper employs Conversation Analysis to investigate a specific interactional environment in Swedish police interviews (PIs): sequences where the interviewee asserts an inability to recollect specific events, and the police subsequently challenge this assertion. The police interview serves as a crucial setting for reconstructing past events and identifying the distribution of knowledge among participants. While previous research has delved into the cognitive mechanisms underlying memory retrieval in PIs, there exists a scarcity of empirical investigation of how memories and their associated knowledge are interactionally managed within this high-stakes activity. Prior Conversation Analytic studies exploring how epistemic dimensions shape social interaction form the theoretical basis for the current study, including research indicating how 'forgetfulness' can be strategically employed as an interactional resource. Only a few studies have targeted recorded high-stakes interviews in Swedish, and the lack of such is problematic since international research have highlighted the influence of cultural and social factors on conditions and outcomes. Data used for this study comprises anonymized audio recordings (N = 51) from a preliminary murder investigation, making a valuable contribution to the understanding of interactional practices in Swedish police interviews. Results indicate a discrepancy in orientation between the police and the interviewee. The former treats the memories as accessible knowables possible to retrieve, in contrast to the latter, who uses the police's challenge as a vehicle for contesting the 'impossible action of remembering.'

Introduction Cancer is a major public health problem globally and has become the leading cause of death and illness in China . The International Agency for Research on Cancer estimated that 19.3 million cancer cases were newly diagnosed worldwide and nearly 10.0 million cancer-related deaths occurred in 2020 . With a rapidly aging population worldwide and an increase in unhealthy lifestyles, cancer has been identified as the primary cause of death, reducing the survival time of cancer patients . An estimated 4.6 million new cancer cases and 3.0 million cancer deaths occurred in China in 2020 . China had a slightly lower cancer incidence rate but substantially higher cancer mortality compared to other countries . Developing countries accounted for >56% of the total new annual incidence of cancer patients, with a total cancer-related mortality rate of 64% . Lung and bronchus cancer was commonly diagnosed and identified as the leading cancer killer in China, with ∼781,000 new cases and 626,000 deaths every year, followed by stomach, esophageal, liver and colorectum cancers . In 2015, the mortality attributed to these five types of cancer accounted for about three-quarters of all cancer mortality . In addition, 16.6% of the total disease burden were attributed to cancer in China . Meanwhile, studies have found that cancer patients often bear considerable medical expenditure. The overall incidence rate of catastrophic health expenditure in cancer patient families was estimated at 60.0% . Patients with cancer from socioeconomically disadvantaged households were particularly financially vulnerable due to the high costs of cancer care, which prevented them from accessing health care. Health care use by cancer patients in lower socioeconomic status groups was limited . A systematic review based on cancer inequalities studies has concluded that there were statistically significant socioeconomic inequalities in cancer biological and precision therapy utilization, and a 1.2-fold gap in cancer therapies treatment between cancer patients with the lowest socioeconomic status and the highest socioeconomic status was observed . The rich cancer patients tended to use more health care. In addition, health care costs might be particularly challenging for those without health insurance who were more likely to pay greater out-of-pocket costs . Hence, the disparity in health care use in China remains a major issue to maximizing total health. The inequality in health has been a major priority of the health system globally . Several studies have contributed an extensive amount of research on the many different dimensions of cancer outcome inequality , including reporting gradients in cancer incidence, mortality, and survival were associated with deprivation and lower socioeconomic status. However, socioeconomic inequalities in health care use or behavior among cancer patients remain largely unexplored, although this type of inequality has also been observed in some high-income countries, such as South Korea, Australia, and England . Previous studies have highlighted systematic differences in cancer care use, with higher incidence rates and inadequate use being more prevalent in lower socioeconomic status groups. Moreover, income substantially affected the use of health care . However, existing research has only focused on the association between socioeconomic status and health care use inequalities among cancer patients; to date, the effects of other socioeconomic and need factors remain unclear. Furthermore, no systematic analysis of health care use inequality and influencing factors among cancer patients in China has been published. Hence, this study sought to close these gaps by measuring socioeconomic inequalities in health care use among patients with cancer in China in order to determine which areas will require more attention in the future. --- Methods --- Study design and data sources This study was based on data collected from the China Health and Retirement Longitudinal Study in 2018, which was conducted by the China Center for Economic Research of Peking University. The survey used a questionnaire to collect data, such as demographic characteristics, socioeconomic status, social security level, and physical health status of patients. Using a multistage probability-proportional-to-size sampling, a total of 19,507 individuals aged ≥45 years were identified. Patients who were reported as having cancer and had no missing values for dependent variables were considered eligible for inclusion. After excluding those with missing relevant variables, a total of 392 individuals were finally included in this study. --- Socioeconomic status The annual per capita household expenditure was adopted as a proxy for socioeconomic status and used to group individuals into five groups, from the lowest to the highest. The quintile of socioeconomic status categories was determined within each county or district and then pooled across all sampled counties and districts because the level of economic development differed between sampling regions. --- Variables Dependent variables Two variables of health care use were employed. Patients with cancer were asked if they had visited a public hospital, private hospital, public health center, clinic, or health worker's or doctor's practice or been visited by a health worker or doctor for outpatient care in the last month and had they received inpatient care in the past year. The answers to these questions were coded as a dummy variable . --- Independent and control variables The following variables were included to investigate the relationship of socioeconomic status and health care use: gender , age , educational level , marital status [single , married or partnered], employment status , impoverished status , region , Hukou type , region of residence , health insurance [no health insurance, Urban Employee Basic Medical Insurance , Urban and Rural Resident Basic Medical Insurance , Urban Resident Basic Medical Insurance , New Rural Cooperative Medical Scheme , or another], number of people in the household, physical examination , self-reported health status , disability , degree of pain , smoking , and alcohol consumption . --- Statistical analysis --- Measurement of concentration index The measurement of the CI proposed by Wagstaff et al. was used to examine the magnitude of socioeconomic inequality according to Equation 1. C = 2 µ cov where h i is the measure of actual health service use, µ is its mean and r i is the relative fractional rank of an individual i in the distribution of the annual per capita household expenditure . According to Wagstaff et al. , the CI is defined as twice the area between the concentration curve and the line of equality, where a concentration curve plots the cumulative proportion of the use of services against the cumulative percentage of respondents, ranked by the annual per capita household expenditure, beginning with the least affluent and ending with the most affluent . The CI ranges from -1 to 1. When the concentration curve lies below the diagonal , the CI is a positive value, indicating the concentration of health inequality in favor of the rich . --- Analysis of decomposition method The decomposition method proposed by Wagstaff et al. was employed to measure factors associated with inequalities. They demonstrated that the health CI can be decomposed into the contributions of individual factors to income-related health inequality, in which each contribution is the product of the sensitivity of heath with respect to that factor and the degree of income-related inequality in that factor. A decomposition analysis estimates how determinants proportionally contribute to inequality in the use of services. A positive value of contribution to socioeconomic inequality means a positive association with health care use; in other words, the variable increases pro-rich inequality and outpatient or inpatient services is more concentrated in the richer population. The overall inequality in health services use is written as: C = j C j + k Ck + GC/µ (2 ) where µ is the mean of y, x j is the mean of x j , C j and C k are the CI of need and non-need variables; and GC is the error term of health care. All analyses were performed using SPSS version 25.0 and Stata version 16.0 . A two-sided value of 0.05 was considered to be statistically significant. --- Results --- Social demographic characteristics of cancer patients A total of 392 cancer-related cases were observed, of which 23.47% had visited for outpatient care during the past month and 40.82% had received inpatient services in the last year. Cancer was most prevalent in male and female individuals aged 60-74 years . About 2/3 of patients with cancer reported retirement and unemployment with their cancer diagnosis. Only 16 reported being uninsured. Of note, individuals from socioeconomically disadvantaged households were less likely to seek outpatient and inpatient services compared to better-off individuals. Other descriptive statistics of health care use and cancer patients' characteristics are shown in Table 1. The CIs for actual outpatient and inpatient services use were both positive, and the values of the indices for inpatient services were much higher than those for outpatient visits . With regard to need-expected use, the CI was not statistically significant in both outpatient and inpatient services, and proportionality was not rejected in either case . --- Distribution of health care use among patients with cancer This study also revealed a 1.2-fold gap in outpatient visits use and a 1.3-fold gap in inpatient services use between the lowest income quintile and the highest income quintile after adjustment due to health needs. Indeed, after controlling for the distribution of needs, a significant pro-rich degree of inequality emerged . As shown in Figures 1,2, the concentration curves of actual and standardized outpatient and inpatient service use were all below the line of equality. --- Decomposition of inequality in cancer care use Table 3 depicts the decomposition results and the contributions of various factors influencing the inequalities in cancer care use. Regardless of outpatient and inpatient services use, socioeconomic status made the greatest prorich contributions-that is, 78.99% and 83.92%, respectively, -followed by high school education and living in a rural village . UEBMI made a great contribution to the pro-rich inequality in inpatient services while having a negative impact on outpatient visits . NRCMS had the opposite effect, but its contribution was relatively small. Among the need variables, a "health-poor" status and smoking had a positive contribution to the pro-rich inequality, while a "health-fair" status reduced the pro-rich inequality . The other variables provided relatively minor contributions to the inequity, as shown in Table 3. --- Discussion Variations in the use of health care among cancer patients have attracted increased attention from both researchers and policymakers in related areas. To our knowledge, this is the first study to examine the association between socioeconomic status and health care use across different socioeconomic populations in China. Our study analyzed the distribution of the use of outpatient visits and inpatient services among patients with cancer from a perspective of equity. The analysis carried out here highlighted that cancer patients from higher socioeconomic status groups were more likely to use health care than those who were worse off. It was also evident that, after controlling for age, gender, and other need variables, there was a clear socioeconomic gradient in health care use. In addition, socioeconomic status and health insurance interacted to influence the risk of inequality in decomposition models. In our study, the CIs for outpatient and inpatient services use were all positive, indicating that there was statistically significant inequality in the use of health care among cancer patients, in line with previous studies from South Korea and Australia . Richer cancer patients appeared to be much more likely to use health care. In addition, this study revealed a greater extent of inequality compared to other research. A possible explanation may be that our study included individuals aged ≥45 years, and most incidence and deaths of cancer occurred in this age range . The health condition of these cancer patients might deteriorate due to inadequate sources of income , with the financial burden of age-related health rising . Compared to the entire population with cancer, the distribution of health care utilization among middle-aged and elderly cancer patients was more unequal. Our study showed that higher inequality was generally in inpatient services in China. It could be explained by the fact that hospitalization costs were very high. Medical expenses and non-medical costs in inpatient services were higher than those in outpatient visits, which exacerbated the burden on health care use . Hence, cancer patients from socioeconomically disadvantaged households could not afford the high medical costs; actually, they tended to abandon medical services or sought cheaper outpatient services instead of inpatient services . Meanwhile, a lower socioeconomic status was related to a shorter survival time in cancer patients . Cancer patients with a higher socioeconomic status survived long enough to use additional inpatient services. Given this, inequalities in the utilization of inpatient services among cancer patients warrant more attention than disparities in outpatient visits. We found that inequalities in health care among cancer patients remain largely determined by patients' financial Results of an Italy survey of individuals aged >50 years also indicated that income was a positive and significant determinant of use in preventive cancer care use . One possible explanation for this may be that, different from other diseases, cancer has more frequent recurrence, shorter diseasefree survival, and higher mortality rates , placing a substantial economic burden on cancer sufferers and their families. Poor households were most likely to face impoverishment and economic hardship, entering a vicious circle of "poverty from illness and disease from poverty" . Health care allocation and use are disproportionately favored by the betteroff with higher education levels and, therefore, may widen inequalities further. It is well-known that health insurance schemes are associated with health care use. Previous studies have shown that insured individuals were more likely to use health care than uninsured ones . An incidence-based study that examined socioeconomic inequalities in Australia found that, apart from providing free medical services in public hospitals, Medicare had policies to protect patients from catastrophic health expenditures , defined as health-related out-of-pocket costs of ≥40% of total non-food household consumption expenditures . In our study, we observed UEBMI's prorich contributions to inpatient service use as well as the limited effects of URBMI and NRCMS, indicating that these health insurance schemes failed to protect low-income cancer patients, especially in terms of inpatient services . This result can be explained by certain reasons. First, although >96% of patients with cancer were covered by health insurance, UEBMI, URBMI, and NRCMS did not reimburse all medical services and items, especially expensive targeted therapies. Second, about 55.36% of participants with cancer in our study were unemployed, bearing the heavy burden of cancer therapy. In addition, these findings may be attributed to differences in the benefit packages between the different health insurance plans . UEBMI had a greater reimbursement rate than other health insurance schemes. The UEBMI beneficiaries were more willing to use expensive drugs and medical compared to the URBMI and NRCMS cancer patients . Evidence from an communitybased study in China has confirmed that, in order to lessen the compensation gap between different insurances, the expansion of benefits packages should be tailored to differences between cancer patients in terms of income, health needs, and other factors . UEBMI had different implications in outpatient and inpatient services on inequality. It could be explained that the cancer treatment choices varied in the different socioeconomic statuses. Due to the high cost of inpatient services, cancer patients with lower socioeconomic status were more willing to use outpatient services to alleviate, while surgical treatment was often chosen among the rich cancer patients . Therefore, for outpatient utilization, the disparities were relatively small. In addition, from the patients' socioeconomic status perspective, cancer patients who were covered by UEBMI were all urban workers or retired workers, they usually had higher income and better education compared to those with URBMI and NRCMS . Hence, they had a stronger incentive to utilize health care, which led to the significant effect on inequalities of UEBMI. UEBMI played a role in protecting the lower-income cancer patients from catastrophic health expenditure and had reduced financial burden in outpatient utilization, while cancer patients with higher socioeconomic status used more inpatient services, increasing the inequalities in inpatient utilization. We did not find an apparent influence in health care use inequalities by age, although greater use by elderly individuals was observed. A possible reason for this result could be found in the sample characteristics, as only 13.52% of participants were aged ≥75 years. However, poverty, limited insurance coverage, education and awareness were factors that contributed to inequalities in cancer patients' health care use, in line with previous reports . Wealth, the health insurance benefits package, and high school education increased the use of health care among cancer patients. Higher-income individuals had greater access to education, healthy dietary habits, and cancer care. This was also a good explanation for the prorich contribution of socioeconomic status to health care use among patients with cancer. Hence, a sustained reduction in socioeconomic inequalities concerning poverty would promote universal equality in health. In addition, more equitable and effective benefits packages committed to provide financial protection against catastrophic illness, such as expanding the public health insurance coverage of inpatient care to cancer patients, should also be designed. Our study has some limitations. First, the diagnosis of cancer was self-reported, which might have led to underor overestimation of the cancer prevalence. The information about health care use was also self-reported, so recall biases could not be avoided. In further research, more data sources and methods should be adopted to control these biases. Second, this study performed a cross-sectional analysis, which prevented us from discussing results based on causal inference. Third, the study sample might be not representative. Our sample size was relatively small and only included individuals aged ≥45 years. Fourth, since URBMI and NCMS have been merged, a comparison between UEBMI and URRBMI could be a better choice in future research . Finally, quality or efficiency measures should be included in inequality research; unfortunately, our survey did not provide relevant indicators . --- Conclusion Significant differences were seen in the distribution of cancer care use across socioeconomic status groups in China, and a socioeconomic gradient was evident. Socioeconomic status and Frontiers in Public Health frontiersin.org health insurance were found to be associated with inequalities. Interventions aimed at reducing inequalities in health care use should focus on improving financial protections for people from socioeconomically disadvantaged households. --- Data availability statement The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s. --- --- --- Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Background: Cancer is a major public health problem worldwide and the leading cause of death in China, with increasing incidence and mortality rates. This study sought to assess socioeconomic-related inequalities in health care use among cancer patients in China and to analyze factors associated with this disparity. Methods: This study used data collected for the China Health and Retirement Longitudinal Study in . Patients who reported having cancer were included. The annual per capita household expenditure was classified into five groups by the quintile method. We calculated the distribution of actual, need-predicted, and need-standardized health care use across di erent socioeconomic groups among patients with cancer. The concentration index (CI) was used to evaluate inequalities in health care use. Influencing factors of inequalities were measured with the decomposition method. Results: A total of people diagnosed with cancer were included in this study. The proportion of cancer patients who utilized outpatient and inpatient services was . % and . %, respectively, and the CIs for actual outpatient and inpatient service use were . and . . The standardized CIs (CI for outpatient visits = . ; CI for inpatient services = . ) were also both positive, indicating that a uent cancer patients used more health services. The annual per capita household expenditure was the greatest factor favoring the better-o , which contributed as much as . % and . % to the inequality in outpatient and inpatient services use, followed by high school education ( . % for outpatient services) and living in a rural village ( . % for inpatient services). Urban Employee Basic Medical Insurance exacerbated the inequality in inpatient services ( . %) while having a negative impact on outpatient visits (-. %). Conclusions: There is a pro-rich inequality in outpatient and inpatient services use among cancer patients in China. A lower socioeconomic status is negatively associated with cancer care use. Hence, more targeted financial protection for poor people would relieve cancer patients of the burden caused by the high cost of cancer care.

Introduction Nigeria's population, estimated at about 200 million as of 2019, is projected to increase to about 440 million by the year 2050 . With a current rate of over 5 children per woman, Nigeria's total fertility rate remains higher than most other sub-Saharan African countries, including those in West Africa . Fertility levels are particularly high in some Nigerian subpopulations, including among residents of the North West and North East regions of the country; Hausa, Fulani, and Kanuri ethnic groups; and those who identify as Muslim or practice traditional religions . Despite high knowledge of family planning in Nigeria--93% of women of reproductive age know of at least one contraceptive method-contraceptive use remains low [Nigeria] and ICF International, 2019). Among married Nigerian women modern contraceptive use increased from 10% in 2013 to only 12% in 2018 [Nigeria] and ICF International, 2014, 2019). There is wide geographical variation in modern contraceptive use in Nigeria, with estimates among married women ranging from 3.9% in the North-West region to 20.0% and 25.4% in the North-Central and South-West regions, respectively [Nigeria] and ICF International, 2014). A plethora of studies have examined individual-and interpersonal level predictors of contraceptive use in Nigeria. These include wealth, education, ethnicity, women's decision-making abilities, and spousal support and communication . A smaller body of research has examined health facilityand community-level factors, including method stock out , health care worker home visits , social norms , and community-level knowledge and literacy , that influence use of contraception. While FP investments in Nigeria have been ongoing for decades, relatively little emphasis on contextual and structural factors may have contributed to low demand for and use of family planning . Greater focus on structural influences, including social norms, that increase acceptance and use of contraceptive methods is needed. Social norms are informal rules that govern behavior in a particular context . Scholars from across the social sciences have attempted to measure and explain the effect of social norms on people's choices and behaviors . While collective norms exist at the social level-typically, the group, community, or national level-individuals' interpretation of these norms, known as perceived norms, exist at the individual, psychological level. Perceived norms are the result of individuals' cognitive processes; thus, understanding the role that social and behavior change communication programs have on influencing norms is operationalized through measurement of perceived norms . While some health-related studies have attempted to measure the influence of collective norms on health choices and behaviors , research on family planning and sexual and reproductive health has increasingly focused on the ways in which perceived norms influence behaviors . Perceived norms are further delineated into injunctive norms-beliefs about what other people approve of or think one should do-and descriptive norms-beliefs about what other people do . The influence of injunctive and descriptive norms on family planning and reproductive health behaviors is mixed. For example, one study evaluating the influence of a male engagement and social norms intervention in the Democratic Republic of Congo found that injunctive norms among married women and descriptive FP norms among married men were associated with future intention to use FP. However, there was no association between descriptive FP norms among women, or injunctive FP norms among men, and future intentions to use FP . Another study in Ethiopia found that, among 15-24 year old male youth, the descriptive norm of knowing a friend who had ever used condoms was associated with use of condoms at last sex, and the injunctive norm of being worried about other people would think if the respondent needed condoms significantly decreased their likelihood of using condoms . Another study implemented in Kenya and Ethiopia found that injunctive norms alone were not associated with women's contraceptive use; rather, women whose current number of sons was lower than their perception of the community ideal had a lower odds of reporting contraceptive use, while women whose personal ideal number of sons was lower than their perceived community ideals had a greater odds of reported contraceptive use . The Nigerian Urban Reproductive Health Initiative , funded by the Bill and Melinda Gates Foundation and implemented by the Johns Hopkins Center for Communication Programs, aimed to increase voluntary modern contraceptives use among women ages 15-49 years through comprehensive demand and supply side programming. NURHI programming was guided by the theory of ideation, which proposes that people's actions are influenced strongly by their beliefs, ideas, and feelings. Ideation factors include personal attitudes and beliefs , and social norms . NUHRI designed and integrated communication methodologies, including those used in mass media campaigns and social mobilization efforts, into each component of the program, including the service delivery ones; and placed intensive and sustained effort and resources into demand generation activities. NUHRI defined demand for family planning as the desire and ability among women and/or men to take action to plan their facilities. The program hypothesized that the demand generation elements would work together to influence ideation factors, including social norms; and these, along with the supply programming, would in turn increase use of modern contraception . Phase 1 of NURHI was implemented from 2009 to 2014 in six Nigerian cities: Abuja, Ibadan, Ilorin, Kaduna, Benin and Zaira. NURHI's demand generation activities consisted of communication campaigns to promote discussion of FP, reduce social barriers, myths, and social stigma, and increase approval of FP methods. Vehicles for the communication campaign included mass media , entertainment-education , and social mobilization to enhance interpersonal communication during client-provider interactions, between spouses, during trade group meetings, and through neighborhood campaigns and social events . Supply side programming included provider training, ensuring security of commodity supplies, and improving the overall clinic environment in target facilities . A subsequent phase of NURHI, Phase 2, was implemented in three states: Oyo, from 2015 to 2018; and Lagos and Kaduna from 2015 to 2020. Phase 2 focused on specific priority audiences of women with unmet need, traditional method users, men and service providers ; and integrated the use of digital and social media into the demand generation strategy . For further details on the NURHI intervention, please see Krenn et al. and Adedini et al. . This study examines whether social norms around FP are related to women's use of modern contraception. Focusing on residents in the cities of Ilorin, Kaduna and Jos, this study also examines whether the relationship between social norms and use of modern contraception differs by varying levels of exposure to NURHI . --- Materials and methods --- Study design and sample The NURHI Sustainability Study examined the continued impacts of the NURHI program on FP attitudes and behaviors 2 years after the end of Phase 1. We used data from the 2017 cross-sectional survey that was part of a 2015 parent study. Three cities with varying levels of the program were included in the current analysis. NURHI Phase 1 only was implemented in Ilorin; NURHI Phases 1 and 2 were implemented in Kaduna; and in Jos, no NURHI program had been implemented. In 2015, a cross sectional survey was undertaken in Ilorin and Kaduna. For the 2015 survey, a two-stage sampling design was used to obtain a representative sample of respondents in each city. First, enumeration areas from the 2006 Nigeria census frame were grouped into primary sampling units ; a random selection of PSUs was then taken. Next, a household listing and mapping was undertaken and then 41 households were randomly selected from each PSU. Following informed consent, all women of reproductive age who had spent the previous night in the household were eligible for participation. The 2015 data were not utilized for the current analysis, but all of the PSUs in Ilorin and Kaduna in the 2015 survey were included in the 2017 survey. In 2017, a second cross-sectional survey was undertaken which included Ilorin, Kaduna and Jos. In Ilorin and Kaduna, we undertook a census of all households located in the sampled PSUs from the 2015 survey to permit matching women to the 2015 sample. All women ages 15-49 who had spent the previous night in the household were eligible to participate in the study after providing informed consent. In Jos, which was not included in the 2015 survey, a two-stage sampling design was used. A total of 56 PSUs were selected from the 2006 Nigeria Census sampling frame. A listing and mapping exercise was undertaken and a random sample of 33 households was selected in each Jos cluster. All women ages 15-49 years residing in or visiting the selected households the night before the survey were eligible to be interviewed after providing informed consent. The initial sample size was 10,535. The analysis in this paper includes only women who reported ever having sexual intercourse. After dropping respondents who reported never having sexual intercourse, the final sample size for this study was 6,396 . --- Study measures Outcome measure: Use of a modern contraceptive method The 2017 NURHI Sustainability Study survey included a question on which contraceptive methods the respondent or her partner was using at the time of the survey. Modern method use consists of use of at least one the following: female sterilization, implant, intrauterine device , injectables, daily pill, emergency pill, male condom, female condom, lactational amenorrhea method , or standard days method . Women were coded one if they currently used any of these modern methods and zero otherwise. --- Social norms The survey included a series of quantitative questions based on vignettes to elicit social norms related to FP use for delaying first pregnancy, spacing pregnancies, and limiting pregnancy. Vignettes are mini-scenarios that ask respondents about their perspectives and attitudes toward one or more fictional characters, and are increasingly used to measure social norms . The study vignettes included an adolescent girl who was sexually active with her 17-year-old boyfriend and was considering using modern contraception; a 21-year-old mother of a 6-month-old baby who wanted to space her next pregnancy while her husband wanted another child immediately; and a 28-year-old woman with four children who wanted to prevent future pregnancies. All social norms survey items were asked on a five-point Likert response scale . --- Individual-level factors Respondent characteristics included age , marital status , highest education level , parity , religion , religiosity , and wealth based on quintile levels. These factors were included given the extant evidence on their relationship to use of modern contraception . --- Data analysis All analyses used weighted data and accounted for the clustered design of the 2017 NURHI Sustainability Study. First, we explored the demographic characteristics of the sample, calculating frequency distributions for the whole sample and by location. Next, we conducted exploratory factor analysis with oblique rotations to determine the latent constructs of social norms. We used scree tests and eigenvalues to determine the number of social norm factors to retain. We dropped items that produced factor loadings below 0.45 or uniquenesses above 0.75, and that did not conceptually fit the factor model. To examine the strength of each FP social norm, we calculated mean scores for each factor by multiplying the items within the factor by their factor loading. We then summed across the products and divided by the number of items within the factor. Last, we fit generalized structural equation models to assess whether FP social norms were associated with respondents' use of modern contraceptives, and whether the relationship differed by city. We fit several models using data from the full sample and used Akaike Information Criterion to determine whether one model fit the data better than others. Subsequently, we used a series of tests of invariance to conduct group comparisons of the selected model by city. --- Ethics The study protocol and all consent procedures and consent forms were approved by the Institutional Review Board at the University of North Carolina at Chapel Hill and by the National Health Research Ethics Committee of Nigeria in Nigeria. --- Results Across all cities, sexually experienced respondents were, on average, 32 years old, and about 84% had been ever married or lived with a man as married. Senior secondary school was the highest level of education for about one-third of women. About 20% of women had no children, almost half had between one child and three children, and the remaining had four to six children. The sample of women who ever had sex from Jos is slightly younger, less likely to be ever married or living with their partner, and have fewer children than their counterparts from Kaduna or Ilorin. In Ilorin and Kaduna almost two-thirds of respondents were Muslim , while in Jos almost two-thirds were Christian . More than threequarters of respondents reported being strongly religious . About 30% of women had used a modern method of contraception at the time of the survey . Results from weighted exploratory factor analysis revealed three social norm factors, or constructs, around FP. The first factor included three items about FP social norms to delay first pregnancy among adolescents . The second included four items about FP social norms to space pregnancies or limit all pregnancies among married women . The third included three items about FP social norms around women's use of contraception when their husbands disagree . Three survey items about FP social norms that did not conceptually or statistically fit were discarded from the final factor model . Mean scores for each factor showed that, overall, FP social norms around spacing and limiting pregnancies was strongest, or most positive, while norms around delaying first pregnancy was weakest, or least positive. Mean scores differed slightly by city, with FP social norms for delaying first pregnancy highest in Kaduna and lowest in Jos; for spacing and limiting pregnancies highest in Ilorin and lowest in Jos; and for using contraception when the husband disagreed highest in Ilorin and lowest in Kaduna . The final structural equation model for the full sample indicated that all three constructs of FP social norms were significantly associated with use of a modern contraceptive method . For every one-unit increase in FP social norms around delaying an adolescent girl's first pregnancy, there was an increase of 0.11 path coefficients in the respondent using a modern method at the time of the pregnancies ; and increased by about 0.08 coefficients for every one unit increase in norms around using contraception even when the husband disagrees . Except for household wealth, use of a modern contraceptive was also significantly associated with all covariates . The models differed by location in the FP social norm constructs that were associated with respondents' use of modern contraception . In Kaduna only, the social norm around adolescents delaying their first pregnancy was associated with respondents' use of a modern contraceptive. In Ilorin only, the norm around using contraceptives when a husband disagrees was associated with use of a modern method. In all three sites, the norm around spacing and limiting pregnancies was associated with respondents' use of modern contraception. --- Discussion and conclusion After adjusting for demographic characteristics and household wealth among all women who ever had sex in three cities in Nigeria, we found that social norms around using FP to delay first pregnancy, to space and limit pregnancies, and when the husband disagreed with contraceptive use were associated with women's use of modern contraception. However, when we examined the data by level of exposure to the NURHI program we found substantial differences. Social norms around using FP to space or limit pregnancies was associated with modern contraceptive use in all three cities. In contrast, FP social norms to delay first pregnancy among adolescents was associated with modern contraceptive use in Kaduna only; and FP social norms for using contraception when a husband disagrees was associated with modern contraceptive use in Ilorin only. The observed differences suggest that the relationship between specific FP social norms and women's use of modern contraception depends on numerous factors at the individual, interpersonal and societal levels , and this may include malleable factors influenced by the NURHI program. For example, increasing access to FP among youth, including addressing provider bias by using human-centered design approaches was a deliberate focus of Phase 2 and in Kaduna only . This component of the program may have supported positive FP norms specifically around adolescents' use of contraception, which in turn may have influenced contraceptive use among a broader cross-section of women in Kaduna. This spillover effect is similar to findings from another study in Nigeria using Performance Monitoring and Accountability 2020 data where authors observed a significant relationship between a facility's delivery of adolescent reproductive healthcare and modern contraceptive use by sexually active women of all reproductive ages . Communities are often less accepting of adolescents' use of FP than that of adult women's , and this is supported by our study as well, given that in all three cities the mean scores for social norms supporting adolescents' use of FP was lower than those of the other measured FP social norms. Positive social norms around adolescents using FP to delay a first pregnancy indicates that individuals perceive members of their community to be open to and/or lenient about FP use in general. That is, this specific norm is likely the most "liberal" of perceived FP norms, indicating that women likely believe that their communities would accept most women's use of FP. However, this FP social norm may influence women's actual use of modern contraception only in certain circumstances, such as when supply side factors, such as access to a range of modern methods and quality of services, are improved. While supply side factors were addressed in Ilorin during NURHI Phase 1, they were strengthened in Phase 2 in Kaduna only. A longitudinal study of quality of services in NURHI sites found that in 2017 Kaduna had better quality services and significantly more new contraceptive users compared to Ilorin . Improved access and higher quality services may be necessary factors to accompany perceived norms that are favorable to FP in order to influence contraceptive behaviors. The observed relationship in Ilorin between FP social norms around using contraception when the husband disagrees and use of modern contraception is more challenging to explain. This finding may be a result of socio-cultural differences between the cities. For example, while both Ilorin and Kaduna are a majority Muslim, there are different compositions of ethnic groups between these two States of Nigeria, Kwara and Kaduna States. The majority of residents in Kwara State are Yoruba people while in Kaduna State they are Hausa and Funali. One multi-level analysis of spatial distribution and factors associated with modern contraceptive use among women in Nigeria found that Yoruba women were more likely than Hausa women to use modern contraception . Given that our study did not control for ethnicity, this may be a salient factor that influenced our results. Varying gender norms among different ethnic groups may influence whether specific FP social norms are associated with use of modern contraception. Gender norms are informal rules and shared social expectations that distinguish expected behavior based on gender and keep the gender system intact . Gender norms is one element of the gender system, a social system that apportions resources, roles, power and entitlement according to whether a person or practice is perceived as male or female, masculine or feminine . Gender norms are embedded within the institutions and narrative of a given culture, produced and reproduced through individuals' actions, and enforced by those who hold power and benefit from compliance to those norms. As such, gender norms have been predominantly conceptualized as a social construct . Cislaghi and Heise suggests, however, that gender norms are at the intersection of the social and individual because the role they play in shaping women's and men's access to resources affects their voice and sense of self and power. Social norms around FP-particularly beliefs about the contraceptive behaviors women in a community would or should practice despite disagreement or opposition from their husbands-incorporate gender norms. Perceptions of what community members support and are doing with regard to a gendered behavior depend on societal and cultural expectations of women's and men's roles, how community members occupy those roles, and whether perceptions of community members' occupation of those roles are accurate. Gendered expectations to prove fecundity may dictate that women refrain from using modern contraception. Gendered expectations to replicate male-dominated power dynamics may dictate that that women use contraception only if their husbands agree. Because social and gender norms and their association with behaviors are contextually specific, it may be possible that a family planning program such as NURHI disrupts unequal gender norms in one city but not in another city , despite greater intensity of the program. Some of our findings diverge from previous research results on social norms and contraceptive use. Most salient, our social norm factors for delaying first pregnancy, and spacing and limiting pregnancies include both injunctive and descriptive items. While our original analysis plan was to separate factors by descriptive norms and injunctive norms, doing so did not produce high factor loadings. Though distinguishing between injunctive and descriptive norms has become a common approach to social norms and behavioral research, the practice is not uniformly supported by all behavior change theorists . For example, revisions of the theory of reasoned action do not make a distinction between these types of norms . Relatedly, the fact that retaining injunctive and descriptive items within a single factor produced better statistical results may suggest that some respondents do not conceptually differentiate descriptive norms from injunctive norms. This study has several limitations. First, this is a crosssectional study and temporality cannot be established. It is possible that women who use contraception are subsequently more likely to perceive positive norms related to family planning within their communities. Additionally, when asked what most people in their community would do or whether most people in a reference group would approve of a particular behavior , some respondents may revert to sharing their own opinions and perspectives . This means that the survey questions in this study may have captured gender norms [operationalized as attitudes toward gender roles and dynamics in relation to family planning ] in addition to, or instead of, perceived social norms . This study also has several strengths. Compared to running separate regression models, using structural equation modeling is a more comprehensive method to analyzing the relationship among latent variables. Structural equation modeling explicitly assesses measurement errors and estimates latent variables via observed variables. Additionally, fully developed models are tested against the data using SEM as a conceptual structure, meaning the relationship among the latent constructs and the observed variables must be pre-specified. The conceptual structure is then evaluated for fit against the sample data . As part of its FP2030 commitment, the Government of Nigeria aims to increase the national contraceptive prevalence rate from 12% to at least 27% through scaling up evidence-based, high impact practices . Both of these efforts can be supported through continued communications and mobilization interventions that focus on shifting social norms around family planning. As seen in the study results, whether and how social norms shift depend on both the type of norm and the context and population within which the norm is being addressed. In order to design interventions that are adequately nuanced and have the best chance of effectiveness, program designers must differentiate and adapt interventions to specific communities. For example, while promoting norms to use family planning to space pregnancies may be an effective strategy across most communities, promoting norms for family planning use among adolescents may be less palatable and therefore less effective in certain communities. Program designers can use various analytic tools, such as the Social Norms Exploration Tool , to explore social norms for specific populations and their reference groups, and use the results to help better design norms interventions that are specific for their communities. --- Conclusion Further research is needed on how collective social and gender norms influence perceived social and gender norms and how both are related to behavior. Challenges remain in measuring collective norms, which exist at a social level, with validity. Aggregating data collected at the individual level is likely to be misleading . Additionally, we need a more granular understanding of the circumstances under which positive FP social norms lead to improved reproductive health behaviors. Whether and how norms lead to a given behavior depend on attributes of the behavior, such as how private or detectable the behavior is and how independently of other people the behavior can be carried out ; characteristics of the individual, such as self-efficacy and self-monitoring ; and interpersonal-and societal-level moderators, such as group identify and group proximity . The relationship between norms and behavior are further influenced by the strength of sanctions of not following the normative behavior and whether the norm is proximately or distally related to the behavior . These characteristics must be considered when designing future social and behavioral programming around family planning and reproductive health. --- Data availability statement The original contributions presented in the study are included in the article/supplementary materials, further inquiries can be directed to the corresponding author/s. --- Ethics statement The studies involving human participants were reviewed and approved by Institutional Review Boards at the University of North Carolina at Chapel Hill in the United States and the National Research Ethics Committee of Nigeria . Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements. --- --- --- Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. --- Author disclaimer The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of CPC or the Bill & Melinda Gates Foundation.

Despite high knowledge of family planning (FP) among Nigerian women, use of modern contraceptives remains low. While FP investments in Nigeria have been ongoing for decades, relatively little emphasis on contextual and structural factors may have contributed to low demand for and use of contraception. From to , the Bill & Melinda Gates Foundation (BMGF) supported the Nigerian Urban Reproductive Health Initiative (NURHI) with the aim of increasing voluntary use of contraceptives among women agesyears in six Nigerian cities. A subsequent phase of NURHI was implemented in three states for the next to years. Using cross-sectional survey data from three cities (two exposed to NURHI, one not exposed), this study examines whether social norms around FP were related to women's use of modern contraception, and whether the relationship di ered by varying levels of exposure to the program (i.e., by city). We identified three distinct FP social norms through factor analysis: norms around delaying first pregnancy; spacing or limiting pregnancies; and using contraception when the husband disagrees. Using structural equation modeling, we found that FP social norms are related to use of modern contraceptive methods, and the relationship varies by city and norm type. The observed di erences suggest that this relationship depends on numerous factors at the individual, interpersonal and societal level, and this may include malleable factors influenced by the NURHI program.

INTRODUCTION In Nigeria, there exists a broad assemblage of government-owned security institutions, of which the Nigerian Police Force is the most prominent. The NPF is a centrally-administered federal security institution with a uniform national structure established to be a vital part of Nigeria's political order. The nature of this political order essentially lies in the enforcement of "specific rights policies, either protecting rights where violations are threatened" , or averting acts of malefaction where necessary. Largely, the NPF has the constitutional mandate of crime prevention and eradication; apprehension and prosecution of criminals; protection of lives and property through proactive policing in Nigeria. The act of policing in democratic Nigeria by the NPF is complex. As an agency of the Nigerian criminal justice system, the NPF engages in a wide array of behavioural problems, each perplexing in its own way. In dealing with the prevailing crime dynamism, protection of human rights, terrorism, kidnapping, maintenance of order, conflict resolution, and other special duties, the NPF is considered to be task-filled. This policing complexity is further problematised by "diversities and contradictions arising from population heterogeneity, urbanization, industrialization, conflicting ideologies on the appropriate socio-political and economic form of organization" , and technological advancement, among others. Given this complex nature of NPF's duties and operational challenges, a high degree of integrity, well-designed arrangements and organisation in discharging their duties are expected. However, these have been defectively upheld over the years. As a matter of course, there have been efforts to reform the NPF's institutional operations, accountability and oversight frameworks such as making the institution more democratic, improving public confidence and mitigating corruption. Yet, all these reformative measures are still not well grounded in the NPF, especially in the area of corruption that is still a major murky water in which the NPF wallows. Ivković describes police corruption as "a form of police misconduct…defined as a violation of penal codes, administrative agency rules, or the codes of ethics." Kempe illustrates this further when he states: …police corruption is any action or omission, a promise of any action or omission, or any attempt of action or omission committed by a police officer or a group of police officers, characterised by the police officer's misuse of the official position and motivated in significant part with the achievement of personal/private or organisational gain or advantage. Going by these archetype conceptualisations, the NPF, characteristically, has reputation for institutional corruption which has undermined its institutional recognition as a law enforcement agency. Though the Nigeria Police Act, signed into law in September 2020, makes provision for a well organised police force driven by the principles of transparency, accountability in operations, protection of human rights and management of resources among others, the act has not engendered any notable transformation among many members of the NPF. As much as there exists what can be relatively considered as a new Police Act , the retention and sustenance of the same culture of corruption and impunity among police officers underlies the fact that as far as policing in Nigeria is concerned, it is still business as usual. Nwite Illustrates the extent of decadence and rot in the NPF with the involvement of a decorated, senior police officer in drug trafficking. According to him: The arrest of Nigeria's hero "super cop" Abba Kyari over his involvement in drug trafficking has once again; put the Nigerian police Force on the spotlight. Kyari who was under investigation for his involvement in a $1.1 million internet scam was suspended by the Police in July. His ability to still run a team that intercepted a drug deal while under suspension has put a question mark on anything integrity left with the Nigerian Police. To many, it is not surprising though. The NPF has overtime, by its actions and inactions, refused to put on the image that will earn it public trust. There are more stories of police involvement in criminal activities than there are of actual criminals in Nigeria -the reason the EndSARS protest happened. In addition to the fold of the NPF's unlawful activities is the aura of corruption that pervades security checkpoint -checkpoint extortion. Security checkpoints, also known as roadblocks in the Nigerian context, which are strategically mounted to forestall crime, have become brazen extortion points, ubiquitous sites of police impunity and abuse of power. In essence, many men and women of the NPF stationed at these checkpoints are known to be more concerned with the naira notes they illegally collect from motorists than diligent policing and prevention of crimes. An editorial opinion of The Guardian validates this illegal strand of NPF excesses thus: "the menace of extortionate security men at check points is nationwide. Successive IGPs have had to order their men off the road because they do not do what ought to by law…they tend to go beyond their briefs." Nevertheless, with this effort, the level of compliance to the discontinuation of extortion-points has not been so convincing. Further efforts were averred in April 2021 by Usman Alkali Baba, the IGP to present a police force guided by the principles of public accountability, conformity with the rule of law, and as compassionate servants and helpers of citizens. However, the conducts of police officers at checkpoints still negate these good intentions of the IGP with regard to present day policing. At different capacities, attempts have been made to instil sanity of ethical policing that underscores norms and style associated with democratic-style policing into the NPF. For instance, successive administrations in the current democratic dispensation, had initiated reforms to revitalise the NPF. In a similar capacity, "Nigeria has a strong record of independent organisations promoting reform. […] many Nigerian academics, analysts, politicians, advocacy groups have recommended ways in which the NP can be reformed" . In a significant way, media, in this case, news, social, web and print among others, have been utilised to draw the attention of the public to the alarming corruption, notoriety and the exigency of reforms in the NPF. The social media, essentially, amongst all, is an archetype digital sphere where NPF and contaminative policing are predominantly portrayed by virtue of unrestricted netizenship. The representations of the NPF, though satirical, capture a critical ambience of the need for reformation and reorientation of the police. This reformative satirical act has been effectively achieved on the social media through digital memes. In describing meme and its characteristics, Shifman affirms that "like many web 2.0 applications, memes diffuse from person to person, but shape and reflect general social mindsets". The NPF digital memes are products of shared societal experiences with the rank and file of the police. These memes are the creations of sociocultural overt reproduction, constructed through edifying artefacts and made prominent in contemporary digital culture. The concept of meme and its semantics which originally is the creation of Richard Dawkins, an Oxford biologist in his book, The Selfish Gene, 1976 and 1989 respectively, has since been reappropriated by the internet and morphed into what Shifman describes as "modern folklore, in which shared norms and values are constructed through cultural artefacts such as photoshopped images or urban legends". Dawkins, having characterised meme as basic unit of cultural transmission that is analogous to genes, portrayed meme as an idea, behaviour or style that spreads from person to person by copying or imitation. Dawkins illustrates meme technically to include tunes, ideas, catchphrases, clothes fashions, and ways of making pots or of building arches. According to Dawkins, "just as genes propagate themselves in the gene pool by leaping from body to body via sperms or eggs, so memes propagate in the meme pool by leaping from brain to brain via a process which, in the broad sense, can be called imitation" .While memes today can certainly denote any of the above, they do not only spread from brain to brain, but now, they leap virally from one internet protocol address to another internet protocol address, via a process which in the broad sense, can be called imitation. Further than what Dawkins considered as meme driven by human representative and re-representative behaviours, Shifman guided by technological transformations in the light of digital technologies, yokes meme with internet, and affirms that "internet meme is commonly applied to describe the propagation of items such as jokes, rumors, videos, and websites from person to person via the internet". Meme is created or based in a real life event that spreads through the internet causing people to replicate it. Sometimes, it spreads in the same way and sometimes it spreads by means of change . Digital technology has provided the digital space that has become a fertile and unfettered ground for the inundation of memes with communicative implications. An investigation of internet meme in the digital space thus opens up the cyber ways of representing socio-political narratives together with their implications and functions. Memes have characteristics that guarantee virality. According to Dawkins , memes possess three characteristics: longevity, fecundity and copying-fidelity. In the words of Shifman : …All three are enhanced by the internet. Online meme transmission has higher copyfidelity than communication through other media, since digitization allows lossless information transfer. Fecundity is also greatly increased -the internet facilitates the swift diffusion of any given message to numerous nodes. Longevity may potentially increase, as well, because information can be stored indefinitely in numerous archives. Beyond these identified characteristics, Shifman stresses that the meme is the best concept to encapsulate some of the most fundamental aspects of the internet in general or web 2.0 culture in particular. Thus, owing to these fundamental aspects , memes have been ascribed three main attributes which are relevant to the analysis of contemporary digital culture: a gradual propagation from individuals to society, reproduction via copying and imitation, and diffusion through completion and selection. All these attributes engender Shifman to claim that memes are "pieces of cultural information that pass along from person to person, but gradually scale in a shared phenomenon. They spread on a micro basis, their impact is on the macro level: memes shape mindsets, form behaviour, and actions of social groups". Internet memes are social constructs consciously designed to communicate deep-seated discourse that signals social concerns and seeks justice. They are cyber weapons now integrated into our digital social life and discourse structure. "…we live in an era driven by a hypermemetic logic, in which almost every major public event sprouts a stream of memes", Shifman . In the past, and at this present time, the NPF and its officers have constituted one of the major public discourses that have provoked internet memes from Nigerian netizens. These memes incorporate, to a large extent, Eno-Abasi's description of the ignominious activities of some men of the NPF: "I will kill you and nothing will happen." "If you say anything again I will just waste you. I've wasted many of your type, so you will not be the first or the last." These and many more obscene lines have become part of the refrain that some officers and men of the Nigeria Police Force bandy when they get berserk and try to ride roughshod over citizens that they are supposed to protect. Across that country, scarcely does a day pass without a trigger-happy belligerent policeman brutalising, robbing, maiming, or extorting an innocent member of the public that has the misfortune of crossing their path. The critiquing of this level of policemen's abuse of power which reflects lawlessness, impunity and corruption by Nigerian netizens in diverse memetic forms is a pop/cyber culture that may appear humorous and inconsequential. However, a critical examination of these pieces reveals that they possess inherent salient points beyond being amusing. There have been a number of scholarly linguistic works on police discourse. Beyond the shores of Nigeria, scholars in the humanities have critically engaged police discourse in the areas of policing and intelligence operations, police interview discourse, police and criminal justice discourse, police comments, policing and racial profiling, dynamics of power and resistance in police interview discourse, respectively . In the same line of discourse, previous studies in Nigeria have investigated the English language competence of the men of NPF, conversational strategy in police-suspect interaction, forensics of police investigation reports and court proceedings, Nigerian police-suspect discourse, and concealment in police-suspect interaction, respectively . However, a linguistic consideration of corruption-related activities of officers of the NPF at police checkpoints has not been devotedly given the required attention. Thus, this current investigation departs markedly from the highlighted works above given that it concentrates on Nigeria policemen's abuse of power digital memes -multi-moulded socially constructed Nigeria police discourse, instantiated digitally to satirise NPF's abuse of power at police checkpoints and privilege reformative advocacy. The rational for focusing on such archetype of meme is because policemen's abuse of power at checkpoints is a popular public discourse in Nigeria. This paper adopts Gunther Kress' social-semiotic theory of multimodality taking into account that it enables a deeper comprehension of multimodal components. Kress' social-semiotic theory of multimodality came to rival the general semiotics , and evidently shows Kress' generalist linguistic orientation in the universals of sign-making and understanding, and displays his ideological engagement which puts the interests of the sign user first and motivation -in Saussure's senseabove all else .The socio-semiotic theory diverges critically from Saussure's fundamental principle that affirms that language should be viewed as a system of arbitrary signs governed by rules which fix the inherent variability of language and keep it stable. As an alternative to this, the theory claims that language is intrinsically fluid and dynamic because the relationship between the material form of the sign and its meaning is not arbitrary but motivated by the sign-maker's interest. As affirmed by Kress and Jewitt , people "use the resources that are available to them in the specific socio-cultural environments in which they act to create signs", so signs are motivated, and not arbitrary. Signs are not static; they are always newly made according to context of use. Thus, the meanings of signs are constantly transformed as sign-makers select the most apt sign from the available resources based on context and sign-maker's interest . Signs exist in all modes. Kress conceptualises mode as "a socially shaped and culturally given semiotic resource for making meaning". Mode, according to Kress includes such examples as image, writing, layout, music, gesture, speech, moving image, soundtrack and 3D objects deployed for representation, communication and essential contributory phenomenon to the meaning of a sign-complex. Humans engage with the world through these socially made and culturally specific resources and they do so in ways that that arise out of their interests. With these magnitude properties of mode, Kress argues that the meaning of any message is distributed across a range of modes and that language which "had been seen as a full means of expression; as the foundation of rationality; sufficient for all that could be spoken and written, thought, felt and dreamt is now seen as a partial means of doing these" . Mode is said to offer meaning-laden means for making the meanings that we wish or need to make material and tangible -realising, materialising meanings. Modes, in their varying affordances, make it possible to make meanings material with specific ontological effects, according to the intentions of rhetor and designer. In realising meanings, design is another concept central to the socio-semiotic theory. It anchors on apt specifications deployed by sign-makers to achieve meaning in totality. Design, in the submissions of Kress , refers to "the use of different modes -image, writing, colour, layout -to present, to realize at times to contextualize social positions and relations, as well as knowledge in specific arrangements for a specific audience". Design in socio-semiotic does not recline on language alone. It is encompassing in nature in meaning representation; enhanced by contemporary media. Based on this background, this paper examines Nigeria policemen's abuse of power digital memes -multi-moulded socially constructed Nigeria police discourse, instantiated digitally to satirise NPF's abuse of power at checkpoints and provoke reformative advocacy. Nigeria policemen's abuse of power digital memes fit into Kress' theoretical underpinnings of socio-semiotic theory because they are modes with sign-complexes and specific designs driven by motivation and interest of sign-maker ; apt contextually. Moreover, Kress' socio-semiotic theory, in this current investigation, is suitable as it facilitates a deeper understanding of Nigeria policemen's abuse of power memes in a digital context. In this regard, this paper is able to bridge digital satire with reformative discourse by investigating how these policemen's abuse of power memes might serve as communicative tools to prompt awareness, dialogue, and influence reformative advocacy in the context of policemen's misconduct at checkpoints in Nigeria. --- METHOD For data, this study gave preference to Opera News by virtue of the fact that this news feed is one of the most used news and content apps in the world. Besides this rationale, the flexibility of Opera News allows African authors and bloggers to create online content and share such with the fastgrowing Opera user base of more than 350 million users worldwide . There are other social media platforms and online forums but this investigation relies on the Opera News for its data specifically because it accommodates a great number of memes of various contents on the Nigeria Police Force unlike others. From the Opera News forum, seven digital police memes that portray abuse of power at police checkpoints were purposively selected. This research endeavour focuses on Nigeria policemen's abuse of power at checkpoints because such discourse has essentially become a complex and multifaceted issue that resonates with the populace because of their experience with the Nigeria policemen, and moreover, reports of abuse by some policemen at checkpoints has gained visibility through user-generated content than other policemen-related discourse. It is important to also state that some of the selected memes comprise known pubic figures in Nigeria. With respect to ethical consideration, Opera News is a public-oriented forum where usergenerated and algorithmic contents are meant for public consumption. More to this is that the public figures used in the memes require lower expectation of privacy because of their profession, and the content with which they have been used involves no potential harm to them. Thus, requesting for permission to carry out a scholarly investigation of the memes that involve them may not be necessary. This study is, therefore, motivated to subject the selected Nigeria policemen's abuse of power digital memes to qualitative analysis through Kress' socio-semiotic theory. In effect, a qualitative analysis in this paper is intended to provide an in-depth understanding of the socio-semiotic variables inherent in the memes by uncovering the nuances of motivations, embedded messages and the contextual aptness of signs, contributing to a comprehensive interpretation. The method of analysis, therefore, involves a comprehensive analysis of motivational factors, interests, modes of communication, design elements, aptness within socio-cultural context, sign-complexes, and functional loads to understand and reformulate the discourse around Nigeria policemen's abuse of power at police checkpoints through the selected digital memes. --- RESULT AND DISCUSSION --- Figure 1. Meme taken from the Opera News Nigeria Figure 1 is a socio-semiotic mode borne out of the meme/sign-maker's social interaction with the officers of the NPF. It is an image with five major signifiers culminating into a sign-complex with functional loads. The meme-maker, no doubt, must have, at a number of occasions, experienced what is fully designed in the entire mode. The mode showcases an embedded culture of the NPF at a supposed checkpoint -a policeman extorting a naira-note from a driver in the full glare of the public. Hence, the meme-maker is motivated to put all the signifiers together because this sort of unenviable police conduct in Nigeria, especially at checkpoints mounted by the officers of the NPF, has been discredited as corrupt practices over time. The motivation that underscores this meme is further established a by the Daily Post reporter thus: The Nigerian Police Force has expressed concern over the activities of some officers collecting money especially from motorist at police checkpoints. The Deputy Inspector General of Police supervising north-west zone, DIG Aminchi Sama'ilaBaraya stated this during an inspection tour and meeting with political stakeholders held at the Jigawa State police command headquarters in Dutse. He strongly condemned the act and other forms of dehumanisation or collection of money especially at motorist checkpoints. The display of unprofessionalism and official corruption by the officers of the NPF constitutes builtin discourse in Nigeria that cannot but attract concerned interest such as the meme above. The picture signifiers: a police man, a naira-note, a driver's hand and a vehicle are apt contextual signifiers of this condemnable act. This is further corroborated by the caption -"Anything for the boys?". "Anything for the boys?" is a recurring question asked by the officers of the NPF at police checkpoints in Nigeria. It is a jocular way of illegally requesting money or other inducement from motorists. Largely, therefore, these signifiers are apt and persuasive given that as sign-complexes, they in unison expresses a signified of satire in the context of this contemporary topical issue such as contemptible bare police extortion which proves the NPF's progressive degeneration of policing; thereby suggestive of reorientation and reformation advocacy of the NPF. The NPF is undeniably a very sensitive security agency constitutionally saddled with the responsibility of maintaining public law and order. However, and regrettably so, some officers of the NPF are neglectful of these civic duties. Figure 2 is created to respond and reflect on such dereliction. As a socio-semiotic mode, it can be ascertained that this meme is a product of the meme maker's direct or indirect encounter with some exploitative officers of the NPF. This unwholesome encounter informs the motivation for this sign. The meme is designed with two major apt signifiers . Naturally, the activity being embodied in the mode -three police men apprehending a man, is reasonably representative of what law enforcement agents should do especially in cases of alleged crime. However, the signifier 'You wear red shirt inside a black car, how many cult you dey [lit. You wear a red shirt in a black car. How many cult groups do you belong to? ostensibly exhibits the nature and goal of such apprehension. Although a mere appraisal of the text signifier and the image signifier shows evident contradiction, given that the individual in question does not 'wear a red shirt but a red cap'. Nonetheless, a critical engagement of these ensemble sign-complexes characterises the meme maker's interest as it affects the NPF. Similar to what motivates the sign maker's interest is what Adediran underscores when he asserts that: Several victims of police harassment recounted their ordeals in the hands of the officers at a press conference in Lagos on Thursday. The press conference organised by the Network on Police Reforms in Nigeria, focused on cases of police corruption and abuse across Nigeria and the need to drive a holistic reform in the Force. Among the excesses of the police force raised by NOPRIN are indiscriminate arrest and detention of young men, violation of human rights, raid of public places by rogue police, use of torture to extort information from suspects, and many other oppressive activities of police officers The internet meme is a highly exploitable tool for civic communication by grass root action since it could reach audiences outside their usual demography .The sign in question, represented by the meme, is a civic mode with apt signifiers. The image signifier and the text signifier portray the level of unprofessionalism and crass abuse of office by some officers of the NPF. The misrepresented notion such as being cladded in a red shirt and in a black car to connote being a member of a cult by a supposed professional body is shockingly bad and myopic. Hence, this collection of image and text is apt signifiers deployed to express a satire that unearths symbolic blind allusions to some wrong societal beliefs initiated by some officers of the NPF to harass and deprive people, especially Nigerian youths of their human rights, especially the right to freedom of dressing; thereby calling for a re-education of the NPF of what their civic duties really are. Figure 3 is analogous to the previously discussed meme. It is a meme, a socio-semiotic mode, with ensemble text signifier (e.g. Me the day #9japolice asked me for my jack after checking all my papers. [lit. This is me, the day Nigeria police officers asked me for my jack after checking all my vehicle documents.]) and image signifier . This meme, ordinarily, depicts one out of the kinds of activities that transpire at police checkpoints. Significantly, the rationale for checkpoints is for the men and women of the NPF to ensure public safety and engage in or check what is lawfully required. Bailey confirms this when he states that "Olumuyiwa Adejobi, the acting Force Public Relations Officer has said that vehicle licence, drivers licence and certificate of insurance, especially for private car owners are the documents that need to be tendered to the police on demand at any checkpoint or routine patrol." This validates that any other request made by the men of the NPF drafted to checkpoints for mere routine checks is unlawful. The interest that therefore motivates this sign-complex is inherent in the deviant conducts of some officers of the NPF at mounted checkpoints who make absurd routine checks and demands of motorists. The vehicle jack being demanded for after checking vehicle documents as represented in the text signifier symbolises every other aberrant demands often made by some officers of the NPF at checkpoints. This is frequently done in order to probe for fault from an ill-informed/informed car owner or driver who may eventually become a cash cow for such men on duty even if such owner/driver has all required documents. The offshoot of such nonstandard gesture from the men of the NPF is often what is represented in the image signifier . This archetypal sign-complex is apposite bearing in mind the hordes of similar human right violation-experience Nigerians have had with the officers of the NPF. As it is, "members of the pubic have been at loggerheads with some police officers at checkpoints over indiscriminate demand of car documents, laptops and even mobile phones receipts" . This pool of image and text signifiers that make up the meme is therefore apt given that it expresses a satiric signified which exposes some police officer's duty deviance; thus hints that police officers should work within the ambits of their duties, and be retrained on human rights best practices. Figure 4 is a socio-semiotic mode with a conversational text signifier and image signifier ).The meme vividly relates what seems to be an amazing conversation between a son or perhaps a daughter and his/her popsy on passing through a police checkpoint without a bribe being demanded by the officers. Of a fact is that requesting for or inducing bribe from motorists appears to be the standard conduct of the police on any highway where checkpoints are mounted in Nigeria. In the words of Nwogbunyama , "we are all used to the checkpoints Nigerian policemen mount on roads. These checkpoints can be on highway, intra state roads and even in streets. …things like extortion, harassment and even bullying are what most Nigerians face." This constancy of checkpoints and what they are put up for is further corroborated by Iremeka , who draws from The Guardian investigation which reveals that the "proliferation of police checkpoints across the country is mostly for economic considerations than provision of security". Hence, the design of the meme above with the signcomplexes is motivated by the diehard exploitative nature of police checkpoints in Nigeria. This obstinate exploitative characteristic of police checkpoints is portrayed in the image signifier -the agape and astonished image of a popular Yoruba comedian and the text signifier (e.g. ADONBILIVIT. [lit. I don't believe it]) -which allude to the incredibility of the text signifier . Essentially, the image signifier accompanied by the text signifier , in this context, is a morphological and syntactical deviation from the norm with the underlying sense of disbelief and sarcastic response made towards the text signifier .This resource foregrounds the constant act of subornment at police checkpoints in Nigeria in place of what is expected by the law. Given this circumstances, it could therefore be affirmed that the image and text signifiers are apt in view of the fact that they bring to the fore a satiric signified which reveals the motif and proliferation of checkpoints; thereby calling for the invocation of the legitimate use of police checkpoints. Figure 5 is a socio-semiotic mode that no doubt represents the meme-maker's social interaction with the men and women of the NPF. It is a mode which comprises a text signifier (e.g. How #9japolice stares at you when they realise all your papers are complete plus you even have fire extinguisher. [lit. How Nigeria policemen stare at you when they realise all your vehicle papers are complete, and in addition to this, you have a fire extinguisher]) and an image signifier ). The two are combined into a sign-complex with functional loads. The officers of the NPF who mount checkpoints in Nigeria often do so with the sole purpose of extortion and less otherwise. Thus, a direct means to extort motorists is to usurp some professional duties which are not directly theirs. They do this, among others, by requesting to check vehicle papers and other roadworthy car packages such as fire-extinguisher. Any vehicle user that does not have this becomes an immediate prey for exaction. In the same vein, they may get frustratingly livid if vehicle owners have all that are being demanded. In such situations, officers of the NPF unsatisfyingly engage other means to induce bribes by simply looking for faults to point at. An online news outlet, Thisdaylive further authenticates how depraved policemen at checkpoints could be thus: Exactly 89 years after its establishment, members of the force have emerged as predators, rather than protectors, {…} and become an icon of unbridled corruption, unprofessional conduct and violence in the country. Particularly common on the list of the groups the police extort most are commercial and private vehicle owners in the country. Most embarrassing are policemen who wilfully extort motorists or others even when they have not contravened the law. Wilful extortion by Nigeria policemen at checkpoints is a by-product of characterised frustration, and this circumstance is what motivates the sign-complex of the represented signifiers as pertinently designed and impelled by the interest of the meme-maker. This sign-complex is, therefore, apt in this circumstance to depict a satiric signified of detrimental frustration of some officers of the NPF; thereby calling for a better oriented and funded force in order to guide against police assault and extrajudicial killings. ) and an image signifier ). This mode, with sign-complexes, has functional loads. For instance, the text signifier , uttered by policemen in the context of police checkpoints in Nigeria often portends an occasion to induce extortion. This is completely a shared knowledge, in particular, for motorists that at such moments, naira-notes necessarily change hands. This comes readily handy from motorists towards a set of already-smiling avaricious policemen to establish the normalcy of such act, and moreover, avoid being delayed by such policemen or a means of distraction to checkmate being checked. As established by Premium Times , "cases of extortion by the officers are so commonplace that most citizens consider them the norm". In the same vein, Onwuka corroborates this model of operation at police checkpoints when he confirms that "the extortion of money from motorists at police checkpoints has become a routine that it has taken on the nature of a standardized tool". This motivated ensemble of text and image signifiers as displayed in the meme in question propels and symbolises a common culture of extortion, where the motorist already knows that s/he has to part with some naira-notes and the officers on duty are privy to this. The circumstance portrayed in the meme is motivated by the interest of the meme-maker to signal certain signified in relation to the operation of some officers of the NPF at checkpoints. The signcomplex that makes the body of the meme is indeed apt given that it exemplifies a force that is material-conscious rather than public safety-conscious. Thus, a satiric signified of checkpoint's main duty negligence is established; by this means convening for a conscientious force in the line of duty. Figure 7 is a socio-semiotic mode with a blend of a text signifier and an image signifier .The synthesis of the two in the body of the mode, therefore, has its functional load. In Nigeria, there are many law enforcement agencies. According to the Constitution of Nigeria, and other federal laws establishing the agencies, each agency has its own clear duties and functions. Nevertheless, there have been controversies on which agency checks a driver's licence between the NPF and the Federal Road Service Commission . Either way, a driver's licence is one of the major documents that vehicle users must possess to navigate Nigerian roads. Hence, as re-counted by Toromade , "according to the spokesperson of the NPF, Muyiwa Adejobi, police officers can only ask you for three things at checkpoints: vehicle licence, driver's licence and certificate of insurance". This, thus, confirms that the demand to view driver's licence is one of the operational jurisdictions of the NPF. But, as stressed by Adejumo , NPF men "compromise their official duty by taking arbitrary 'fines' from motorists whose vehicle licence or insurance papers are outdated or missing". This official circumstance of misconduct, therefore, motivates the design of the meme at the centre of discourse. The text signifier and the image signifier embody the men of the NPF, who regale at the instance of traffic infractions such as that of using expired driver's licence by motorists in Nigeria. This further implies a set of officers of the NPF who privileges and exploits breach of law for financial gains. On the whole, therefore, these signifiers are apt and persuasive given that as sign-complexes, they altogether articulate a signified of satire in the context of some Nigeria policemen who are supposed to sustain and keep the law, but encourage lawlessness because of illicit pecuniary gain derived through driving with an expired driver's license thereby suggestive of reorientation and reformation advocacy of the NPF. --- CONCLUSION In a divergence from the concentration of the identified extant literature on police discourse, this paper has examined Nigeria Police digital memes -multi-moulded socially constructed Nigeria police discourse, instantiated digitally to satirise NPF's abuse of power at checkpoints and prompt reformative advocacy. In interpreting the selected digital police memes, this paper engages the theoretical underpinnings of Gunther Kress' social-semiotic theory of multimodality. Through its theoretical apparatus, functional loads such as contemptible open extortion, symbolic blind allusions, duty deviance, wrong motif towards checkpoints, detrimental frustration, checkpoints' main duty negligence and lawlessness were established as what the selected police digital memes critically reveal as regards the NPF and checkpoints in Nigeria. Precisely, the functional loads characterise the prevalence of corrupt practices within the force, wilful demonstration of ignorance in the face of systemic policing issues, ill-commitment of some Nigeria police to duties, wrong motif towards checkpoints and their potential misuse. With these findings, this current endeavour maintains that the satirical philosophy embedded within most Nigeria policemen's abuse of power digital memes draws attention towards reformation and reorientation of members of the Nigeria Police Force. Thus, with respects to extant studies on police discourse, this investigation, therefore, contributes to a nuanced understanding of how digital communication tools can serve as catalysts for reformative advocacy, particularly within the context of law enforcement reforms. Since this investigation has concentrated on the abuse of power at checkpoints by some Nigeria policemen,

Several Studies have investigated Nigerian police discourse. However, digital memes deployed to reflect corruption-related activities at police checkpoints have attracted inconsequential research. This paper, therefore, examines selected Nigeria Police digital memes with the aim to analyse qualitatively embedded socio-semiotic functional loads instantiated digitally to satirise corruptionrelated activities of some Nigeria policemen at police checkpoints and how these privileges reformative advocacy in Nigeria. Seven digital memes that thematised Nigeria policemen's abuse of power at checkpoints were purposively selected from Opera News Nigeria, an app that allows African bloggers to create online content and share such with a fast-growing subscriber base. In interpreting the selected digital police memes, this paper adopts the theoretical underpinnings of Gunther Kress' (2010) social-semiotic theory of multimodality. Through its theoretical apparatus, results established that functional loads such as contemptible open extortion, symbolic blind allusions, duty deviance, wrong motif towards checkpoints, detrimental frustration, checkpoints' main duty negligence, and lawlessness characterised the police memes. These characteristics, in sequence, highlight the prevalence of corrupt practices within the force, wilful demonstration of ignorance in the face of systemic policing issues, ill-commitment of some Nigeria policemen to duties, wrong motif towards checkpoints, and their potential misuse. This paper affirms that the satirical philosophy inherent within most Nigeria policemen's abuse of power digital memes points towards reformation and reorientation of members of the Nigeria Police Force. This investigation, therefore, contributes to a nuanced understanding of how digital communication tools can serve as catalysts for reformative advocacy within the context of law enforcement reforms.

INTRODUCTION Drunken driving is a significant risk in road traffic. Drunk drivers are more likely to cause fatal accidents and injuries than other drivers . Driving under the influence of alcohol or drugs is not just a risk for other people, as the offenders suffer negative health and social effects themselves as well. Several studies have found a high mortality among DUI offenders . Causes of death that are especially associated with abuse of alcohol are more prevalent among the DUI population . Studies link alcohol use and abuse to social factors such as low socio-economic status, neighbourhood poverty or deprivation, unemployment and family structure . In a Finnish nationwide survey, study on young adults substance use and dependence was linked with unemployment and low education . However, it has also been suggested that other environmental social factors, even more than the family, may affect alcohol use in adolescence . In a cross-national survey of youth drunkenness, however, the effects of family affluence and parental occupation were found significant in only some countries. For girls, there was only little support for the effect. For Finland, the study showed that family affluence was not connected to drunkenness in adolescence, but parental occupation did affect drinking . Some results have shown no connection between parental social position and excessive alcohol use in adulthood . Drunken driving is affected by social factors, similarly to alcohol use. In register-based studies where all the information is obtained from administrative registers, drunken driving has been linked with low parental education and living conditions , poor performance in school and single-parent family background in Finland, and social class of origin and low level of education in a nationwide Swedish study . Connections between DUI and low education , low income and occupational status have been reported in surveys with self-reported information on social position. Also recidivism of DUI has been associated with occupational status low education and not being married . A nationwide register-based study found strong evidence for social factors correlating with DUI of drugs . A drunk driver is not necessarily a problem user of alcohol or a marginalized person. Drunk drivers nevertheless are at higher risk of alcohol abuse. Previous studies show that drunk drivers have health problems that are strongly linked with alcohol use . Knowledge about the social backgrounds of drunk drivers is needed to deal with the public health problem. The study investigated the social determinants of drunken driving. It was hypothesized that poor social position is associated with higher risk of drunken driving in men and women while the strength of the association may vary by age and sex. For this purpose, a register of DUI arrests was combined with registers of social variables. Extensive register data on drunk drivers linked with various register data enabled us to use a large number of variables in different models, as the data is exceptionally extensive. Such nationwide register-based studies have previously been conducted only in Finland and Sweden. --- MATERIAL AND METHODS The Finnish legislation prohibits the operation of a motor vehicle in road traffic when the driver's blood alcohol concentration exceeds 0.05% . The offence is considered aggravated when the BAC is 0.12% or above . Since 1998, also precision breath testing with limits of 0.22 and 0.53 mg of alcohol in 1 l of exhaled air has been legally binding. The police may perform random road-side alcohol testing from the driver's breath with no particular suspicion. When preliminary testing suggests DUI, a further precision breath test or blood sample is required. This study is based on the database of all suspected drunken drivers apprehended by the police between 1993 and 2007 in Finland. Information on whether the driver was convicted for DUI was not available. The first suspected drunken driving case of each driver was selected for analysis. The drivers in the study had no previous record of DUI during at least 5 years prior to the arrest. All drivers with a history of suspected DUI of drugs were excluded from the data. Only drivers within motorized road traffic arrests were studied. Data on the social background of the study subjects were obtained from the employment statistics also maintained by Statistics Finland. This annual statistics covers the population's economic activity and employment. The data include information on the parents' education and income, the subject's own education, marital status, employment status, income, possession of a car and socio-economic status. The data linking was done by using personal identification codes. To protect the privacy of individuals, the data were anonymized, and a 50% random sample was drawn from the original data of DUI suspects. An additional sample was drawn from the general population, as a reference group for the DUI offenders. The reference population was matched for age and gender. The study design, data collection and linking were approved by the Institutional Review Board of the National Institute of Health and Welfare. Two age groups were analysed separately: young persons aged 15-24 years, and the economically most active population aged 25-64 years. Information on social variables was from the year prior to the first arrest, as this was considered to precede the DUI offence. Information on socio-economic status was not available for each year, so the latest available information prior to the offence was used in these cases. Education was classified by Statistics Finland as high , medium or low education. This classification was used for the working-aged population. To study the impact of education, the younger group was divided into three categories: low education and age 20-24; medium or high education and age 20-24 and persons aged 15-19. In this latter group, separation according to education was not possible. Five income categories were established within the age and sex groups for analysis. The people with no income formed one group, and the population with an income was divided into quartiles. Because income is based on tax registers, only the income under taxation was available. Level of urbanization had three levels: urban, semi-urban and rural. Possession of a car is a binary variable, indicating whether a person owns a car or is registered as a main user of a car owned by someone else . Logistic regression models were fitted to the data to examine the risk factors for drunken driving. Observed risk is reported as odds ratios and their 95% confidence intervals. Single-variable models were used to compute the initial ORs. Variables were entered in the subsequent models one by one in order presented in Tables 2345. The order of entered variables was determined by the factors' estimated temporal proximity to the arrest, i.e. parental background is the most distant, while possession of a car is considered the most proximate to the incidence . The final model was fully adjusted for all variables presented. No interactions between the variables were included. --- RESULTS The data included altogether 81,125 people arrested for suspected drunken driving and a reference population of 86,279 people. Of these, 87% of the DUI suspects were males, and 28% were apprehended more than once between 1993 and 2007. One-third of the first-time offenders were between 15 and 24 years of age, and two-thirds were between 25 and 64 years of age. Mean BAC was 0.141% for men and 0.138% for women. Distributions of the social and economic variables are presented separately for both age groups and sexes in Table 1. --- Social determinants of DUI among youth The ORs for the younger age group are presented in Tables 2 and3. The results for univariate models showed that the subjects coming from lower income families had higher ORs for drunken driving. Father's income in lowest quartile had odds 2.2 for men and 2.0 for women. Mother's income in lowest quartile had odds 1.9 for men and 1.7 for women. When model was adjusted for subsequent variables in Models 2-8, the odds of income attenuated, but the association still remained for the two lowest income quartiles in men. Parental education had similar odds for both sexes in the initial models. Low and medium level parental education were associated with higher odds of drunken driving for their offspring. After adjusting for other variables, the ORs attenuated but tended to remain. The father's education had higher odds than the mother's education for both sexes and education levels in the fully adjusted model . Level of urbanization had some or no effect on the odds of drunken driving for the age group. An urban living environment was associated with a slightly lower odds of drunken driving in the fully adjusted models. The odds were constant throughout the different models. Education displayed a similar pattern for both sexes. Men aged 20-24 years with no secondary education had the highest ORs . When labour market status and income were adjusted for, the odds for education became higher for women . Those with a low education committed more drunken driving offences after 20 years of age than others with the same income level. Unemployment correlated strongly with the odds of drunken driving. Long-term unemployment was more harmful than shorter unemployment in the univariate model. Being a student meant lower odds for drunken driving for men . In the univariate model, being a conscript in the defence --- Continued forces was connected with less drunken driving . When income was adjusted for, the conscripts showed elevated odds for drunken driving. Low odds were therefore a consequence of low income. Being outside the work force for other reasons also was a risk factor after income was adjusted for. The subject's own income had a reverse association with drunken driving compared with the parental income. A higher income was associated with higher odds of drunken driving. There was a gradient by income group, and it was significant for both sexes. Living alone was a risk factor for drunken driving . Being single was associated with higher odds for drunken driving than being married. The odds were higher among women than among men . Divorced persons, however, had the highest odds . Possession of a car was statistically significant for both men and women . --- Social determinants of DUI among working-aged people As in the case of young persons, an urban living environment was associated with less drunken driving than a rural environment . The difference was small but statistically significant. Education had a gradual effect on the odds. A lower education signalled higher odds of drunken driving for both men and women . The odds for education attenuated after other variables were adjusted for . Socio-economic position was related to drunken driving among both sexes. In the univariate model, blue-collar workers and entrepreneurs had the highest odds for drunken driving. After adjusting for urbanization and education, the odds for blue-collar workers and entrepreneurs attenuated. In these cases, education explained the high odds. Lower white-collar workers had a higher odds initially , but after adjusting for all other variables, the ORs attenuated . For farmers, the results were different for the univariate model and the fully adjusted model . Adjusting for education makes the higher odds disappear, and farmers do not commit more drunken driving offences than other people with the same level of education. Compared with employed people, unemployment had the highest odds for drunken driving. For people with long-term unemployment, the odds were the highest initially . When model was adjusted for subsequent factors , especially income, long-term unemployment did not differ from short-term unemployment. Unlike in the young age group, work disability pension correlated with elevated odds of drunken driving . Other types of pensions and being outside the work force for other reasons meant higher odds of drunken driving for men, but not for women. The income of working-aged people functioned differently than the income of young persons. The ORs of drunken driving had a gradient with lower income showing higher odds; the ORs for the lowest income quartile were 3.0 for men and 2.5 for women. After all adjustments, the two lowest quartiles showed elevated odds of drunken driving, whereas the group with no income showed no difference compared with the two highest quartiles. Living alone was a risk factor for drunken driving also for people aged 25-64 years. Being single was associated with slightly elevated odds. Being divorced or widowed were associated with higher ORs. For women, possession of a car had high odds in the univariate model as well as in the fully adjusted model , but among men the association was negligible. --- DISCUSSION --- Main results The association between social background and drunken driving was studied separately for both sexes and in two age groups. It was assumed that the effect of social background varies between men and women, and between young people and working-aged people. However, the associations were largely similar throughout the age and sex groups. In general, poor social position was associated with a higher probability of drunken driving. Some interesting differences in the associations were nevertheless found. For all groups observed, low level of education, unemployment, living alone and being divorced had highest odds for drunken driving. Income functioned in different ways in the different age groups. A high income entailed a high risk of drunken driving for young persons. For working-aged people, low income was associated with a higher risk. The result for the working-aged population is in line with other indicators of poor social position. Working at a young age often means early exit from education. Working and a higher income may thus indicate more drinking for young people . Possession of a car was associated with drunken driving among young persons, as well as working-aged women. For working-aged men, such an association did not exist. A low level of parental education and low Mother's income Highest quartile 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Third quartile 1.3 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Second quartile 1.6 1.2 1.2 1.2 1.2 1.2 1.2 1.1 Lowest quartile 1.9 1.3 1.3 1.2 1.2 1.3 1.2 1.2 (1.1-1. 3) No income or person missing 1.5 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Father's income Highest quartile 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Third quartile 1.4 1.1 1.1 1.1 1.0 1.1 1.0 1.0 Second quartile 1.7 1.2 1.2 1.2 1.2 1.2 1.2 1.1 Lowest quartile 2.2 1.4 1.4 1.4 1.4 1.4 1.3 1.3 No income or person missing 1.7 1.2 1.2 1.2 1.2 1.2 1.2 1.2 Mother's education High 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Medium 1.8 1.4 1.4 1.4 1.3 1.3 1.3 1.3 Low 2.1 1.5 1.5 1.4 1.4 1.3 1.3 1.3 Father's education High 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Medium 2.0 1.6 1.6 1.6 1.5 1.5 1.5 1.5 Low 2.3 1.7 1.7 1.7 1.5 1.5 1.5 1.5 Level of urbanization Rural 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Semi-urban 1.0 1.1 1.1 1.1 1.1 1.1 1.1 Urban 0.8 0.9 0.9 0.9 0.9 0.9 0.9 Age × education 20-24, high 1.0 1.0 1.0 1.0 1.0 1.0 20-24, low 2.4 2.1 2.2 2.4 2.5 2.5 Labour market status Employed 1.0 1.0 1.0 1.0 1.0 Unemployed up to 12 months 1.6 1.5 1.9 1.9 1.9 Mother's income Highest quartile 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Third quartile 1.2 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Second quartile 1.3 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Lowest quartile 1.7 1.2 1.2 1.2 1.2 1.2 1.2 1.2 No income or person missing 1.2 0.9 0.9 0.8 0.9 0.9 1.0 0.9 Not necessarily all of the factors discussed here directly expose people to drunken driving. In some cases, the person's social background might rather indicate alcohol use, risk-taking behaviour, or other factors, which in turn worsen his or her social position, leading to sickness or unemployment, for example. Factors such as marital status and employment may play a role through major life events, such as divorce or lay-off from work. --- Methodological considerations The strength of this study lies in the quality of the data with regard to its extent and nationwide coverage. Finnish register data have also been evaluated as a good research tool . The possibility to analyse 50% of all arrested drunk drivers during a 15-year period, with a reference population, makes the statistical inference reliable. The socio-economic data contain a large set of variables which enabled diverse analyses. The case-control study design allows only the study of associations between variables, not causal relations, and this must be kept in mind when interpreting the results. With regard to indicators, register-based data are more limited than data which are collected specifically for a certain study, because the indicators and time points of data collection cannot be strictly controlled by the researcher. When we consider the nature of drunken driving, only the arrested drivers are known, as the true prevalence and the true socio-economic differences remain hidden. --- CONCLUSIONS Although road safety in Finland has improved, one-fourth of fatal road traffic accidents have involved a drunk driver . At the same time, DUI offenders have a high mortality rate due also to various other causes, including serious injuries, poisonings and chronic diseases . Preventing drunken driving by focusing on reducing its causes could improve not only road safety, but also public health by improving the health of this high-risk group. Social differences obviously remain a big problem in the phenomena of drunken driving. They arise from more prevalent alcohol abuse among less educated, poorer people and those living alone. Thus, the problem is not simply one of traffic control and anti-DUI legislation. Drunken driving and the social differences observed within it should also be seen as a public health problem of alcohol abuse and as a problem of the society's alcohol and social policies. Interventions against DUI behaviour should not concentrate only in prohibiting the driving, but also limiting the substance abuse. Narrowing the socio-economic health differences is now an important goal of the policy programmes in Finland . As socio-economic differences in drunken driving can already be found in the age group of 15-24 year-old persons, and alcohol-induced disorders are persistent among them, early intervention is of utmost importance for young drunk drivers. --- parental income of young people also had some correlation with drunken driving. The main findings of a poor social background were expected, as they are in line with earlier studies on social determinants of alcohol abuse and drunken driving. Similar results have been found for drivers arrested for DUI of drugs . In Finnish cohort-based studies, a connection between low parental education and crime has already been established for boys . The association between parental education and drunken driving was now found also for girls. Although some studies indicate that poor parental social factors affect substance abuse , a recent nationwide survey of almost 21,000 students in the USA also suggests that there may be more substance use among children coming from a high socio-economic background . Also a cross-national survey found an association between high parental occupation and a low level of drunkenness for both boys and girls in Finland . In our study, however, low parental income was connected to more drunken driving among young people. Already in the age group of 15-24-year-olds, significant differences were detected between social groups. As parental education, income and socio-economic position are associated with drunken driving, social and economic differences are formed early in life. Unemployment and low education have also been linked with substance use and dependence among 19-34-year-old persons . Despite the efforts and policies to diminish health inequalities between various social groups, the differences persist and have even grown in Finland . Alcohol use and abuse, binge drinking and alcohol-related problems are known to be linked with socio-economic position . The results of this study demonstrate a connection between several socioeconomic factors and drunken driving. Level of urbanization Rural 1.0 1.0 1.0 1.0 1.0 1.0 Semi-urban 1.0 1.0 1.0 1.0 1.0 1.0 Urban 0.8 0.9 0.9 1.0 1.0 1.0 Education High 1.0 1.0 1.0 1.0 1.0 1.0 Medium 1.9 1.9 1.4 1.4 1.3 1.3 Low 2.2 2.1 1.9 1.7 1.5 1.5 Socio-economic status Upper white-collar 1.0 1.0 1.0 1.0 1.0 Lower white-collar 1.5 1.3 1.2 1.2 1.1 Blue-collar 2.5 1.8 1.6 1.5 1.5 Farming 1.3 0.9 1.0 0.8 0.9 Level of urbanization Rural 1.0 1.0 1.0 1.0 1.0 1.0 Semi-urban 1.0 1.1 1.0 1.0 1.0 1.0 Urban 0.8 0.8 0.9 0.9 0.9 0.9 Education High 1.0 1.0 1.0 1.0 1.0 1.0 Medium 2.0 2.0 1.6 1.5 1.5 1.4 Low 2.1 2.0 2.2 1.9 1.9 1.9 Socio-economic status Upper white-collar 1.0 1.0 1.0 1.0 1.0 Lower white-collar 1.5 1.2 1.1 1.1 1.1 Blue-collar 2.9 1.9 1.7 1.7 1.6 Farmer 1.2 0.8 0.8 0.7 0.9 Enterpreneur 2.6 1.9 1.9 1.8 1.

The aim of the study was to examine the association between social background and drunken driving. Methods: A Finnish register on suspected drunken driving was combined with data on social background. There were 81,125 drivers arrested for drunken driving and 86,279 references from 1993 to 2007. Results: A low level of education, unemployment, living alone and divorce were strongly associated with drunken driving. In addition, for persons aged 15-24 years, low parental education and income, high own income and possession of a car correlated with higher odds of drunken driving. For working-aged men and women, low income was associated with a higher risk of drunken driving. For working-aged women, also possession of a car was a risk factor. Conclusions: Social factors are associated with drunken driving. In general, people with a lower social position are more prone to drive after drinking. Social differences are visible already in youth, whereas working and own income of young persons signal different risk mechanisms for youth than for working-aged people. Measures for preventing drunken driving are needed within public health policies.

program aimed at increasing the SET supply. Finally, we describe innovative approaches used outside of education that can also be used to strengthen ST. --- SYSTEMS THINKING IN ACTION: ANALYSIS OF SPECIAL EDUCATION TEACHER EDUCATION PROGRAMS We used Stroh's four-stage ST process as the overarching framework for analyzing SET programs' role in the SET workforce crisis. When considering the content in Table 1 many of the past and present workforce solutions have been or are currently supported by the U.S. Department of Education Office of Special Education Programs 84.325 K and D funded projects as well as through IDEA flow through funds. The solutions to bridge the gap identified in Table 1 might be considered as a basis for future funding efforts to improve the availability of effective SETs. Moreover, stakeholders could add additional or alternative solutions to current practices or in place of existing solutions. Although these are clearly not exhaustive, the content included in Table 1 serves, in part, not only to synthesize and illustrate ST ideas, but also as the basis for stakeholder discussions about what needs to change and why. --- Systems Thinking in Action: Analysis of School Districts The approach and content delineated in Table 2 also emerged from Stroh's four-stage ST process for analyzing both districts' and schools' roles in the SET workforce crisis. Although we replicated the ST process, used in Table 1, the content included in Table 2 differs. Specifically, the content pertains to district and school related SET preparation, recruitment, and retention. Systems Thinking in Action: SET Programs and School Districts Using known system issues to analyze two parts separately reflects important aspects of ST. When stakeholders carry out Stroh's four-stage process in isolation, the results typically reflect short term solutions. Although short term solutions may be vital in responding quickly to a crisis, they often backfire over time . By contrast, longer term solutions are generated when diverse system stakeholders convene and intentionally use the ST processes and tools to identify root causes, assume shared responsibility, commit to change, and carry out modifications or interventions. When applying a longer term, ST approach , SET educators and district personnel convene with other key stakeholders. Together, these diverse stakeholders use ST processes, such as Stroh's action steps and stages, to gain new insights into the crisis achieving longer, rather than shorter term solutions. --- SYSTEMS THINKING ACTION STEPS AND STAGES FOR GENERATING SOLUTIONS In this section, we describe ST action steps and stages stakeholders can use to analyze, innovate, and improve results based on their unique SET workforce needs. These action steps and related stages have the potential to offer stakeholders not only greater understanding of the complexities in the SET workforce crisis but also how to intervene effectively. --- Action Step 1: Understand the "Big Picture" To understand the "big picture" , stakeholders should use current specific SET workforce data related to their program or district. For example, if SET faculty from several geographically connected universities are working to address the SET crisis, they should join with districts in their region to collect and analyze personnel data-allowing them to understand the nature of shortages in their area. As they examine data, stakeholders may find a surplus of certified SETs, who either separated prematurely from the workforce, or never entered it. Rather than solely recruit a new supply of SETs, these stakeholders should make efforts to understand this reserve pool and consider incentives to hire them for full or part-time work. --- Action Step 2: Increase Awareness of and Foster Shared Responsibility for the Crisis One of the tenants of ST centers on optimizing the relationships between the parts of the system . Neither SET university nor district personnel are solely responsible for the workforce crisis, so neither can solve the crisis alone. Through this partnership approach, diverse stakeholders can cooperate, rather than compete, to achieve better short and long-term results. Drawing on Action Step 1, diverse stakeholders can combine their recruitment efforts by jointly identifying and targeting workforce surplus supply. --- Action Step 3: Take a Deeper Dive to Influence the Whole System Although the first two steps matter, they are insufficient to change the entire system and yield better results . According to Stroh Consider international direct hires Develop agreements for free community college credits/degrees. --- Time for traditional SET preparation Offering Alternative Certification options Alternatively prepared teachers more likely to leave Provide more intensive induction and mentoring support for underqualified SETs --- Specific SET shortage areas Offering cohort programs to fill targeted areas Recruit paraprofessionals, substitute teachers, or high school students in grow your own program Insufficient numbers of individuals interested in becoming SETs Determine specific numbers of teachers needed to teach students in specific exceptionalities. Identify adults from foster care system as they have college support and understand diversity of issues . Consider online games for recruitment; used in STEM to recruit students --- Inadequate clinical experience Enhance clinical experiences by determining the scope, selecting priority activities, identifying products/outcomes, assessing outcomes, and providing ongoing feedback Limited access to clinical sites and/or inadequate supply of supervisors, mentors, coaches Use technology to increase supervision, mentoring, and coaching, during coursework and clinical experiences Limited fiscal resources at district, state, and/or national levels. Apply for grants to increase SET salary, signing bonuses, and/or offer additional compensation for other roles . Consider pay for teachers higher than administrative positions . Using artificial intelligence to automate some of the routine tasks to reduce the overall SET workload Low status Business as usual . Low SET status remains unchanged. Engaging in marketing through those in the profession who are viewed as "positive" ambassadors. Partnering with public television station to create a Teaching Network Channel . --- Inadequate preparation Fostering partnership programs between universities and schools Providing residency models Partnerships are often fraught with conflict Residency models may provide SET candidates with insufficient preparation Increasing technology enabled opportunities for practice-based SET professional development , Video Coaching [Coogle et al., 2017]) --- Poor working conditions Producing SET survival books and guides and "stress busting" strategies . May result in victim blaming and limit improvement in working conditions. Partnering with district and national teacher unions to advocate for improved conditions. Providing leadership development about supporting SETs and improving working conditions . Employing teams of professionals to create support networks . Using Glassdoor. com to improve working situation . --- Lack of supportive leadership preparation Providing principals/leaders with preparation about disability, special education, and supporting SETs. --- Lack of preparation in general and tends to focus on legal aspects of special education Facilitate collective responsibility for students with disabilities across the school . TABLE 2: Systems Thinking Analysis for School District Personnel diverse stakeholders take deeper dives into the system, they do so to identify and understand the parts of the system, the connections between the parts, how the system has functioned and is currently functioning, allowing them to identify modifications that might yield better results. Thus, diverse stakeholders ST efforts can be guided by using Stroh's four-stage framework. --- Stage 1 Building the foundation for change begins when diverse stakeholders convene and acknowledge the SET workforce related issues each faces as well as what they want to change. However, cultivating collective readiness for change involves preparing stakeholders to use ST processes while engaging in difficult conversations. --- Stage 2 Facing existing realities requires understanding and acceptance of the problem . For example, SETs and district personnel may recognize that under-preparing SETs is a problem that contributes to their departure from the workforce and adversely impacts educational outcomes for students with disabilities. This insight may lead the stakeholders to realize that when attempting to recruit individuals from the SET reserve pool, they need to consider how to address this underlying issue . Also, the stakeholders might need to acknowledge they harbor different views about what SET knowledge, skills, and dispositions SETs need. This understanding and acceptance leads stakeholders to Stage 3. --- Stage 3 Committing to change involves making an explicit choice . After SET faculty and district personnel have established readiness and identified the current realities, they continue moving forward by committing to change. At this stage, the realization of what needs to change to achieve key outcomes occurs when stakeholders acknowledge the costs of the status quo, the costs and benefits associated with changing and not changing, and the solutions and trade-offs needed for both. This stage is often considered a crucial turning point. For example, when SET faculty and district personnel realize their separate attempts, producing short-term results only allow them to cope with the SET workforce crisis, rather than ending it; they may be more likely to commit to a collective approach to change. --- Stage 4 Bridging the gap between the undesired and desired outcome takes place when diverse stakeholders move from understanding and affirming to acting . When applied to the SET workforce crisis, stakeholders not only engage in joint recruitment efforts, which target individuals in the workforce pool, but also address the problem of under-preparation through collaborative approaches that offer low-cost certificate/licensure options, employment incentives , improved working conditions, innovative approaches , and opportunities for SET leadership. Stakeholders also engage in continuous ST improvement by jointly monitoring and adjusting their approaches regularly . --- SPECIAL EDUCATION TEACHER PREPARATION: SYSTEMS THINKING IN THE REAL WORLD In this example, we describe how faculty at the University of Central Florida relaunched a previously discontinued SET preparation program using the four-stage ST process. Stage 1. Establishing a readiness for change was a foregone conclusion this university faced with a decision of whether to completely discontinue its undergraduate teacher preparation program in 2016, or to focus on resurrecting it. Under-enrollment and lack of faculty led the program faculty to temporarily suspend the program years earlier, and by 2016 the final student graduated. Local school districts were in a crisis with unfilled special education positions, so in response to local need the faculty members committed to focusing on relaunching the program in 2017, with a change-model approach in mind in partnership with several Central Florida local school districts. Using a popular change model from the field of business, the program coordinator and doctoral students used principles from Kotter's 8 step change model as the framework . The first step of the model, establishing a sense of urgency, is an obvious need in special education as SET positions are left unfilled by qualified teachers. Communicating this urgency to the College and university provided a way to promote change to existing systems as quickly as possible and cleared the path to building a more accessible and attractive undergraduate program which was key to this successful relaunch. Stage 2. The "existing realities" to navigate in attempting to implement change began with a close examination of how and why the program was structured in its original form, state requirements impacting program design, and existing college and program area policies that hindered recruitment and retention of students. Some realities hindering recruitment were quickly identified: course scheduling hindered working in schools and taking classes, internship requirements created an economic inequity as some students could not afford to quit working to fulfill the 40-hour per week requirements, program admission requirements created a bottleneck and frustration for students, and an increasingly online experience threatened the development of collaboration skills critical for teachers. Given that new SETs with substantial field experiences are significantly more likely to stay , enhancing these experiences became a cornerstone of the program relaunch. A teaching residency was created with partnering school districts to address the issue of teachers leaving the field due to feeling underprepared for the challenges they experience in the classroom . The program allowed students to find positions as assistants in special education classrooms to complete two semesters of internship. Rather than creating a "paid internship", the model allows for internships to be layered over the job. Students fulfilled the job requirements of the school's position and completed coursework online or in seminars after work hours. Doctoral scholars were prepared and used a coaching model to support undergraduate students during these internships through weekly online discussions and goal-setting sessions. Changing the practice of clinical experiences and determining modifications that might yield better results was no small task. Gaining "buy-in" from colleagues willing to consider new paths forward was critical to the momentum for change. Stage 3. Committing to change involved not only commitment from university stakeholders but also community partners. Understanding that School-University partnerships allowed districts to play a direct and productive role in preparing their teachers while allowing them to fill vacancies with teachers who were better prepared, more diverse, and more likely to stay ; thus, the program was committed to strengthening these partnerships. In one large district, for example, a long-standing MOU was changed to include language supporting the completion of university internships while on-the-job as paraprofessionals and teaching assistants. Other local districts followed suit and examined their MOUs with the university to find places to support students interested in the profession. Another area of examination and change included program admission requirements. At the time the program was relaunched, test requirements were a barrier for students to enter the major. A system was put in place to allow provisional admission and support for test preparation so students could begin coursework rather than facing unnecessary delays. While investigating the need for program changes it was also determined that not all students interested in working with students with disabilities wanted to work in traditional classrooms. A separate track, a partnership with communication disorders, was created for students to work in other settings . Stage 4. Bridging the gap between the undesired and desired outcome is occurring at the time of this article is being written. The number of teachers in the program and entering the districts continues to grow but an unintended outcome is that more students are enrolling in the special education major, but not seeking to complete the teaching licensure requirements. Thus, some of the ST that needs to continue lies beyond the teacher preparation program and even the districts involved. The next level of ST that needs to occur involves a need for national, state, and local messaging about the state of teaching and the work conditions for the SET workforce to further impact both the undesired and desired outcomes of this project. --- EXPLORING INNOVATIVE WORKFORCE SOLUTIONS TO STRENGTHEN SYSTEMS THINKING In this section, we provide short summaries of ideas from other fields to offer additional examples of solution-centered, innovative workforce approaches aimed at reducing shortages. The ideas are presented with notations of how they might be employed or have been employed in universities and/or districts. Although these ideas are not yet research-based approaches to SET workforce recruitment, preparation, and retention, they are worth considering and evaluating throughout Stroh's recommended four-stage ST process. --- Supporting Mental Health Companies such as, LinkedIn, Starbucks, Bumble, and Mozilla provide employees with mental health days to focus on their well-being. The purpose of mental health days is to support employee's productivity and retention by encouraging self-care. LinkedIn found success in providing all employees with one paid week off to enhance mental health and to cope with burnout. Fidelity Investments took a different, preventative approach and extended the time off for holidays by three days. SET preparation program faculty and school district personnel could consider similar health and wellness approaches by offering mental health days that support workforce preparation and retention. Nurses deal with high levels of loss of life in their work, thus "Death Cafés" have been used as a form of debriefing . These cafes used internationally, guide informal discussion on topics of death, loss, secondary trauma, and illness. Healthcare workers, particularly within the ICU reflect on distressing events and develop a sense of community and support among coworkers to prevent burnout. Similar types of stress cafés could be created to help preand in-service SETs talk virtually about challenges with others . In a systematic review, Tolksdorf et al. found combat fatigue in Intensive Care Units was reduced when employees' work settings promoted higher levels of autonomy, decreased job overload, ensured employee safety, reduced exposure to violence, and decreased working hours. SET preparation program faculty and school district personnel could consider similar approaches for reducing SET fatigue. Some have attempted to do so by providing longer breaks, mindfulness kits, emotional support and breaks after a crisis, or by offering incentives, such as onsite daycare, free car washes, massages, therapy dogs, or even pet daycare. However, little is known about the extent to which these approaches are used and whether they have been studied systematically. We suggest funding to consider interventions that improve working conditions and supports to address SET teacher preparation, retention, and recruitment. Like the "What Works Clearing House," a national database could support the development of a knowledge base and the identification of approaches that could be used within an ST approach. --- Matching Needs to Shortages The vocational rehabilitation system also faced shortages of qualified rehabilitation professionals . To address this need, one university implemented a five-year training program with the goal of increasing the skills of VR counselors to effectively meet the needs of persons with disabilities. Unique features of this program included customized employment strategies such as personalizing the employment relationship between job seekers and employers by matching interests or talents. Additionally, the university offered a scholarship opportunity with a service payback requirement which received a high level of successful placement within the field. Smith and colleagues found that financial incentives, mentorship, networking, and professional learning opportunities paired with careful selection of scholars whose career interests matched the intent of the program led to an increase in the number of students pursuing a master's degree in vocational rehabilitation. This same type of model often is aligned with Office of Special Education Programs 84.325K grants, but how this might be sustained in partnership with district, state, and federal resources is a pathway for SET educators to consider. --- Global Application Outside of education, countries worldwide are taking novel steps to address worker shortages. In Germany, companies facing labor shortages tend to respond with more training for lowskilled workers . The practice of using 'voice', or incorporating employee training interests or preferences, was found to increase participation in these trainings particularly when organizations have formalized HR practices and structures supporting employee representation . Meanwhile, in the face of IT shortages, cyber security, and other technology-related fields, the European Union recommends enterprises to ensure their current technology professionals remain up to date on skills and acquire proper or new certifications to meet the demands of the evolving field . Additionally, the EU recommended embedding industry expertise in courses and having businesses offer certifications or collaborate with others on the development of courses or certifications . How might a similar approach in teacher preparation, through associations like the Council for Exceptional Children, with the Teacher Education Division combined with the Division of International Special Education , be used to strengthen ST and address SET recruitment, preparation, and retention shortages globally? --- SPOTLIGHT ON SYSTEMS THINKING FOR BOLSTERING RECRUITMENT AND ELEVATING PROFESSIONAL STATUS SET faculty and school district personnel also may use the ideas offered below as a basis for how ST might be employed to bolster recruitment and elevate professional status. These ideas are intended to be generative and are worth considering and evaluating throughout Stroh's recommended four-stage ST process. • Realign and clarify workforce, including rehiring, retooling, recycling, and continued use of those who could or will retire. Finding short-term ways to keep retirees as reading or mathematics coaches or as first year mentors for even one day a week was a successful approach by one large urban district. • Offer scholarships, in addition to or instead of TEACH grants, to recruit for high-need schools. Universities have coordinated scholarships across organizations into a single database to recruit teachers at the university aligned with getting the district leaders to provide "paid" student teaching internships while others have harnessed foundation and Title 1 funding to provide richer financial support for teachers . • Employ teams of professionals to create support networks . One university hired clusters of faculty members to address targeted areas of needs instead of the traditional approach of hiring in a department one at a time. • Identify "positive" ambassadors to shift workforce recruitment and retention. One district had celebrities talk about their favorite teachers while another had weekly promotions from diverse teachers sharing positive experiences. The current narrative in SET cannot change without directing a new narrative. • Increase economic support. Some districts are offering signing bonuses for schools with extreme and persistent shortages. A master's cohort in these same sites are using Title 1, scholarship, and endowment funds to ensure SETs move up the pay scale, with efforts to create a strong cohort of leaders in these schools. From the over 100 teacher leaders funded to date, over 75 remained in the same schools and placements 5 years later . Offer apprenticeships and ensure the new employees have the most enticing jobs , or encourage paid internships. In the previous real world example illustrating Stroh's recommended four stage ST approach in SET preparation, the ST team led by Hines and colleagues at the large urban university increased undergraduate enrollment from 0 to 100 in a year. --- CONCLUSIONS Special education teacher shortages continue to be chronic and pervasive, interfering with the provision of a free and appropriate education to students with disabilities . We realize fully this longstanding problem will not be remedied quickly. However, we believe the collective power of key stakeholders in special education, policy, leadership, and practice can come together in unprecedented ways to no longer talk about shortages but to turn work towards producing timely, innova-tive workforce research and solutions. Toward this end, we encourage stakeholders to explore what a ST framework offers and how it might inform a new research agenda centered on interventions to improve teacher recruitment, preparation, retention, and effectiveness. The special education workforce and the students with disabilities and families they serve deserve no less. --- ABOUT THE AUTHORS Marcia L. Rock, Ph.D.

Chronic and pervasive special education teacher (SET) shortages have interfered with state, district, and school efforts to recruit and retain effective teachers for students with disabilities. Unfortunately, these shortages have worsened post-pandemic due to early retirements, low unemployment rates, and career changes. The purpose of this article is to provide a systems thinking (ST) framework to help stakeholders consider the complex and interacting systems in which these shortages occur (i.e., teacher preparation, district and schools, society). We consider specific elements within these systems, their interconnections, with a focus on identifying steps and ideas stakeholders can use to understand contributors to the shortage crisis, while providing strategies and innovative ideas for greater sustainability. We also offer real examples of ST solutions used within teacher education programs, schools, and other professions. To further bolster ST, we conclude with examples of innovations outside of education with ideas to bridge these concepts into potential pathways to address SET shortages.

Introduction Funding targeted for HIV care programs in sub-Saharan Africa has produced tremendous results over the past several years, most notably the delivery of antiretroviral therapy to almost 4 million people in sub-Saharan Africa by 2009 [1,2]. In Kenya, the number of people receiving antiretroviral therapy has increased from about 11,000 in 2003 to more than 138,000 patients in 2007 largely as a result of receiving the President's Emergency Plan for AIDS Relief funds [3,4]. This kind of directed "vertical" funding has allowed for specialized staff training, more rapid and efficient program implementation, and betterequipped facilities-including free laboratory services and medications-as HIV programs have been scaled-up [5]. These results may not have been possible in such a short time using an integrated approach to health care delivery. Nevertheless, the HIV epidemic has increased pressure on the Kenyan health care system as a whole by increasing the workload for health personnel-whose numbers have not increased proportionally to the demand [6]-and straining infrastructure capacity and public expenditure. Vertical HIV programs may exacerbate the situation further, as the concentration of resources leads to HIV clinics being better equipped and their workers better compensated with financial and nonmonetary incentives [7,8]. The resulting attrition of personnel from general health services may weaken important primary health care services [8,9]. Integration of HIV services into primary care can be defined in multiple ways. In the context of this paper, we define it as colocation and sharing of services and resources for HIV care and primary care, such as clinic space, clinicians, health education, pharmacy, laboratory services, and training. Integration of HIV services into primary care addresses the issue of skewed resource allocation, allowing people to access the health care they require regardless of HIV status. Available evidence suggests that integration offers several potential advantages at this point in the response to HIV in sub-Saharan Africa [10,11]. Colocation of HIV and primary care services maximizes the use of available health facility structures and ensures that funds targeted for rehabilitation or construction of HIV care facilities will also benefit primary care [7]. Decentralization of HIV care services into primary health care clinics results in greater acceptability of services, increased referrals and enrolments into HIV care, and improved patient retention compared to services at specialized hospitals [12][13][14][15]. Patient outcomes may also be better, and costs lower, at primary health clinics [16,17]. Support systems such as monitoring and evaluation , supply-chain management, laboratory networking, and counseling services can also be shared when HIV services are integrated into primary care services. Pfeiffer et al. reported that integration results in the inclusion of antiretroviral therapy related data into national health databases [7]. Such joint information systems also increase efficiency [18]. In addition, integration allows staff to share the workload for HIV-positive and HIV-negative patients, resulting in more efficient use of resources and reduced patient waiting time [18]. A program of decentralization and integration of HIV care services into the primary care system in Mozambique reported improvements in access to care, quality of care, and efficiency in service delivery [7]. Integration may increase the positive effect of resources allocated to HIV: a study in Rwanda provides observational evidence that introduction of HIV care services results in improved staff capacity at primary health care facilities due to the additional in-service training given to health care workers [19]. Increases in the uptake of other services, especially antenatal care, were also associated with integration, suggesting that a mutually beneficial interaction can be created between HIV care and other primary care services. Despite the evidence suggesting system-wide and national-level benefits to integration, the patient-and service-level effects are unclear. Whereas the overall benefits of integration are likely to trickle down to patients and providers, it is also possible that the resulting reorganization of health care delivery may disrupt service provision and potentially cause dissatisfaction among patients, particularly in the short term. Further, integration of specialized services into primary care services may not always result in better patient-and service-level outcomes [20]. For example, integration of HIV services with sexual and reproductive health services may be hindered by increased patient burden, inadequate staffing, and resistance from existing health care workers [21]. Similarly, integrating services for sexually transmitted infections into routine health services may result in lower utilization and reduced patient satisfaction [22]. Also, integrating services for family planning and maternal and child health care into routine primary care may reduce knowledge of family planning [23]. There is minimal evidence to suggest that stigma associated with HIV care may reduce with integration [24]. Stigma may be reduced if a patient's HIV status cannot be determined by the general public simply by observing the physical location where a patient is receiving care. However, integration of services may lead to other opportunities for accidental disclosure of one's HIV status, so the effect of integration on stigma is unclear. A study assessing HIV/AIDS stigma on health service utilization demonstrated that anticipated HIV stigma can be a barrier to HIV testing acceptance among pregnant women at clinics that were part of a clusterrandomized trial evaluating the effect of integration of HIV and antenatal care services [25]. In 2008, a system that integrated HIV services into primary care was piloted in rural health facilities in Nyanza Province in western Kenya. We assessed the patient-level effect of this integration with a focus on patient satisfaction and perceived stigma. --- Methods From December 2008 to January 2010, we conducted an evaluation of the pilot integration project at three health facilities to assess the patient-level effect of integration. Family AIDS care and education services implemented the integration in collaboration with the Suba District Health Management Team of the Kenya Ministry of Health. FACES is a collaboration between the Kenya Medical Research Institute and the University of California, San Francisco , funded through the US Centers for Disease Control /PEPFAR [26]. --- Study Site. The pilot was conducted in Suba District in Western Kenya, which has a population of 214,463 [27] spread over the mainland and 10 inhabited islands. The 26.3% HIV prevalence in Suba is the highest in Kenya, which has a national average of 7.1% [3,28]. At the time of the pilot, there were 40 health facilities of varying capacity staffed by about 200 healthcare workers, 144 of whom were nurses. In contrast, the Suba District Annual Operations Plan 2008-2009 recommends 934 healthcare workers, including 591 nurses [6]. Of the 40 facilities, 20 provided HIV care and treatment services to a total of approximately 15,000 patients. Three health facilities underwent integration and were evaluated during the pilot period: one subdistrict hospital and two health centers, representing the two main levels of health care in the district. Although the subdistrict hospital and health centers chosen for this evaluation represent different levels of health --- Features of Integration. We addressed the different aspects of integration as shown in Table 1. The implementation of these activities also constitutes our definition of integration of HIV care with primary health care services. Further features of the integrated health facility are described below. --- Clinic Space and Clinicians. Before integration, HIVpositive patients were seen separately on one to four dedicated days each week. After integration, they were seen on any day of the week in the general outpatient clinic; all patients followed a uniform patient flow, and patient load was evenly spread out among the existing staff. --- Health Education. Integrated health education was conducted for all patients regardless of HIV status in a single shared waiting bay. This approach was seen as a way of reducing stigma against HIV and broadening health education reach. --- Pharmacy/Supplies. With integration, all drugs were stored in the same storage space and dispensed from the same pharmacy; supply and inventory management at the clinics under evaluation were also integrated. The integrated pharmacy became the central repository and dispensing area for therapeutic food supplements that would otherwise have to be stored in a separate food store. Orders for all drugs and food supplements were made within the same supply-chain network. Training and mentorship for pharmacy and supplies management were offered to both Ministry of Health and FACES staff who then worked together in the integrated pharmacy and were able to share the daily workload. --- Laboratory. Introduction of HIV services brought with it a well-equipped central laboratory at the district hospital; after integration, these lab facilities were used to conduct other tests not necessarily related to HIV. For example, a new automated biochemistry machine to be used for measuring liver enzymes to monitor the progress of patients on ART was also used for other patients when clinically indicated. Supportive supervision for the district lab network prior to integration was infrequently conducted by the MOH District Medical Laboratory Technician; however, after integration it was conducted by a combined FACES and MOH team. Laboratory supervision was more frequent and regular after integration because FACES had a more reliable transport system. --- Clinical --- Study Participants. Study eligibility at baseline included adult patients ≥18-year old who were HIV-positive and already enrolled in HIV care , attending a return visit to study facilities in rural Suba District. Exclusion criteria included patients <18-year old, patients not enrolled in HIV care, and patients enrolling in care that day. Inclusion and exclusion criteria remained the same at the followup with one exception: individuals enrolled in HIV care as well as those accessing general outpatient services were included, reflecting the integration of these services. Patient participation involved a self-administered questionnaire on service satisfaction and perceived stigma. The questionnaire was available in the language of the patient's choice , and all data collection was anonymous. A sample size of 10% of the patients enrolled into HIV care at each health facility was targeted: 94 of the 941 patients actively enrolled at baseline, 107 of the 1,065 patients actively enrolled at three-month follow up, and 142 of the 1,421 patients actively enrolled at 12-month followup, for an overall sample target of approximately 343 patients. A consecutive convenience sampling approach was used. Trained Monitoring and Evaluation staff employed by FACES approached each departing patient throughout the day, determined eligibility verbally, obtained verbal consent, and carried out the assessment. Patients were advised that they were free to participate or not without affecting their services, and no incentives were provided. Willing participants completed the self-administered questionnaires in private and upon completion placed the questionnaires in a secure box to maintain privacy. The assessment was conducted over a 3-day period at baseline and followup . --- Measures 2.4.1. Patient Satisfaction. Patient health facility satisfaction was assessed using a questionnaire that included questions adapted from the AIDS Clinical Research Trials Group tool. The questionnaire is a standard FACES program tool that was initially adapted, piloted, and validated for the local population at FACES-supported health facilities in Kisumu District and then used for this evaluation in Suba District . Items included satisfaction with the office , with clinicians , and with each of seven departments: reception, community health assistants , health talks, nursing/counseling, laboratory, clinical, and pharmacy. Individual items addressed wait time, perceived usefulness of HIV education, and overall satisfaction with the clinic. Each item included three or four response options, such as "Very satisfied", "Satisfied, " and "Not satisfied;" or "All the time, " "A lot of the time, " "Some of the time, " and "None of the time". The two measures of office satisfaction and the three measures of clinician performance were each combined into linear scales for analysis. All other satisfaction measures were coded as binary responses with only the highest level of satisfaction coded as one. --- Perceived Stigma. Patient perception of stigma at the health facility was assessed using three items adapted from standardized items validated for use in this context [29]. Items covered privacy and confidentiality, equitable treatment for HIV-positive people, and discomfort receiving care at the health facility due to the possibility of other people finding out one's HIV status. Responses were grouped into agreement and disagreement for analysis. --- Data Management and Analysis. The data were entered and cleaned using SPSS Software and were merged across clinic sites and time before being analyzed using Stata 11.0 . Change from baseline was assessed using truncated linear regression and logistic regression with robust standard errors. All models were assessed for effect modification by sex and analyzed within sex if the interaction term with either time point was significant at 𝑃 < 0.20. Survey language, sex, and being on ART were included as confounders in the final models based on showing an association with the outcome in unadjusted analysis . --- Ethical Review. Ethical review committee approval was obtained locally and internationally; the protocol was reviewed for human subject concerns and approved by the KEMRI Ethical Review Committee and UCSF Committee on Human Research. --- Results Of the targeted sample of 343 surveys, 295 surveys were collected cumulatively over the three time points: 58 participants at baseline , 104 participants at three months , and 133 participants at 12 months . Table 2 shows patient demographics; of the 256 respondents whose sex is known, 94 were men and 162 were women. At baseline 28 were on ART, with this proportion increasing to 61 individuals at three months and 78 at 12-month followup. These consistently high percentages demonstrate that a large proportion of respondents were in HIV care even when the survey was administered to an integrated patient population. --- Patient Satisfaction. Patient satisfaction with the clinic office and with each department remained high throughout the assessment period . By 12 months after integration, respondents had higher odds of being very satisfied with reception services , HIV education , and wait time . Women rated clinician performance more favorably at both the 3month and 12-month assessment compared to baseline . Respondent ratings of overall satisfaction declined significantly by the 3-month assessment , although by the end of the assessment, this association was no longer present. --- Perceived Stigma. All stigma-related items showed significant effect modification by sex . At 12 months post-integration, men reported significantly higher levels of agreement that people with HIV were treated the same as others , while women's responses did not change significantly. Patient agreement with providers maintaining privacy and confidentiality did not change significantly within either sex during the assessment period. Men were less likely to agree that they were not comfortable receiving care at the clinic by the 3-month survey . Women were more likely to express discomfort at both time points compared to baseline: 3-month aOR 2.65, 95% CI 1.01-6.99; 12-month aOR 3.37, 95% 1.33-8.52. --- Discussion Our findings suggest that integrating HIV care with primary care services does not negatively affect individual patients and may offer some benefits that extend beyond the healthsystem level to the individual patient. Assessment of the patient-level effect suggests that patient satisfaction remained high and that integration did not heighten perceived stigma. Before and after integration, patients generally agreed that care was provided confidentially and equitably regardless of HIV status although women expressed increased discomfort with receiving care at integrated clinics. In rural Kenya, other investigators from our group have found that overall client satisfaction with integrated HIV services among pregnant women is associated with satisfaction with administrative staff, satisfaction with health professionals, and convenient wait times and encounters with a receptionist [30]. As an initial assessment of the patient-level effect of integration, these findings may be a promising indication that integration may extend the resources concentrated within HIV care to a broader patient population without diminishing patients' perceived satisfaction. Patient education sessions appeared to have a greater effect when done in an integrated setting. The significant increase in satisfaction with HIV education as well as reception and wait times may be an indication that the increased staff training positively affected patient-provider interaction. The increased satisfaction with HIV education may have also resulted because the mixed population of HIV-positive and general patients found the sessions more beneficial than HIV-positive patients alone, or because of the broader scope of topics after integration. While we found that patients trusted healthcare providers to keep information confidential and to treat all patients equally to similar degrees before and after integration; women were less comfortable about receiving care at the integrated clinic. Since the follow-up samples included HIV-negative patients, it is possible that this finding relates to their discomfort with integrated care or that it represents continued high levels of apprehension regarding others suspecting one is HIV positive. The effect of integrated care on inadvertent disclosure bears further investigation. Overall, our findings suggest that despite the challenges involved in integrating HIV care into routine health care, it is possible to pursue integration without significant disruption of patients' experience, and in fact with some benefits observable even within a 12-month period. Our finding that patient satisfaction remained high with integration may be evidence that the restructuring required to achieve the system-wide benefits of integration [10,11] does not result in patient dissatisfaction. Our findings that stigma did not significantly worsen with integration support some of the findings by Topp et al. [24]. --- Limitations. This study provides important integration findings. However, there are limitations to consider. The first limitation was the absence of a control group. It is possible that the changes observed could have been due to factors beyond the intervention. Second, the health facilities and the small number of patients interviewed within them may not be representative of the target population, limiting the generalizability of the findings. However, the proportion of respondents recruited out of the targeted number of participants increased substantially from 62% at baseline to greater than 90% at the followup. Third, the patient population changed between baseline and followup due to the nature of the intervention. In particular, the inability to distinguish between patients receiving HIV care and general care patients in the follow-up surveys limits the ability to determine how these two different patient groups responded to the intervention. Although we controlled for ARV status in multivariate analysis, we are unable to fully identify the impact of the intervention on patients receiving HIV care versus general patients. However, the consistently high percentages of patients on ART were an indication that a large proportion of respondents were in HIV care even when the survey was administered to an integrated patient population. By interviewing only patients who attended clinic and only tracking the number of willing participants, we were unable to capture the experiences of those who may have avoided clinic due to integration or who may have refused to participate, potentially introducing bias. Individual confounding factors may also influence the findings. Fourth, the questionnaire was self-administered; therefore, only literate respondents completed the questionnaire. In addition, the sampling strategy was not systematic and was therefore subject to selection bias from the research team. Other biases inherent in our evaluation design included social desirability bias and courtesy bias. Finally, the study may have benefitted from qualitative interviews to provide more insight into the results, especially in relation to improvements in service delivery. However, this evaluation was done in the context of a mature HIV program in a real-world setting. We demonstrate the utility of operational research within HIV programs to improve program quality. --- Conclusions This study is a step towards developing a model of care that integrates HIV care into primary care in resource-limited settings. Given the potential positive aspects, it is worth exploring integration as one innovative way of improving primary care services that receive little donor funding, while at the same time maintaining the achievements of donor-supported HIV care at the patient level. Larger, cluster-randomized or stepped-wedge and longitudinal studies should be conducted to confirm these findings and address other critical issues, including the effect of integration on quality of care, longterm health outcomes, and cost-effectiveness. --- AIDS Research and Treatment --- Conflict of Interests The authors declare that they have no conflict of interests. --- Authors' Contributions Thomas A. Odeny designed and conducted the study, interpreted the data, and drafted the paper. Jeremy Penner conceptualized and designed the study, provided mentorship during study conduct, and made substantial contributions to the interpretation of the results and drafting of the paper.

HIV departments within Kenyan health facilities are usually better staffed and equipped than departments offering non-HIV services. Integration of HIV services into primary care may address this issue of skewed resource allocation. Between 2008 and 2010, we piloted a system of integrating HIV services into primary care in rural Kenya. Before integration, we conducted a survey among returning adults ≥18-year old attending the HIV clinic. We then integrated HIV and primary care services. Three and twelve months after integration, we administered the same questionnaires to a sample of returning adults attending the integrated clinic. Changes in patient responses were assessed using truncated linear regression and logistic regression. At 12 months after integration, respondents were more likely to be satisfied with reception services (adjusted odds ratio, aOR 2.71, 95% CI 1.32-5.56), HIV education (aOR 3.28, 95% CI 1.92-6.83), and wait time (aOR 1.97 95% CI 1.03-3.76). Men's comfort with receiving care at an integrated clinic did not change (aOR = 0.46 95% CI 0.06-3.86). Women were more likely to express discomfort after integration (aOR 3.37 95% CI 1.33-8.52). Integration of HIV services into primary care services was associated with significant increases in patient satisfaction in certain domains, with no negative effect on satisfaction.

Introduction Currently, when inclusive education is widespread in most educational systems, including the Czech Republic, research on the attitudes of adolescents toward disabled individuals, emphasizing key findings, trends, and gaps in the knowledge, is a significant contribution to theory and practice. Adolescent attitudes have their own developmental dynamics, just as the development of the reality of inclusive education shows elements of development over time [1]. There is a need to describe differences in adolescent attitudes toward disability, providing insights into factors that influence these attitudes, such as cultural values, education, and personal experiences. The impact of media portrayals on perceptions of disability and adolescent attitudes toward disabled individuals is also significant, with a need for discussion of both positive and negative portrayals and their potential consequences [2]. As far as the school environment is concerned, it is especially expected here to work with attitudes toward persons with disabilities and their education [3,4]. The results of research in this area should be used for policies and initiatives aimed at promoting equality, accessibility, and inclusion for people with disabilities to ensure their rights and quality of life [5,6]. Studying adolescent attitudes toward disability can provide valuable data for predicting future societal trends, informing strategic planning, and resource allocation in areas such as education, healthcare, employment, and available infrastructure. There is a gap in society's perception and willingness to accept persons with disabilities in the school environment because of low levels of awareness and a lack of informed experience [7,8]. School inclusion has a very wide range of definitions, reflecting national perspectives on this way of education. In general, inclusive education can be characterized as the best way to respect the diversity of the student population with opportunities for modification of the material and social environment [9][10][11]. Inclusive education can be traced back to basic principles that are very similar to those in other countries [12]. It is about striving and creating conditions for the involvement of all pupils and students in the educational process. In an inclusive learning environment, all pupils, regardless of health, social, cultural, or other disadvantages, learn together and teachers compensate for these disadvantages and special educational needs and use support measures. This includes modification of teaching methods, support by a teaching assistant, or, in some cases, support provided by a personal assistant. Inclusive education aims to educate all pupils and students together, to promote and put into practice tolerance of difference, and to maximize collaboration between pupils [13]. In addition to inclusive education options, there is also the possibility of educating children with special educational needs in special classes and schools. With this in mind, this study aims to explore the attitudes of non-disabled pupils toward co-education with pupils with disabilities. The school environment, especially the social climate in the classroom, is a significant factor influencing the academic achievement of all students [14,15]. Children's relationships in the classroom influence their motivation to learn, but also affect the teacher's interest in the work and motivation to perform the teacher's work [16,17]. In the Czech literature, there are no articles that deal with the relationship of nondisabled pupils to persons with disabilities in relation to inclusive education. The target topics are teacher support, support for pupils in inclusive environments, and teachers' attitudes toward inclusive education. For this reason, a research project was designed to provide the first information on adolescents' attitudes toward individuals with disabilities and basic knowledge of them. The characterization of attitudes in this age group is very important for the creation of the social climate in the school classroom and the design of appropriate interventions led by a special pedagogue or psychologist. The aim was to find out the current attitudes of pupils without disabilities toward classmates with any disability. It was necessary to find out the situation and to create a set of information for teachers of the previous educational stage to try to include topics and activities that would have a positive influence on the attitudes of pupils without disabilities toward classmates with disabilities. Children cannot be expected to acquire this information and competence in their own families. --- Inclusive Education in the Czech Republic Inclusive education in the Czech Republic is based on the national and international legislative framework. At the national level, the Education Act [18] defines the types and level of support provided for children, pupils, and students with special educational needs in Section 16, "Education of children, pupils, and students with special educational needs". The support measures have five levels and are applied to education in an inclusive framework at the primary and secondary levels, depending on the level of the pupil's needs. Similarly, support is provided to students who need it at the university level. The current practice of education of children and pupils with special educational needs is regulated by Decree No. 27/2016 Coll. [19], on the education of pupils with special educational needs and gifted pupils, and its latest amendment . These documents detail all levels of support measures in terms of the content and form of support and also list the purpose of compensatory aids, special textbooks, and special teaching aids, their division into levels according to their use by the pupil, and the rules for their use by the school, including their standardized financial requirements. The support offered to inclusive education-to pupils and teachers-is characterized by Decree No. 72/2005 Coll. [20], on the provision of counseling services in schools and school counseling facilities, as amended . This describes the counseling facilities, institutions, and professions that are included in the counseling system used in inclusive education. These include diagnostic services, assistance in the construction of individual education plans, and advice for teachers concerning the necessary modifications to the educational process. This includes both methodological recommendations and counseling to make the necessary adjustments to the psychosocial climate in the classroom or school. In the Czech Republic, in the 2022/2023 school year, there were 3657 primary schools with pupils aged 6-15. In this period, there were 88,903 pupils with special educational needs. There were 84,344 pupils in inclusive education and 4559 pupils in special classes [21]. From the international perspective, the Convention on the Rights of the Child, which was negotiated on 20 November 1989 in New York [22], and entered into force under 104/1991 Coll., and the Convention on the Rights of Persons with Disabilities, negotiated on 13 December 2006 in New York [23], are very important. The Convention on the Rights of Persons with Disabilities, implemented in signatory countries since 2008, provides the basic framework for equality for persons with disabilities, guarantees the full enjoyment of all human rights, and promotes their active participation in society, which includes education. The significance of these two fundamental documents can be summarized as a basic position on the rights of persons with disabilities, which includes the support provided, the protection offered, and the regulation of living conditions leading to equality, and, as regards all human rights and fundamental freedoms of persons with disabilities, also the promotion of respect for their inherent dignity in the same way as for non-disabled persons. These documents reflect a qualitative shift in the world's response to disability in its understanding of disability and the possibilities for living life to the full. The principles of inclusive education and equality of access to education are set out in the Czech national legislation. Attention is paid to persons with any kind of disability or disadvantage, following the conditions set out in particular in the Convention on the Rights of Persons with Disabilities . The educational process is conceived as a dynamic interaction between the disabled pupil and the environment in a broader context that includes all participants and offers possibilities for modification so that a particular person with a disability can reach his or her full potential and fulfil the requirements of the educational program. Inclusive education is essentially built on four pillars: the pupil with special educational needs, the teacher, classmates, and parents. The fourth pillar, parents, is divided for some research purposes into two groups: parents of the pupil in inclusive education and parents of non-disabled pupils. --- Inclusive Education and Stigma Inclusive education is now an established educational format in many countries, reflecting the developmental dynamics of society. This is linked to both the successes of this approach and the barriers to it. The most common causes of difficulties are insufficiently prepared teachers, an insufficient level of special education support, the school climate with non-disabled pupils, and often parents of non-disabled pupils [24][25][26]. It should be noted that special education also features difficulties and barriers affecting success. These include teacher preparedness, stigmatization of children, or a lack of communication between the school and families [27,28]. The multilevel approach to stigma and public health, as described by Cook et al. [29], is a model that focuses on the complex and multidimensional nature of health-related stigma or health disparities. This approach recognizes that stigma can be triggered by a variety of factors such as cultural and structural issues, interpersonal interactions, and personal factors such as relationships and behaviors. The stigmatization of people with disabilities is not a new issue and needs to be addressed. Stigma can start in the school environment, which is extremely competitive in all areas of a child's life. This also includes the difference caused by the presence of a disability [30,31]. For education in inclusive settings, it is important to understand some stigma caused by disability/special educational needs as a significant factor influencing the access of individuals with disabilities to education in general and to their potential with an impact on aspirational levels. The biopsychosocial perspective is a holistic approach that considers the biological, psychological, and social factors influencing an individual's health and well-being. This perspective is particularly relevant in the field of education, where it can inform the understanding and support of students with functional disabilities. The International Classification of Functional Disabilities and Health provides a framework for assessing and classifying these disabilities. It recognizes that disability is not solely determined by physical impairments but also by the interaction between impairments and environmental factors. Referencing the biopsychosocial perspective of the ICF in education allows for a comprehensive understanding of functional disabilities. By considering biological, psychological, and social factors when supporting students with disabilities, educators can create inclusive learning environments that promote optimal development [32]. These resources may vary depending on individual characteristics and the context. Overall, a multilevel approach to disability offers a comprehensive and systemic view that allows for the development of strategies to prevent and reduce the negative impact on the child, which can be reflected throughout the educational complex. The theoretical level is fundamental in the stigma model because it provides a basis for understanding and defining stigma. This model assumes that the stigmatized person is perceived and evaluated on the basis of various factors that are located on three dimensions: the cognitive dimension-this dimension includes the beliefs and stereotypes that individuals have about the stigmatized person. For example, they may think that they are dangerous or incapable; the emotional dimension-this dimension refers to individuals' emotional reaction to the stigmatized person. For example, they may feel uncomfortable; the behavioral dimension-this dimension refers to the behaviors and skills needed for social interaction. These three dimensions are interrelated and interactions between them can lead to stigma becoming more deeply embedded in individuals and society as a whole. Understanding these dimensions can help to reduce stigma and improve the lives of children who are educated in inclusive settings. For the educational setting, the tools that can be used to reduce the stigma against children with disabilities in mainstream schools are essential. On the basis of knowledge of how stigma works, strategies can be developed to combat it. These strategies can include public awareness and education, and support for specific students who are stigmatized [33]. --- The Aim of the Study This study focuses on the third pillar in the above list-classmates who are aged 11-16. This is the period of puberty, which is often described as difficult for the individual experiencing it and those around them. For this study, the period of puberty is characterized as a stage that is significant for the formation of attitudes toward environmental phenomena and persons in the school environment [34]. This environment, in the form of inclusive education, provides the space for the emergence of such a specific attitude as the relationship with the student with disabilities. The school environment introduces a very important element to this process, which is competence [35,36]. The high level of competition in school is another element that enters into the social relationships between children and also into the construction of attitudes-children try to conform to group dynamics while being left to themselves in the competition for the best personal results [37]. The attitudes of adolescents can be assessed as complex, consisting of cognitive, emotional, and conative components. During the adolescent period, attitudes tend to be changeable, malleable, and influential [38,39]. Socially desirable attitudes depend on the objective conditions of an individual's upbringing and socialization process and the adolescent's subjective relationship to the phenomenon. The tendency to radicalism that appears in this period is caused by the desire to have one's own opinion and the tendency to reflect the group opinion of classmates, and this is framed by a lack of experience and a tendency to hypercriticism [40]. In this schema, one's own opinion and attitudes are formulated in the school environment. Thus, attitudes toward minorities or differences often become a very fragile area for the construction and implementation of relationships between non-disabled members of the school class and a child with disabilities [2, 41,42]. For this reason, the topic of relationships and the attitudes of non-disabled students toward a classmate with a disability was chosen for this study. When it comes to the relationship with a classmate with a disability, the determining factors are relationships and the environment [43]. Intervention programs can be used to educate and shape attitudes toward people with disabilities and seek to achieve attitudinal change based on information that reflects the reality of the school environment in which the disabled pupil is to be found [44]. Attitudes toward persons with disabilities are based on respect for differences, tolerance, and the ability to accept a classmate with a disability. Educational intervention should include encouragement to recognize and understand difference, education for empathy, and understanding of the specific needs that a disability brings to a student. In this way, the three basic areas on which the coexistence of the two "worlds" is based should be addressed. These are knowledge, skills, and emotional adjustment [45]. The research was focused on the age category of pupils who are in different grades in the Czech education system. The aim of the research was to obtain basic information about the attitudes of these pupils toward pupils with disabilities. On the basis of the research objective, two main research questions and four "subquestions" were identified: I. What are adolescents' attitudes toward individuals with disabilities? II. What knowledge do adolescents have about individuals with disabilities? --- 1. Does the respondent's gender affect their attitude toward individuals with disabilities? 2. Does the age of the respondent affect their attitude toward individuals with disabilities? 3. Does a previous encounter with an individual with a disability affect their attitudes toward individuals with disabilities? 4. Does adolescents' knowledge about individuals with disabilities relate to their attitudes toward individuals with disabilities? The following hypotheses were then established for the research questions: H1. There is no statistically significant relationship between the respondent's gender and their attitudes toward individuals with disabilities. --- H2. There is no statistically significant relationship between the age of the respondent and their attitudes toward individuals with disabilities. --- H3. There is no statistically significant relationship between a previous encounter with an individual with a disability and attitudes toward individuals with disabilities: Presence of the individual in the family A friend with a disability Any encounter H4. There is no statistically significant relationship between knowledge about individuals with disabilities and attitudes toward individuals with disabilities. The hypotheses were constructed to describe the existing relationship between the respondent's attitude and his/her age and gender. In planning the research, the results in these areas were important for recommending targeted interventions. The items concerning previous experience and the level of awareness were of similar importance. These findings should also influence the form of the planned intervention. Teachers of pupils aged 6-11 in the first stage of primary school play a key role in how children perceive the world around them. Teachers need to teach children about a variety of topics, including those related to disability. One of the most important tasks for teachers in Key Stage 1 is to teach children how to behave and interact appropriately with people with different types of disabilities. It is important to stress that everyone has the right to equal respect and dignity and that everyone should be respected regardless of whether or not they have a disability. Teachers should also teach children about ways to cope with disability and ways in which disabled people can be supported. Learning about how to seek help, for example with special aids, so that people with disabilities can be as independent as possible, and linking them to organizations that help and support people with disabilities, can be very useful for children. For example, it may be possible to invite people with disabilities into the classroom to present their lives and answer children's questions. Children can learn simple ways to help people with disabilities in their daily lives. Disability education should be part of everyday learning in the first year of primary school. This learning can also contribute to reducing the stigma against people with disabilities and creating better conditions for their inclusion in society. In addition to creating a haven for children with different types of disabilities in particular, and equipping them with the skills and knowledge needed for future independence, primary school teachers can also transform children's behavior and their relationship to disability through their work. When today's children grow up in inclusive classrooms, they feel more comfortable working with people with different kinds of disabilities, which can lead to a future society that is broader and more open to the different needs and experiences of people with disabilities. Disability education at the first level of primary school is therefore not only important for creating an inclusive society but also for developing empathy and respect among children. It is one of the basic tasks of teachers to teach children how to behave properly toward people with disabilities. --- Materials and Methods --- --- Design A mixed design was chosen for the research. A questionnaire and a semi-structured interview were used to collect data. --- Measurement A questionnaire which had a total of 19 items was chosen for the data collection. The first 2 items were questions on demographic data , while the remaining 17 items were related to the issue under study. The questionnaire was designed by the authors of this paper on the basis of a questionnaire constructed by [1], which had 12 items, each in the form of a statement to which respondents were asked to select a response in the form of a value on a scale from 1-5, where 1 indicated total agreement with the statement, 2 agreement with the statement, 3 a neutral attitude toward the statement, 4 some disagreement, and 5 total disagreement. All 12 of these items were used to construct the questionnaire for the research presented here. The questions of the questionnaire were constructed to obtain data to assess the respondents' attitudes in the area of three factors , which emerged from the factor analysis. The interview to obtain qualitative data was conducted following the protocol [46]. In translating the items into English, we followed the Standards for Educational and Psychological Testing [47]. The items were translated by an independent translation service, then submitted for consultation with two different special pedagogues and one psychologist. The proposed modifications were consulted with the submitter and a pilot version of the 12 items was created. These items were presented to 15 respondents aged 11-16 years. Suggestions for revisions of the items were again consulted with the translator, and a final version of the 12 original items in the Czech language was created. All the experts who participated found no inconsistencies in the translations of the instrument. The individual items of the original questionnaire were clearly worded, which made the translation easy and understandable. The remaining items were created by the authors of the questionnaire following the research objectives and were again subjected to consultation by two special pedagogues and one psychologist. In this case, too, piloting was carried out with 15 respondents. The last item was left as an opportunity to say anything on the topic. These five items explored the respondents' experiences with persons with disabilities in their own family and social environment. The respondents' awareness was determined by completing a task in which respondents were asked to create pairs of terms that belonged together. The responses from these items were then correlated with items from the original questionnaire. The questionnaire was distributed electronically. Principals of schools across the country educating adolescents aged 11-16 years were contacted. The questionnaire was administered during school hours. Teachers were instructed not to help or influence the adolescents. It was then a random selection within the chosen age range. In the course of contacting schools to ask them to complete the questionnaires, a request was also made to participate in a semi-structured interview. A total of ten respondents answered this call. The interviews were conducted in person. To preserve anonymity, only these ten respondents were contacted to obtain qualitative data in interviews, a sufficient number for this method. The data collection took place from April to November 2022. --- Data Processing The data were fixed in Microsoft Excel 2016; statistical processing was performed in Microsoft Excel and IBM SPSS 29 . The last question in the questionnaire was open-ended and was then assessed qualitatively using open coding. The interviews were again analyzed using open coding. --- Results The questions explored the respondents' experiences with persons with disabilities in their own family and social environment were evaluated first . For all three questions, in the case of an affirmative answer, the type of disability was also collected. In the first case, intellectual disability was the most represented , while in the second and third cases, multiple disabilities were the most represented . It should be added that the term "don't know" included all the responses from respondents who wrote that they did not know, were not sure, or might have had experience but did not know what type or kind of disability it was. The primary question of this research was adolescents' attitudes toward individuals with disabilities. There was a total of 12 questions related to this issue, all of which took the form of a statement. For all the questions, the respondents used all the values of the scale. Basic statistical values were calculated for all items. The mean value for 9 out of the 12 questions was 3. The detailed results show that for six questions, the responses are distributed in such a way that about 40% are neutral responses and the remaining 60% are evenly split between agreeing and disagreeing with the statement. These questions are: --- 7. If I had a family member with a disability, would I tell my friends? 8. I feel comfortable in the company of people with disabilities. 9. I would go to a place where people with disabilities could be. 12. People with disabilities act like children. 14. Individuals with disabilities have a circle of friends similar to mine. 18. An unemployed person with a disability should have priority in employment over an able-bodied person. For Question 18, although there is a slight tendency toward disagreeing answers, it is not as great as for the other questions. The frequencies and percentages of each response for the questions are shown in Table 3. For the remaining six questions, the answers were always on one side of the scale. In the case of one question, this was a significant shift of the rating toward neutral to negative: 13. Children with disabilities should study in normal schools. Only 21% of the respondents strongly agreed or agreed with this statement. In contrast, almost 50% disagreed or strongly disagreed with the statement. For the other questions, it was the other way around-a tendency toward positive to neutral evaluations: 10. Interesting and varied work is suitable for individuals with disabilities. 11. If I am with an individual with a disability, I know how to treat them. 15. Individuals with disabilities should have the same opportunities in life as others. 16. Individuals with disabilities should live together with non-disabled people. 17. Individuals with disabilities can do both team and individual sports. The strongest tendency toward positive evaluation was for Question 15 . Overwhelming agreement was elicited from 50% of the respondents. A total of 77% of the respondents included total agreement and agreement . The second research question was: What is the adolescents' knowledge about individuals with disabilities? There was one question related to this issue, which was divided into five sub-questions. The respondents were asked to connect two related concepts each time. The adolescents have a basic understanding of each area. In the area of knowledge of physical disability, 95% answered correctly, while in the other areas the range was 86-88%. The respondents were expected to combine two concepts that are related to each other . The author of the questionnaire [1] based the interpretation of the results on factor analysis. Three factors emerged from the factor analysis. In addition to the statistical methods mentioned below , factor analysis was also chosen in this case. As a first step, the suitability of the data for factor analysis was examined. We calculated the Kaiser-Meyer-Olkin value = 0.84 , and an "antiimage matrix" was constructed in which the values of the measure of sampling adequacy were greater than 0.5 for all items; more precisely, they ranged from 0.8 to 0.91. The final step involved determining commonality, i.e., the degree of correlation of an item with the other items entering the factor analysis. The commonality values for all items ranged from 0.4 to 0.6. The above data show that the data are suitable for factor analysis. The KMO value showed the average suitability of the data for factor analysis. The other values, MSA and commonality, confirmed the suitability of the items for factor analysis. We extracted three factors that explained 46% of the variance in the items. The factors that were extracted did not correspond to the three scales in the original questionnaire. The principal component analysis method was used to extract the factors and the number of factors was selected on the basis of Kaiser's rule. For easier interpretability, the factors were rotated using the varimax method. The factors listed correspond to the different components of attitudes as defined as acceptance , feelings , and competence in the introduction. Compared to the original questionnaire, the type and number of items that saturate each factor are different. However, the aim of this research was not to review the original questionnaire. Thus, we will continue to work with the results we obtained. An independent samples t-test was used to evaluate Hypothesis H1. First, the hypothesis was tested in relation to all responses and then in relation to the individual factors. In the first case, there was a significant difference in the scores for boys and girls in terms of their attitudes toward individuals with disabilities; t = 6.67, p < 0.001. Using a sign scheme, it was then found that the girls have statistically significantly more positive attitudes toward individuals with disabilities. In the case of testing each factor, it was found that the null hypothesis could be rejected in the case of acceptance and feelings. On the other hand, in the case of competence, the null hypothesis must be accepted. Again, using the sign scheme, it was found that the girls show statistically significantly positive acceptance of individuals with disabilities and statistically significantly positive feelings toward individuals with disabilities: Acceptance t = 4.91, p < 0.001 Feelings t = 7.42, p < 0.001 Competence t = 1.52, p = 0.065 On the basis of the results, we reject the null hypothesis and accept the alternative hypothesis. There is a statistically significant relationship between overall attitudes and gender and between acceptance and feelings and gender. However, this is not confirmed in the case of competence and the null hypothesis must be accepted. A chi-square test of independence was used to test Hypothesis H2. The calculated values of χ 2 = 175.97, p = 0.93 indicate that the null hypothesis cannot be rejected. There is no statistically significant relationship in terms of age and overall attitudes toward individuals with disabilities. However, when one looks at individual factors, a statistically significant relationship with age cannot be confirmed in the case of acceptance and feelings. For competence, the opposite is true. On the basis of the sign scheme, it was found that the eleven-year-olds were statistically significantly more likely to choose responses agreeing with the statements and the fifteenyear-olds were statistically significantly less likely to choose responses that agreed; that is, more of the eleven-year-olds reported knowing how to behave in the presence of individuals with special educational needs. On the basis of these results, the null hypothesis cannot be rejected. There is no statistically significant relationship between overall attitudes toward individuals with disabilities and age and between acceptance and feelings and age. The null hypothesis can be rejected in the case of age and competence. There is a statistically significant relationship between age and competence. The chi-square test of independence was used to test the hypotheses of H3. In the case of Hypothesis H3a , the values of χ 2 = 185.81, p < 0.001 can indicate rejection of the null hypothesis and acceptance of the alternative hypothesis. For Hypothesis H3b , the χ 2 values = 104.17, p = 0.05 lead to rejection of the null hypothesis and acceptance of the alternative. And in the case of Hypothesis H3c , the values of χ 2 = 113.72, p = 0.012 again lead to rejecting the null hypothesis and accepting the alternative. The above suggests that there is a statistically significant relationship between personal experience with an individual with a disability and attitudes toward such individuals. Using the sign scheme and correlations, it was possible to find that previous personal contact influenced attitudes toward disability rather positively . --- Table 6. Correlation between personal experience and attitudes. --- Previous Contact Correlation Individual with a disability in the family r = -0. Using correlations, it was possible to find that previous personal contact influenced attitudes toward disability rather positively . The sign scheme showed a statically significant tendency toward positive feelings for respondents who had previous contact with an individual with a disability. Further, those who had no prior contact were significantly less likely to choose affirmative responses for the feelings factor. --- Correlation Acceptance Correlation Feelings Correlation Competence Individual with a disability in the family NS r = -0.09, n = 1806, p < 0.001 r = -0.07, n = 1806, p = 0.003 Friend with a disability r = -0.07, n = 1806, p = 0.006 r = -0.12, n = 1806, p < 0.001 NS Previous contact with an individual with a disability r = -0.12, n = 1806, p < 0.001 r = -0.12, n = 1806, p < 0.001 NS In the case of the presence of a disability in the family, there is a statistically significant relationship to all factors. For the existence of a disabled friend, this statistically significant relationship exists only for feelings, and for contact with an individual with a disability, for acceptance and feelings. There is a statistically significant relationship between prior experience of contact with an individual with a disability and overall attitudes toward individuals with disabilitieswe accept the alternative hypothesis. For each factor, we reject the null hypothesis for the presence of an individual with a disability in the family. In the case of a friend with a disability, the null hypothesis can only be rejected for feelings, and for the last factor, for both acceptance and feelings. The chi-square test of independence was chosen to test Hypothesis H4. The calculated values for all items χ 2 = 491.68, p < 0.001 indicate that the alternative hypothesis, i.e., that there is a statistically significant relationship between knowledge and attitudes, should be accepted. The sign scheme showed that those who have a significantly good level of knowledge have significantly better attitudes toward individuals with disabilities and, at the same time, those who have a low level of knowledge about the world of individuals with disabilities have significantly worse attitudes toward these individuals. The possibility of rejecting the null hypothesis is then confirmed by the values for the individual factor domains: Acceptance χ 2 = 225.54, p < 0.001 Feelings χ 2 = 177.65, p < 0.001 Competence χ 2 = 14.64, p < 0.001 Again, the sign schema showed that good knowledge of the world of people with disabilities leads to better overall perceptions of these individuals, whether in the areas of acceptance, feelings, or competence. The null hypothesis can be rejected for both overall attitude and individual factors. The interviews were transcribed and analyzed using open coding. The following categories gradually emerged: regret, admiration, taboo, inclusion and segregation, differentiation --- Pity • Disability is a very tragic thing. --- • I feel sorry for them; it must be hard to live with a disability. --- • It's hard. --- • Because I find this topic sad even though I know we should talk about it but I feel sorry for these people. --- Admiration • I would like to say that a lot of people shy away from disabled people and it's a shame. People should know that they tend to be the nicest people. They should learn more about them and treat them like everyone else in life. --- • Disabled people are fine, I did the Swan Lake dance in primary school and it was an experience. --- Taboo • Why did you have such sensitive questions here? Could anyone have been hurt or mentally harmed? • I don't like to talk about this topic. --- • I don't feel comfortable talking about it. --- Inclusion and segregation • I think people with disabilities or other disabilities should have the same opportunities as non-disabled people but I would make schools separate because there is a higher risk of bullying. --- • I know from experience that some people with disabilities would be able to integrate into normal life quite easily, but most disabled people would not be able to get on properly and get along with people in 'normal' society. --- • We are learning more about this topic every day because one of my parents works in a special school and since I know how teaching is done in these schools, I think that being with qualified staff can help with inclusion among students with similar disabilities. However, pupils with physical disabilities should be taught to the same extent as able-bodied pupils. --- Differentiation --- • Depending on the type of disability The categories that emerged from the data collected in the interview corresponded with the categories that emerged from the analysis of Item 19 in the "opportunity for the expression" questionnaire. --- Discussion A total of 1806 respondents aged 11-16 years were approached to investigate adolescents' knowledge and attitudes toward individuals with disabilities. Two primary research questions were asked: --- What Are the Attitudes of Adolescents toward Individuals with Disabilities? Statistical evaluation shows that the adolescents have rather positive attitudes toward individuals with disabilities. These attitudes are stronger among the girls than the boys. Similar conclusions were reached by Loeffler and Greitemeyer [48]. Different research conclusions were reached by Sharma, Pratap Yadav, and Sharma [49]. The results also showed that the girls are also more open to sharing information and attitudes. Using a sign scheme, the evidence was that if girls have someone with a disability in their family, they are significantly more likely to disclose this fact than boys are. Age no longer plays a role in these attitudes. However, on deeper examination, age was shown to play a role in relation to competence. The younger the individuals are, the more positive their view of individuals with disabilities. Looking more closely at the attitudinal assessment, the results show that, in general, adolescents have no problem meeting individuals with disabilities. Almost 70% of the respondents agree that individuals with disabilities should have equal opportunities, and almost 60% state that these individuals should live together with non-disabled individuals, i.e., they lean toward the idea of an inclusive society. The exception is the attitude toward co-education. There, on the contrary, almost 50% of the respondents reject this attitude. This finding among the sample of children in the present study is not consistent with the claim that has been made elsewhere that a positive attitude toward inclusive education is not a majority, but rather neutral with a preponderance of positive responses [50]. Here, a twofold explanation is offered; the first possibility is that it is an attitude where the individual generally agrees with inclusion but then in the case of personal space this is a problem. The answers to the questions that dealt with attitudes to the personal space then rather indicate that attitudes were normally distributed in the sample of respondents . This means that it is not possible to say unequivocally how attuned the adolescents are to face-to-face encounters; it can only be stated that some are and some are not. Another reason may be that young people perceive social inclusion to be highly important. Thus, special education is not seen as segregating, but rather as a space in which to receive an education that is tailored to individuals with special educational needs. This education does not discriminate against them, but rather enables normal participation in society in the areas of work and social inclusion [51]. It has also been shown that previous experiences with individuals with disabilities can exert a positive influence on attitudes toward individuals with disabilities [52]. This is most significant in the area of feelings. A statistically significant association was shown for all three types of encounters. This effect is then perceived as positive. Often, it is a fear of the unknown that stems from a lack of information and can persist into adulthood and have a negative effect on professional life [53]. It is interesting to note that when questions were asked about the type of disability in friends, the following types of disability did not appear once: autism spectrum disorder, attention deficit hyperactivity disorder, and specific learning disability. These are the types of disabilities that are perhaps the most challenging in terms of relationships as a result of the difficulty in communication and the competencies needed to engage in activities together, particularly play. In this context, it is worth noting that the responses to the questions on personal experiences with individuals with disabilities revealed that 79% of the respondents had no family members with disabilities and 67% had no friends with disabilities. This is understandable as nowadays individuals with disabilities naturally appear in society and participate fully in different areas of life. The phenomenon of "friend" versus contact is interesting. From the above data on inclusive education, it is more than evident that almost every school classroom has at least one pupil with SEN. However, here the fact that most respondents probably do not perceive them as friends, or perceive as friends only those with milder types of disabilities, becomes apparent. This corresponds with the finding above, where only 21% of the respondents perceive joint education as appropriate. Five basic codes emerged during the open coding assessment of the free responses and interviews: pity, admiration, taboo, inclusion and segregation, and differentiation. The open-ended responses revealed that approximately 20 respondents indicated that they felt tremendous regret when they encountered individuals with disabilities. Approximately 20 respondents said they found the topic unpleasant and painful and did not want to talk about it. In one case, we were even warned that this was not an appropriate topic and that the questions might cause hurt. In two cases, there was admiration. Even though this only occurred twice, a separate category was created from these responses. We perceived a strong emphasis on this view from these respondents. The last two categories are intertwined. Attitudes toward the disabled are driven by a variety of factors. Pity can be triggered by compassion for a person who is struggling with difficulties and limitations. Admiration may arise from respect for a person who is able to overcome his or her difficulties and achieve success. Taboo can be associated with negative stereotypes and prejudices against persons with disabilities. Inclusion is associated with an attitude where persons with disabilities are fully integrated into society without any discrimination or segregation. Segregation, on the other hand, is a view in which persons with disabilities are separated from the rest of society and are subjected to various types of discrimination. Differentiation can be linked to the provision of special services and support for people with disabilities to enable them to reach their full abilities and potential. It is important to engage in dialogue and seek solutions that lead to better inclusion of persons with disabilities in society without any restrictions. The results of the research presented here show the main categories that make up attitudes toward persons with disabilities and also the themes that can be used for possible intervention. Importantly, for about 150 respondents, their attitudes would depend on the type of disability. These respondents reported that their greatest fear was coming into contact with individuals with intellectual disabilities. They stated that they did not know how to treat them or were embarrassed about not understanding their reactions. The ten interviews revealed the same. The respondents reported that they stood by their general attitudes toward individuals with disabilities, but that they would change their attitudes a little in the case of individuals with intellectual disabilities: "that I am afraid of people with disabilities. . . I imagined people with intellectual disabilities and I didn't think about, for example, physical disabilities as a result of an accident. I wouldn't be so worried there. . . that different types of disability cannot be put on the same level in an evaluation. . .". Prior to the actual research, we considered whether to construct the questionnaire to measure general attitudes or to separate attitudes and different types of disability. Given the current talk of an inclusive society, we decided to construct one on general attitudes [54]. In line with the inclusive concept, no distinctions should be made on the basis of the type of disability. Even so, some adolescents do not perceive these attitudes in the same way. --- What Is the Knowledge of Adolescents about Individuals with Disabilities? The results presented show that almost 80% of the respondents have a clear basic knowledge of individuals with disabilities. Only 15 individuals did not assign even one answer correctly. Good knowledge then leads to positive attitudes toward individuals with disabilities and vice versa. In the free response question and follow-up interviews, it was found that most information about individuals with disabilities is obtained by young people at school and from the media , as well as in the family or through personal contact with an individual with a disability. The application of the research results has an impact on education. On the basis of the findings, it is possible to develop and strengthen educational programs aimed at educating and informing teenagers about persons with disabilities. These educational measures can help to reduce prejudice, stereotypes, and discrimination. The area of prevention is also important. On the basis of the facts that were established, it is possible to propose prevention aimed at strengthening positive attitudes and openness to discussing individuals with disabilities. Prevention may include workshops, presentations, or discussion groups that allow adolescents to better understand and respect the needs of individuals with disabilities. Skills and their development could also provide support for adolescents to feel comfortable interacting with individuals with disabilities, especially intellectual disabilities. This may include creating a friendly environment in which questions and discussions about disability are open and providing information on how to communicate and treat individuals with different types of disabilities. Education should also include training in empathy and understanding. Adolescents should be encouraged to understand the barriers and challenges faced by people with disabilities and how they can contribute to creating an inclusive and respectful environment. A significant role is also played by the involvement of adolescents in community activities or volunteer projects where they can meet individuals with disabilities. This interaction could help raise their awareness, change attitudes, and provide them with an opportunity to build personal relationships with these individuals. One of the findings is the difference in the attitudes of boys and girls. The findings indicate that boys, compared to girls, have fewer positive attitudes toward individuals with disabilities and are less open to discussing the topic. Therefore, special programs or initiatives aimed at supporting the positive attitudes of boys and their openness toward communication and interaction with individuals with disabilities should be targeted. Overall, these findings should be used to create an inclusive environment that embraces and respects diversity. --- For Further Research Follow-up research activities should focus on the analysis and detailed description of attitude components such as awareness, behavior, and emotional adjustment of adolescents. It would also be advisable to conduct research using the same methodology in other age groups to gain cross-sectional results with information on the development of attitudes toward the disabled in the non-disabled population to gain a deeper understanding of the issue. Another task would be to analyze curriculum documents at all levels of the educational system of the Czech Republic. By analyzing the texts, it would be possible to name opportunities suitable for educating or informing about persons with disabilities. --- Conclusions The research has shown that more than 70% of the adolescents aged 11 to 16 who were surveyed have basic knowledge of individuals with disabilities. Further, these adolescents were shown to have an overall positive attitude toward individuals with disabilities. This relationship is statistically significantly better for the girls than the boys, and the girls are also more open to discussing individuals with disabilities. At the same time, even when attitudes are positive, they still depend on the type of disability. The greatest insecurity or discomfort is shown when interacting with individuals with intellectual disabilities. Furthermore, positive attitudes appeared to be directed toward social functioning but not toward learning together. Here, rather segregationist attitudes became clear. For future research, it would be interesting to focus on attitudes toward specific types of disability. The results obtained could then be reflected in the educational process. In today's inclusive society, it is crucial to foster relationships between non-disabled children and their disabled classmates. These relationships promote empathy and understanding and create an environment where all students feel valued and included. To achieve this, several strategies can be implemented. Firstly, it is essential to educate nondisabled children about disabilities. As a result of their awareness being increased, children will better understand the challenges their disabled classmates face and also their unique abilities. Group projects or classroom activities encouraging collaboration among students with diverse skills can help break down barriers and build friendships. By highlighting each student's strengths, regardless of ability, a sense of belonging will be cultivated among all classmates. Promoting peer support programs can significantly enhance relationships between non-disabled children and their disabled classmates. Pairing students together for mentoring or buddy systems allows for regular interaction outside the classroom, fostering genuine connections based on mutual trust and support. Another option is the intervention of a special educator or psychologist to help correct the relevant deficits. --- Implication and Contribution to Further Research Subsequent research activities should be directed toward the analysis and detailed description of components of attitudes such as awareness, behavior, and emotional adjustment. --- Study Limits A random sampling method was used to select the sample, but it does not reflect the total population of the age group under study. This results in the limitation of the study through "sampling bias". To obtain more accurate results, it would be desirable to focus more on a larger representation of schools by type of study-vocational training schools and schools preparing for further study. Overall, a larger number of respondents always leads to more conclusive results. The study was based on "volunteering" to work with the researchers. --- Data Availability Statement: The data presented in this study are available on request from the corresponding author. The data are not publicly available for privacy reasons. --- Author Contributions: Conceptualization, A.S., M.P. and P.P.; methodology, P.P. and M.P.; software, A.S.; validation, P.P. and M.P.; formal analysis, P.P.; investigation, M.P.; resources, M.P. and A.S.; data curation, P.P.; writing-preparation of original draft, P.P.; writing-review and editing, M.P.; visualization, P.P. and M.P.; supervision, A.S. All authors have read and agreed to the published version of the manuscript. --- Informed Consent Statement: The survey participants were informed about the purpose of the research and the anonymous publication of the results. As this was an online survey, if they agreed to complete the questionnaire, they were agreeing to the terms and conditions. All participating schools had general informed consent from the pupils' parents. ---

The presented research is a study of the attitudes of adolescents toward individuals with disabilities and a description of the level of knowledge about this issue among a selected group of adolescents. The study is focused on the school social environment, specifically on pupils without disabilities and their attitudes toward pupils with disabilities, which is one of the factors influencing their school success. The target group was 1806 pupils between the ages of 11 and 16 from the entire Czech Republic, approached by random selection. The research tool for the quantitative approach was a questionnaire developed by the authors of this article. The questionnaire section with open questions served for the qualitative part of the research. The results show that 70% of the respondents have basic knowledge about people with disabilities. Furthermore, it was shown that these adolescents have an overall positive attitude toward people with disabilities. This attitude is statistically significantly better among the girls than the boys, and the girls are also more open to discussing individuals with disabilities. At the same time, even if attitudes are positive, they still depend on the type of disability. The greatest uncertainty or discomfort is manifested when interacting with individuals with intellectual disabilities.

Introduction In developed countries, overall, 8.6% of all livebirths are born preterm [1], however, inequality is pervasive [2,3]. As preterm birth is associated with higher mortality [4], poorer neurological development, behavioural, social, and learning difficulties [5], being born preterm presents a threat to children's health and ability to reach their life potential. A socioeconomic gradient in the risk of preterm birth is well documented [2] even in countries with universal access to antenatal care [6,7]. Inequity in preterm birth thus presents a significant public health challenge requiring identification of pregnant women at particularly high risk of preterm birth in order to target interventions. When examining inequality in preterm birth, epidemiological studies have generally considered single risk factors [8] without considering that these might interact. In this way, disadvantaged socioeconomic position, mental health conditions, and younger and older age are all found to be independent risk factors of preterm birth [2,9,10]. In pregnant women, disadvantaged socioeconomic position is associated with both young maternal age and mental health conditions [11]. Generally, mental health conditions are the leading cause of illness among women aged 15 to 44 years [12]. In developed countries, 15.6% of all pregnant women experience a mental health condition [13], but young pregnant women are at particularly high risk [14]. Despite these associations between socioeconomic position, mental health conditions, and age, it is unknown whether these independent risk factors interact in their contribution to the inequality in the risk of preterm birth. Analyses of additive interaction can clarify this by examining whether the observed joint effects of the exposures are greater or less than the expected based on summing their independent effects on preterm birth [15]. In this way, examining additive interaction can help identify subgroups where limited resources can be of best use [16]. Therefore, it is a highly relevant public health measure [17] although it is not commonly used. To identify relevant subgroups for intervention in order to reduce inequality in preterm birth, the aim of this study was to estimate the risk of preterm birth according to different combinations of socioeconomic position and mental health conditions in different age groups and examine additive interaction between these three risk factors. --- Methods --- Design This study was a Danish nationwide register-based cohort study. --- Setting In the Danish tax-based healthcare system [18], the antenatal care is free of charge and used by the majority of all pregnant women [19,20]. --- Data sources Data on the mother and the child was retrieved from the following nationwide registers: The Danish Medical Birth Register [21], Danish National Patient Registry [22], The Danish national prescription registry [23], Danish National Health Service Register [24], and Statistics Denmark's registers on population and education [25]. Linkage between databases was conducted on an individual level by means of the personal registration number, a unique identifier assigned all Danish individuals and used in all public registers in Denmark enabling linkage between them [26]. Statistics Denmark conducted the linkage and anonymised data. Information on databases associated with codes and algorithms identifying exposures and outcome are available in Supplementary tables s1-s3 [Additional file 1]. --- Study population We identified all liveborn infants in Denmark in the period 1 January 2000 to 31 December 2016 and their mothers. We included first-time mothers who gave birth to a singleton liveborn infant. To increase the probability that all mental health conditions of the study population were registered, we excluded women not living in Denmark in the five-year period prior to birth as this was the period where mental health conditions were considered. Women with missing data on parity, gestational age, education, or maternal age were excluded . --- Variables The outcome measure was preterm birth, defined by the World Health Organisation as birth before 37 completed weeks of gestation [27]. In Denmark, gestational age is corrected according to early ultrasound examination [28] received by > 90% of all pregnant women [20]. Maternal age was categorised into the following three categories: ≤23, 24-30, and ≥ 31 years because of the lowest risk of preterm birth in Danish women aged 24-30 years [10]. The socioeconomic measure of interest was highest maternal educational level attained at birth of the child, as educational level is a strong predictor of preterm birth [6]. In accordance with the International Standard Classification of Education [29], we categorised maternal educational level into three categories: Low educational level was defined as primary school, equivalent to 10 years of mandatory education, corresponding to the ISCED level 0-2. Intermediate educational level was defined as ISCED level 3-4, consistent with secondary or vocational education. High educational level was defined as ISCED level 5-8, corresponding to a short-cycle tertiary education or above. As mean age at commencement of study at ISCED level 5 is in the mid-twenties in Denmark [30], we expected only few first-time mothers ≤23 years at the high educational level, and therefore high and intermediate education were merged for this age group resulting in two educational categories. Mental health conditions were categorised in three mutually exclusive severity groups: Minor mental health conditions were defined as mental health conditions managed in the primary healthcare system measured as contact to private psychologist, at least two psychometric tests or two sessions of talk therapy with general practitioner, or at least two redeemed prescriptions of benzodiazepines or antidepressants. Moderate/severe mental health conditions were defined as contact to a private psychiatrist or mental health conditions managed in a psychiatric hospital . No mental health conditions were assigned if none of the criteria above was met. Because mental health conditions often are enduring or recurrent [31], all contacts, conditions, and medication were considered in a window of 5 years before the birth of the child. --- Statistics Poisson regression with a robust error variance [32], was used to estimate the relative risk of preterm birth in the different combinations of education and mental health conditions. The analysis was stratified by agegroup, and in all age strata the reference was women with the combination of high education and no mental health conditions. With the aim of identifying women at particularly high risk of preterm birth as relevant subgroups for intervention in order to reduce inequality in preterm birth, no adjustments were conducted as adjustment may remove important effects between social positions [33]. In order to measure additive interaction, we performed two interaction analyses where we calculated attributable proportions defined as the proportion of the risk that is due to interaction in the doubly exposed groups [16]. In the first analysis, we examined additive interaction between education and mental health conditions in each age stratum. In this analysis the doubly exposed groups were those exposed to low or intermediate education and minor or moderate/severe mental health conditions : p = the absolute risk of preterm birth in each combination of education and mental health conditions, E-indicates high education , and M-indicates no mental health conditions . AP takes values between -1 and + 1. AP > 0, AP < 0, and AP = 0 suggests positive, negative, and no additive interaction, respectively [16]. In the second interaction analysis, we examined additive interaction between age and the different combinations of education and mental health conditions. In this analysis, we calculated AP in the groups that were doubly exposed to both age ≤ 23 or ≥ 31 years and each of the different exposed combinations of education and mental health conditions : A-was age 24-30 years and EM-was the combination of high education and no mental health conditions . Given that the proportion of women registered with a mental health condition increased during the study AP = RR E+M+ -RR E+M--RR E-M+ + 1 RR E+M+ = p E+M+ -p E+M--p E-M+ + p E-M- p E+M+ AP = RR A+EM+ -RR A-EM+ -RR A+EM-+ 1 RR A+EM+ = p A+EM+ -p A-EM+ -p A+EM-+ p A-EM- p A+EM+ period [34], we performed supplementary analyses, including all main analyses described above for women giving birth in the periods 2000-2008 and 2009-2016, separately. Furthermore, we performed a supplementary analysis of the risk of extreme preterm birth, defined by WHO as birth before 28 completed weeks of gestation [27], because these children are facing the largest risk of death, disability and use of resources [35]. For some short-term mental health conditions, including information 5 years prior to birth might be too long. Therefore, sensitivity analyses were performed considering maternal mental health conditions 2 years instead of 5 years before the birth of the child. Analyses were conducted using Stata version 15.1, College Station, TX, USA. --- Ethics No ethical approval is required for register-based studies in Denmark [26]. --- Results A total of 415,523 women fulfilled the inclusion criteria . Table 1 presents the total number and percentage of women in each age stratum with different combinations of education and mental health conditions. A larger proportion of the 62,363 women aged ≤23 years had a mental health condition compared to the 227,839 women aged 24-30 years and the 125,321 women aged ≥31 years . In the 27,252 women aged ≤23 categorised with high/intermediate education, only 1535 had attained a high education corresponding to ISCED level 5. During the study period, 6.3% gave birth prematurely to a liveborn infant. In the women aged ≤23, 24-30, and ≥ 31 years 6.5, 6.1, and 6.7% gave birth prematurely, respectively. In all age strata, the absolute risk of preterm birth increased with decreasing educational level and increasing severity of mental health conditions . The highest absolute risk was 12.9% [95% CI: 11.2;14.8%] in women aged ≥31 years with low education and moderate/severe mental health conditions. The highest relative risk in women aged ≤23, 24-30, and ≥ 31 years was found in those with low education and moderate/severe mental health conditions who had 25% [95% CI: 14-37%], 53% [95% CI: 39-68%], and 123% [95% CI: 93-158%] higher risk of preterm birth, respectively, compared to the unexposed reference groups . In the first interaction analysis, we found positive additive interaction between low education and minor and moderate/severe mental health conditions in women aged 24-30 years and between low education and moderate/severe mental health conditions in women aged ≥31 years . The AP in the last-mentioned group indicates that 21% [95% CI: 7;34%] of the absolute risk of preterm birth in this doubly exposed group was due to additive interaction as illustrated in Fig. 2. The second interaction analysis where we examined additive interaction between age and the combinations of education and mental health conditions are presented in Fig. 3. We found negative additive interaction between age ≤ 23 and the combinations of low education and no, minor, and moderate/severe mental health conditions . The expected joint effect based on summing the independent effects of both age ≤ 23 years and the combination of low education and moderate/ severe mental health conditions was 19% greater than the observed effect indicated by the AP of -0.19 [95% CI: -0.33;-0.04]. We found positive additive interaction with age ≥ 31 years in most of the eight combinations of education and mental health conditions . For women with the combined exposure of low education and moderate/severe mental health conditions, further being exposed to age ≥ 31 resulted in an AP of 0.32 [95% CI: 0.21;0.44%] indicating that 32% of the risk of preterm birth among these women could be explained by the interaction itself. The supplementary analyses stratified by study period are presented in Supplementary tables s4-s14 [Additional file 1]. Overall, the risk of preterm birth decreased from 6.5% [95% CI: 6.4;6.6%] in 2000-2008 to 6.1% [95% CI: 6.0;6.2%] in 2009-2016, corresponding to a decrease in the overall risk in all three age strata . However, the relative risks and the AP measures did not vary remarkably between the first and the last part of the study period. The supplementary analysis of the risk of extreme preterm birth showed to some extent similar patterns of the risk increasing with lower educational level and the severity of mental health conditions, see Supplementary tables s15-s16 [Additional file 1]. In general, for women aged ≥24 years the relative risks were higher than in the main analyses. However, there were Fig. 2 The results of the first analysis of additive interaction between educational level and mental health conditions in pregnant women aged ≤23 , 24-30 , and ≥ 31 years. a illustrates an example of how to read the attributable proportion in the bar charts a limited number of extreme preterm births in our study population, resulting in very broad confidence intervals. Our sensitivity analyses, where we considered maternal mental health conditions 2 years instead of 5 years before the birth of the child, showed increased risks for the women with mental health conditions . However, the AP measures were similar to the AP measures in the main analyses. --- Discussion In this study, we found that the risk of preterm birth increased with decreasing educational level and increasing severity of mental health conditions in all age groups. However, this inequality increased substantially with increasing age. The positive additive interaction between low education and mental health conditions in women aged 24-30 and ≥ 31 years, found in the first interaction analysis, indicates a higher impact on the risk of preterm birth when doubly exposed in these two age groups. The second interaction analysis of the additive interaction between age and the combinations of education and mental health conditions revealed negative additive interaction with age ≤ 23 years and positive additive interaction with age ≥ 31 years. This indicates that with increasing age, the impact of education and mental health conditions, both separately and in combination, are more consequential to the risk of preterm birth. The results from this interaction analysis further suggests that when resources are limited, intervention strategies may have the potential to prevent a larger proportion of Fig. 3 The results of the second analysis of additive interaction between age and the combinations of education and mental health conditions. a illustrates an example of how to read the attributable proportion in the bar charts. b presents the results of the analyses of additive interaction with age ≤ 23 year, and c presents the results of the analyses of additive interaction with age ≥ 31 years. In both and , the unexposed group was women aged 24-30 years with the combination of high educational level and no mental health conditions preterm births if targeting women with higher age further combined with lower educational levels and mental health conditions. In 2009, a shift was seen in the national Danish antenatal guidelines from universal towards differentiated services, as recommended by WHO [36], with the intention of giving greater priority to disadvantaged pregnant women [37]. Potentially, this differentiated antenatal care could have reduced the inequality in preterm birth. However, our supplementary analyses revealed that inequality remained unaffected before and after 2009, despite the decrease in the overall risk of preterm birth. The direction of increasing risk of preterm birth in women with mental health conditions and/or decreasing educational level found in this study is consistent with other Danish and international studies examining the exposures separately [3,6,7,9,38]. However, the suggested mechanisms underlying the inequality in preterm birth are complex and not fully understood [39,40]. Often, inequality in preterm birth has been attributed to socially patterned lifestyle [41]. However, lifestyle factors such as smoking, alcohol consumption, and body mass index, only explain a small part of the educational disparities in preterm birth [6,41]. Studies have found that preterm birth is associated with psychological and social stress [42,43] which is considered leading to preterm birth through neuroendocrine, inflammatory, and immunological mechanisms [39,42]. Socially disadvantaged pregnant women may lead more stressful lives due to exposure to stressors such as unemployment, financial hardship, discrimination, unstable social relations, and lack of social support [40,44]. The combination of low or intermediate education and mental health conditions could entail further accumulation of stressors compared to women separately exposed and therefore explain some of the higher risk of preterm birth. It is suggested that the risk of negative birth outcomes increases with longer duration of exposure to risk factors and that repeated exposure to stressors could increase the risk of preterm birth [45]. Longer duration could explain some of the more consequential impact on preterm birth of low education and mental health conditions for pregnant women at advancing age found in this study. --- Strengths and limitations There are several strengths to this study. The national Danish registers contain high-quality data covering the entire population [26]. This resulted in a nationwide cohort which enabled examination of the risk of preterm birth in numerous combinations of educational level, mental health conditions, and age allowing for identification of relatively specific high-risk subgroups. Minor mental health conditions are often undiagnosed [46] and therefore not included in register-based studies. By including medication and contact to general practitioners and private psychologists, we were able to identify pregnant women with minor mental health conditions which we found had a noteworthy higher risk of preterm birth. There are also some limitations to this study. We only identified mental health conditions of women who sought medical care and were registered with contact to the primary or secondary healthcare system or redeemed prescriptions within a window of 5 years before the birth of the child. Therefore, some women with mental health conditions might have been misclassified with no mental health conditions. The consideration of mental health conditions in a window of 5 years may also have classified some women as having a mental health condition at childbirth even though they were actually recovered. Accordingly, our sensitivity analyses of mental health conditions considered 2 years instead of 5 years before childbirth showed slightly increased risks for the women with mental health conditions. However, this finding did not change the overall interpretation of the study results. Although misclassification of the highest educational level attained is unlikely due to mandatory registration of completed education by the educational institutions [25], the merging of high and intermediate education in women aged ≤23 years might have led to bias towards the null. However, the proportion of women who had reached a high education within this age group was limited. Stillbirths is associated with preterm birth [47] but were not included in this study. Though stillbirths do not count numerous births in Denmark [48], we may have underestimated the true burden of the inequality in preterm birth by not including stillbirths, which are inversely associated with maternal educational level [7]. Due to the exclusion criteria, the results might not be generalisable to women with multiple pregnancies, as these women have an increased risk of preterm birth, and to multiparous pregnant women, because previous preterm birth increases the risk in later pregnancy [5]. The women excluded due to the criteria of a Danish registered address, are likely to be recent immigrants. Hence, the results might not be generalisable to all immigrant pregnant women. --- Implication of findings The substantial increased risk of preterm birth found in women with combinations of decreasing educational level and increasing severity of mental health conditions emphasises the importance of identification of these disadvantaged pregnant women in the antenatal care to reduce the inequality in preterm birth. Routine antenatal psychosocial risk assessment may increase awareness of these psychosocial risks [49]. Our results indicate that such psychosocial assessment should not neglect minor mental health conditions. Systematic screening during pregnancy is important to ensure that pregnant women with mental health conditions and lower educational levels are referred to relevant, tailored services and that such specialized, supportive interventions are freely available to women, based on their individual needs. In many countries, public health policy has focused attention on younger mothers and their adverse perinatal outcomes [10]. However, despite the larger proportion of mental health conditions in women aged ≤23 years our findings reveal that intervention strategies with the purpose of reducing inequality in preterm birth should target women with higher age further combined with lower educational levels and mental health conditions, especially when resources are limited. That the inequality remained unaffected from the first to the last part of the study period despite greater priority to disadvantaged pregnant women implicates that improved intervention strategies are needed targeting disadvantaged pregnant women with lower educational levels and mental health conditions, and particularly those aged ≥31 years. Further studies are needed to examine variables that drive the heterogeneity across the social positions found in this study in order to improve intervention strategies targeting disadvantaged pregnant women with lower educational levels and mental health conditions. --- Conclusion Substantial inequality in preterm birth remains with increasing risk in disadvantaged pregnant women with decreasing educational level and increasing severity of mental health conditions. The inequality in preterm birth increased with increasing age. Thus, more awareness of women with higher age further combined with lower educational levels and mental health conditions is needed in the prevention of the inequality in preterm birth. • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research ? Choose BMC and benefit from: ? Choose BMC and benefit from: --- --- --- Additional file 1: Supplementary --- --- --- Competing interests The authors declare that they have no competing of interests. ---

Background: Inequality in preterm birth is a world-wide challenge that has proved difficult for maternity care services to meet. Reducing the inequality requires identification of pregnant women at particularly high risk of preterm birth in order to target interventions. Therefore, the aim was to estimate the risk of preterm birth in women with different combinations of socioeconomic position, mental health conditions, and age.In this nationwide register-based cohort study, we included all first-time mothers that gave birth to a singleton liveborn infant in Denmark between 2000 and 2016. The absolute and relative risk of preterm birth (< 37 weeks of gestation) was examined in different combinations of educational level (high, intermediate, and low) and mental health conditions (no, minor, and moderate/severe) in three age strata (≤23, 24-30, and ≥ 31 years). We estimated the relative risk using Poisson regression with a robust error variance. As additive interaction can help identify subgroups where limited resources can be of best use, we measured the attributable proportion to assess the risk that is due to interaction of the different exposures. Results: Of the 415,523 included first-time mothers, 6.3% gave birth prematurely. The risk of preterm birth increased with decreasing educational level and increasing severity of mental health conditions in all age strata, but most in women aged ≥31 years. The highest absolute risk was 12.9% [95% CI: 11.2;14.8%] in women aged ≥31 years with low education and moderate/severe mental health conditions resulting in a relative risk of 2.23 [95% CI: 1.93-2.58] compared to the unexposed reference group in that age strata. We found positive additive interaction between low education and mental health conditions in women aged 24-30 and ≥ 31 years and between age ≥ 31 years and combinations of mental health conditions and educational levels.The inequality in preterm birth increased with increasing age. To reduce inequality in preterm birth focused attention on women with higher age further combined with lower educational levels and mental health conditions is essential.

INTRODUCTION At the end of December 2019, a patient in Wuhan, China, reported pneumonia due to an unknown microbial pathogen . The new Coronavirus was later identified as a pathogen and temporarily named the coronavirus novel 2019 . On January 30 2020, the World Health Organization announced that the emergence of the coronavirus novel was a public health emergency of international concern . Infectious diseases have become one of the main threats to global public health in the 21st century. The disease can have an impact on physical and mental health. Many studies have shown the impact of infectious disease outbreaks on people's mental health, such as severe acute respiratory syndrome in 2003 and the 2009 influenza A epidemic. These types of epidemics direct people to experience psychological problems such as stress disorders post-trauma, psychological distress, depression, and anxiety . On April 27, 2020, 2,990,559 people were confirmed to be infected with COVID-19 and caused death in 207,446 people worldwide. The COVID-19 outbreak caused public panic and mental health pressure, with more and more cases being diagnosed and several countries appearing to have confirmed cases and public concerns about infections . COVID-19 cases in Indonesia, as of April 27 2020, had reached 9,096 people, with 765 deaths spread across 35 provinces . In Central Java alone, COVID-19 case number 19 was the fourth rank in Indonesia, with cases of 666 people and the number of dead 58 people . By looking at the relatively short distribution process, this number is certainly not small and will continue to grow. The Ministry of Health of the Republic of Indonesia and the COVID-19 Task Force team continue to strive to raise public awareness about prevention and intervention strategies by providing daily updates on surveillance and active cases on websites and social media. In addition, many private media or internet users also release and transfer information on media social, such as YouTube, Instagram, and WhatsApp. Social media can cause information overload, which can lead to differences in perception and problems in mental health . WHO points out that the driving factors underlying fear, anxiety, and stigma are triggered by misinformation and rumors, especially those spread through social media. Neria and Sullivan show that indirect exposure to the media can increase Post-Traumatic Stress Disorder . A study by Choi et al. also shows that exposure to social media might be positively related to the formation of risk perception during the MERS outbreak in South Korea. But no studies examine the relationship between social media exposure and mental health problems and perceptions of COVID-19, especially for adolescents. The use of social media among adolescents today cannot be avoided anymore. Adolescents use social media almost daily to find information or convey their activities. The results of a survey conducted by Suratman show the number of adolescents who use Facebook as many as 65 million users, Twitter 19.5 million users, Google+ 3.4 million users, LinkedIn 1 million users, and Path 700 million users . One psychological impact during the COVID-19 pandemic was psychosomatic, a disorder of the mind that causes physical complaints without having the disease . One of the psychosomatic causes lately is anxiety triggered by news related to COVID-19. Because every minute, we can be exposed to various news about COVID-19 through social media, WhatsApp groups, and TV news. As a result, the COVID-19 pandemic threatens physical and mental health . The high number of COVID cases in Central Java and the large number of adolescents accessing social media can undoubtedly impact adolescent perceptions of COVID-19 and cause mental health problems. Until now, no research has examined the relationship between social media exposure to perceptions and adolescents' mental health in Central Java. This study aimed to describe adolescent perceptions related to COVID-19, the prevalence and distribution of two major mental disorders, anxiety and depression, in adolescents in Central Java and examine this relationship with social media exposure. --- MATERIAL AND METHODS --- Research design This study used a cross-sectional method where the independent and dependent variables were taken simultaneously. However, the cross-sectional test is weak in testing the causality relationship between variables. In this study, this weakness was not a problem because the independent variables used were the existing conditions of the respondents. --- Population and research sample The population was adolescents in Central Java aged 10-24. The estimated sample obtained using the Lameshow formula was 1357 adolescents. --- Ethical considerations This research was approved with Ethical Clearance Letter Number 057/ KEPK/EC/2020. It was issued by the Health Research Ethics Commission at Universitas Negeri Semarang on February 22, 2020. --- Data collection The study used snowball sampling techniques to distribute questionnaires online. The online questionnaire was created using the Google form platform at . The online questionnaire was distributed to circles of friends on WhatsApp. When a participant filled out the questionnaire, they passed it on to a circle of friends on WhatsApp to expand the sample size and distribution. In this study, three questionnaires were used, namely the WHO-Five Well-Being Index , Generalized Anxiety Disorder , and a questionnaire developed by researchers using the theoretical framework approach Health Belief Model . The WHO-5 and GAD-7 questionnaires are standard instruments from WHO, which were adopted into the Indonesian version. For the preparation of the HBM questionnaire, the researcher used a literature review. After the questionnaire was compiled, the linguists and substance experts were consulted. After making improvements according to the input of experts, validity and reliability tests were then carried out. Depression was measured using the Indonesian version of the WHO-Five Well-Being Index , which consisted of five question items that reflected the presence or absence of depression symptomatology . Respondents were asked to report their feelings in the past 2 weeks on a 6-point scale starting from every time to never . Points less than 13 indicate depression . Anxiety was measured using scale Indonesian version of Generalized Anxiety Disorder , which consisted of 7 symptoms measured . Participants were asked how often they were disturbed by each symptom during the past two weeks. The response choices were "not at all," "several days," "more than half a day," and "almost every day," each getting a score of 0, 1, 2, and 3. A score of 10 or more was the cut-off point for identifying anxiety. All these instruments have been validated for adolescents in Indonesia. Social Media Exposure was measured by asking how often respondents during the past week were exposed to news and information about COVID-19 on media social, such as Facebook, Instagram, Twitter, YouTube, WhatsApp, and others. Response options were "less", "often" and "frequently". Social Media Exposure was divided into three categories, namely less , sometimes , and frequently . Health condition was divided into three categories, namely Not Good , Good , Very Good . Perception consists of vulnerability, the seriousness of the impact, the benefits of preventive action, barriers to prevention, and self-confidence to take preventive action . The perception was measured using instruments developed by researchers using the theoretical framework approach Health Belief Model. All of the instrument was tested for validity and reliability. Multivariable logistic regression was used to identify the relationship between exposure to social media with mental health problems and adolescent perceptions after controlling for covariates. --- Statistical analysis The mean and standard deviation were used to analyze numerical data, and the percentage was used to determine the frequency. To determine the correlation between variables and correlation analysis was used non-parametric statistics. A p-value <0.05 was considered as statistical significance. SPSS version 26 was used as the software for the analysis. Social Media Exposure: Less ; Sometimes ; Frequently . Health condition: Not Good ; Good ; Very Good . --- RESULTS Table 1 shows the demographic of respondents and social media exposure. This study involved 1357 respondents, who were drawn randomly from the entire population in Central Java Province. Out of 1357 respondents, the mean age was 19.1 ± 2.5 years . The proportion of "less", "sometimes", and "frequently" SME was 38.5% , 54.4% , and 7.10% , respectively. As shown in Table 1, it is known that the largest proportion of men "sometimes" of SME was 49.9% , and the largest proportion of women also "sometimes" of SME was 56.3% . Based on age, the highest percentage of respondents was 19-24 years old, namely 57.4%. Among respondents aged 19-24, the largest proportion "sometimes" of SME was 57.8% . Based on education, the highest percentage of respondents had middle education, namely 90.3%. For respondents with middle education, the largest proportion "sometimes" of SME, was at 55.5% . Based on the marriage status, the highest percentage of respondents were unmarried, 99.1%. For no marriage respondents, the largest proportion was "sometimes" of SME, 54.5% . Based on occupational, the highest percentage of respondents were students, 93.5%. Among student respondents, the largest proportion "sometimes" of SME was 54.4% . Based on region type, the highest percentage of respondents who live in rural areas was 69.9%. Among respondents who lived in rural areas, the largest proportion was "sometimes" of SME, which was 53.3% . Based on health conditions, the highest percentage of respondents declared a good condition, namely 48.7%, and in very good condition was 47.5%. For respondents with good conditions, the largest proportion "sometimes" of SME was 55.7% . Based on the statistical analysis results, it was known that the factors associated with the frequency of use of social media were age, education, and health conditions. Respondents, other than students and workers, used social media to entertain or seek information. Most students used social media to do assignments or take online classes during the analyzed period. For workers , most social media use was to do their work or assignments because they must work from home during this COVID-19 pandemic. Teachers must continue to teach their students through social media or special programs according to the provisions of the agency. Table 2 shows the prevalence of perception during the COVID-19 pandemic and related factors. Of 1357 respondents, those with good perception were 94.8 ± 1.2% , and those with bad perception were 5.2 ± 1.2% . In theory, what affected a person's perception were gender, age, education, marriage status, occupation, region type, and health condition. Of the 1287 respondents who had good perceptions, 71.3% were women. The risk of women having a good perception was 2.5 times greater than men's. Among respondents with good perceptions, 36.1% were aged 15-18, and 57.7% were aged 19-24. Based on education, respondents with good perceptions, 90.1% had middle education, and based on marriage status, respondents who had good perceptions, 99.2 was no marriage. Unmarried respondents were at risk of having a good perception 37.6 times greater than respondents who divorced/widow/widower . Based on occupation, respondents with a good perception of 93.6% were students, and based on region type, respondents with a good perception of 70.5% lived in rural areas. People who lived in risky rural areas had a good perception of 1.7 times greater than respondents who lived in urban areas . Based on health conditions, respondents with good perceptions were 49.3% in good condition and 47.3% in very good condition. People who were in good condition were at risk of having a good perception of 0.3 times than respondents who were in not good , and people who were in very good condition were at risk of having a good perception of 0.4 times than respondents who were in not good . Based on the statistical analysis results, it is known that the factors related to perception were gender, marital status, type of residence, and health conditions. The perception that a person has will influence the person's attitude and behavior. People with good perceptions during the COVID-19 pandemic will think positively and always try to take COVID-19 prevention measures. In addition, a good perception will also make a person always think positively, which could increase body immunity . Table 3 shows the prevalence of stress during the COVID-19 pandemic and related factors. Of the 1357 respondents, those with low stress were 74.5 ± 2.3% , and those with high stress were 23.2 ± 2.3% . In theory, what affects a person's stress level were gender, age, education, marriage status, occupation, region type, and health condition. Of the 1011 respondents who had low stress, 67.5% were women. The risk of women having low stress is 1.7 times greater than men's . Among the respondents with low stress, 39.5% were aged 15-18, and 53.2% were aged 19-24. Based on education, respondents with low stress, 89.2% had middle education, and based on marriage status, respondents with a good perception were 98.9 was no marriage. Respondents aged 10-14 had a 2.4 times greater risk of low stress than those aged 19-24. Based on occupation, respondents who had low stress, 92.7% were students, and based on region type respondents who had low stress, 71.3% lived in rural areas. Based on health conditions, respondents with low stress were 44% in good condition and 53.4% in very good condition. People who were in good condition were at risk of having low stress 0.5 times than respondents who were in not good , and people who were in very good condition were at risk of having low stress 0.2 times than respondents who were in not good . Based on the statistical analysis results, it was known that the factors associated with the stress level in society are gender, age, and health conditions. A person's stress level is influenced by the person's ability to manage thoughts and cope with the stress experienced . A person's emotional maturity is influenced by sex, age, and health condition. This emotional maturity will affect people's ability to stress coping strategies. People with highstress levels during the COVID-19 pandemic will cause their immune conditions be suppressed, thereby increasing the risk of contracting COVID-19 . Table 4 shows the prevalence of anxiety during the COVID-19 pandemic and related factors. Of the 1357 respondents, those with low anxiety were 70.2 ± 2.4% , and those with high anxiety were 29.8 ± 2.4% . In theory, what affects a person's stress level were gender, age, education, marriage status, occupation, region type, and health condition. Of the 953 respondents with low anxiety, 68.6% were women. Among the respondents with low anxiety, 38.2% were aged 15-18, and 55.5% were aged 19-24. Based on education, of respondents who had low anxiety, 89.7% had middle education, and based on marriage status, respondents who had low anxiety 99.2 was no marriage. Based on occupation, respondents with low anxiety 92.9% were students, and based on region type, respondents with low anxiety 70.5% lived in rural areas. Teachers were at risk of having 0.75 times low anxiety than respondents who are not yet working . Based on health conditions, respondents with low anxiety were 44.2% in good condition and 53.1% in very good condition. People in very good condition were at risk of having low stress 0.3 times more than respondents who were not good . Based on the statistical analysis results, it is known that the factors associated with community anxiety are occupation and health conditions. A person's anxiety level is influenced by the person's ability to manage thoughts and cope with the stress experienced . The ability to manage thoughts can be influenced by occupation and health conditions. This emotional maturity will affect people's ability in stress coping strategies, thereby reducing the risk of experiencing high anxiety. People with high anxiety during the COVID-19 pandemic will cause their immune conditions to be suppressed, thereby increasing the risk of contracting COVID-19 . There was a significant difference in SME between people with poor perception and those with good perception . As shown in Table 5, it was known that the largest proportion of people with "less" good perception of SME was 57.1% , while the largest proportion of people with "sometimes" good perception of SME was 55.6% Social media has become an integral part of daily life for adolescents in Indonesia . During the pandemic, social media has played a significant role in disseminating information about the virus and providing a platform for individuals to connect and share their experiences. However, social media exposure during the pandemic negatively affects adolescent mental health . Social media exposure during the pandemic is associated with increased symptoms of anxiety and depression among adolescents . The research of this study was in line with a study conducted by Budury et al. that social media exposure was significantly associated with increased symptoms of anxiety and depression among Indonesian adolescents during the pandemic. The study also found that social media exposure was associated with decreased life satisfaction and social support, which are important protective factors for mental health. A study conducted by Reihm et al. found that adolescents who spend more than three hours per day on social media may be at heightened risk for mental health problems, particularly internalizing problems. --- CONCLUSION From the results of the study obtained data that are associated with social media exposure in adolescents in Central Java during the pandemic are age , education , and health conditions . Then social media exposure in adolescents during a pandemic affects perception and stress but does not affect depression . With a variety of psychological conditions that could have arisen in adolescents during the COVID-19 pandemic, the regulation of the use of social media is important to note. Adolescents' perceptions and mental health were significantly affected by social media exposure. The study found that increased social media use was associated with increased anxiety and stress. Additionally, exposure to COVID-19-related news on social media was associated with increased stress and anxiety. Publisher's Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. Open Access: This article is distributed under the terms of the Creative Commons Attribution 4.0 International License , which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author and the source, provide a link to the Creative Commons license and indicate if changes were made. --- CONFLICT OF INTEREST The authors declare no conflicts of interest.

Context: Many adolescents exposed to social media information during the coronavirus pandemic (COVID-19) appeared in Indonesia, especially in Central Java. Aims: To assess adolescent perceptions of COVID-19 and the prevalence of mental health problems and examine their relationship with social media exposure. Methods: This study used a cross-sectional design in adolescents aged between 10-24 years in Central Java province and was conducted from May to June 2020. The snowball sampling technique was used to distribute questionnaires online via WhatsApp. The estimated number of samples who filled out the questionnaire was 1357 respondents. The instruments used in this study were WHO-Five Well-Being Index (WHO-5), Generalized Anxiety Disorder (GAD-7), and a questionnaire developed by researchers using the theoretical framework approach Health Belief Model. Multivariable logistic regression was used to identify the relationship between social media exposure with mental health problems and adolescent perceptions after controlling for covariates. Results: From the results of the study obtained data that were associated with social media exposure in adolescents in Central Java during the pandemic as age (p = 0.0001), education (p = 0.025), and health conditions (p = 0.016). Then social media exposure in adolescents during a pandemic affected perception (p = 0.0001) and stress (p = 0.005) but did not affect depression (p = 0.108).The findings indicated that social media exposure had a significant effect on perception and stress levels. Adolescents who had higher exposure to social media during the pandemic experienced changes in their perception and reported higher levels of stress. However, the study did not find a significant association between social media exposure and depression among adolescents during the pandemic. This suggests that while social media exposure may influence perception and stress, it may not have a direct impact on depression.

Introduction Infertility or childlessness is a global reproductive issue for both sexes yet often neglected and not discussed in public. It is generally believed that more than 70 million couples suffer from infertility worldwide [1]. In Sub-Saharan Africa, the prevalence differs widely from 9% in the Gambia [2], 21.2% in northwestern Ethiopia [3], between 20 and 30% in Nigeria [4,5] and 11.8% among women and 15.8% among men in Ghana [6]. There are varying opinions on the definition of infertility. The lack of consensus on the prevalence of infertility is a consequence of differing definitions of infertility, the varying periods of time over which it is studied, and a failure to differentiate analytically between voluntarily childless and involuntarily childless [7]. However, the Ghana Demographic Health Survey reports that voluntary childlessness is not common in Ghana, and currently married women with no live births are likely to be those in relationships with fertility problems [8]. Infertility has been defined as failure to conceive after one year of regular unprotected sexual intercourse in the absence of known reproductive pathology [9]. However, epidemiological studies have revealed that in a normal population of heterosexually active women who are not using birth control methods, 25% will become pregnant in the first month, 63% within 6 months, and 80% within one year. By the end of a second year, 85% to 90% will have conceived [10]. Because some couples, who are not infertile, may not be able to conceive within the first year of unprotected sex, the World Health Organisation therefore recommends the epidemiological definition of infertility, which is the inability to conceive within two years of exposure to pregnancy [9]. Infertility may be primary or secondary. Primary infertility refers to infertility of women who have never conceived and secondary infertility refers to infertility of women who has conceived at least once before. The use of the ability of the female to conceive as a measure to differentiate between primary and secondary infertility is however problematic as it places couple infertility on the doorsteps of the female partners. In the African culture, the true meaning of marriage is only fulfilled if the couple conceives and bears children [11]. Africans consider their child to be a source of power and pride, and children act as insurance for their parents in old age. The most important aspect of bearing children is an assurance of family continuity. Anthropological and sociological studies bear testimony to the considerable suffering associated with involuntary childlessness due to negative psychosocial consequences such as marital instability, abuse and stigmatization [11,12]. A study among women seeking infertility treatment in Southern Ghana revealed that infertile women used their internal coping strategies by keeping their fertility problem to themselves as a result of the stigma associated with it whilst others coped by drawing on their Christian faith [13]. There are varying findings on the contribution of the various sexes to couple infertility. An increasing body of social science and biomedical evidence suggests that nearly 40-50% of infertility is attributable to problems suffered by men. The underlying cause of infertility may be a male factor , a female factor or a combination of problems [14]. Another study states that about 50% of infertility is due to the female, 20-26% to the male and 26-30% is unexplained , but these findings are all from non-African populations [6]. Studies have revealed that the most common cause of male infertility is due to a problem in the sperm production process in the testes. About two thirds of infertile men have sperm production problems. Low numbers of sperm are made and/or the sperm that are made do not work properly [15]. A number of factors can disrupt the production of sperm including undescended testis, infections such as mumps, heat, sperm antibodies, torsion, varicocele, drugs or radiation damage. Blockages in the tubes leading sperm away from the testes to the penis can cause a complete lack of sperm in the ejaculated semen. This accordingly is the second most common cause of male infertility and affects about three in every twenty infertile men, including men with the common problem of having an earlier vasectomy [15]. In some men, sperm antibodies can develop which can lessen sperm movement and block egg binding during fertilisation. About one in every sixteen infertile men has sperm antibodies [16] and this may cause male infertility. In women, poor quality eggs may cause infertility, a blocked fallopian tube could prevent the egg and sperm from uniting, or the woman may not ovulate regularly -a problem that is sometimes results in irregular menstrual cycle [17]. There is a wellestablished link between a woman's age and infertility. Women over age 35 years have an estimated 50 percent chance of becoming pregnant naturally [16]. As women age, their fertility is affected by the quantity and quality of their eggs. In reality, the number of eggs available in the ovary gradually declines. As menopause approaches, an increasing number of cycles are not ovulatory, and therefore unable to result in conception [17]. Moreover, an older woman's eggs are most susceptible to chromosomal changes that may produce abnormal embryos [18]. A study in Bawku in Northern Ghana identified tubal damage, male factor, anovulation, uterine factors and unexplained as the five main causes of infertility. This study also states that about 20% of infertility in Bawku in Northern Ghana was due to male factors [19]. In spite of these statistics, women still endure the worst of the blame for infertility problems. Leading male Obstetricians and Gynaecologists are often quick to attribute couple infertility to female factors and can therefore be blamed for the belief that is widely held [20]. Infertility is socially constructed in many cultures that is, men and women are meant to become parents and that women are especially socialized to become mothers [21]. Infertility is a problem both medically and socially in Ghana. Medically because there are reports of high prevalence of infertility among couples with inaccessible fertility services. Infertility is also a problem socially because we live in a society where womanhood and manhood are generally tied to motherhood and fatherhood respectively. Despite these problems of infertility, not much has been done to understand community's perception of infertility and childbearing in Ghana. It is in the light of this that this study examines community perception and understanding of childbearing and childlessness in Northern Ghana. --- Materials and Methods --- Ethics Statement Ethical approval for the study was received from the Ghana Health Service Ethics Committee. During the approval process for the study, the committee was explicit about the need to maintain confidentiality and anonymity whilst emphasising on the need to obtain verbal or written consents. In line with the approved procedure of obtaining consent for the study, verbal or written consent were obtained from participants. Verbal consent was obtained from participants who had difficulty reading the consent form and those who opted to give verbal consent. To those who gave verbal consent, the researchers read and translated the consent form into the preferred language of the participant. They were further made to recommend a neutral member of the household to act as independent witness in the consent process. A cover sheet containing the demographic information except the names and locations that were coded and kept separately was used to document those who gave verbal consent. This procedure for obtaining verbal consent was approved by the Ethics Committee of the Ghana Health Service. To ensure confidentiality of participants that gave written consents, codes were used on the form instead of their names. The specific locations of the participants were also not reported as this could lead to easy identification of the infertile couples that took part in the study. In addition, only codes were written on interview transcripts. --- Study Area The Upper West Region is the smallest Region in Ghana with a population of 702,110 with 989 settlements. The region covers a total land area of 18,476 km 2 , with a population density of 32 persons per square kilometre [22]. The region is divided into eleven administrative districts and the people speak three main local languages namely: Wali, Dagaare or Sisali. The people of Upper West Region practice patrilineal mode of inheritance and is a typical patriarchal society with male dominance in decisionmaking. Polygamy is a common practice both by members of the Islamic community and those who profess African Traditional Religion. In all, there are sixty-five sub-districts, five district hospitals located in four districts, a regional hospital in the regional capital. The region has a fertility rate that is the average number of children per couple of five, which is above the national average of four [22]. --- Data Collection In-depth interviews , Focus Group Discussions and Key Informant Interviews were the main data collection methods used in the study. Four trained data collectors were used in the study. These research assistants were put in groups of two, a male and a female. Male interviewers were assigned for male respondents and female interviewers were assigned for female respondents with children. Feminist theorists argue that the positionality of the researcher: gender, class, and race affect all aspects of the research process, from the framing of the research question to the analytical approach as one's own social location influences the full scale of research choices [23]. In-depth interviews are social interactions [24] and hence race, class, and gender inequalities are inherent in these interactions and can therefore affect the results. Hence, interviewer/interviewee homogeneity was adopted to overcome this challenge of the effects of differences in gender between the interviewer and interviewee. Concerning couples with infertility , the principal researcher who supervised the data collection carried out the interviews for both partners. This was done because it was anticipated that some respondents might break down emotionally in an attempt to narrate unpleasant past experiences. The principal researcher's background in nursing and psychology therefore put him in a better position to offer counseling for those who break down in the course of the interviews. Interviews were either conducted in English or local language depending on the language the respondent was comfortable with and the duration of the interviews was between 30-45 minutes. The interviews were tape-recorded with the permission of the participants as well as note taking. In polygamous family, the male's partner and the individual wives were interviewed starting from the most senior of the wives in line with community's norms. Younger wives in some instances were interviewed first upon the request of the most senior wife in consultation with the husband. The same codes were written on the male and female interview guides and each recording was started by first mentioning the code on the interview guide to ensure data collected could be analysed as belonging to a couple for comparison to be made. The codes also differentiated between the male and female partners that constituted a couple. Field notes were written immediately after each interview. The field notes covered the initial reactions to the interview, including the first analytical reflections from the interview content, and any useful observations that would not be captured by digital recording. This covered the demeanor of the respondents, his or her body language and mood, and any informal conversation that took place before or after the interview. --- Selection of Study Participants An informal discussion with the community members and Community Health Volunteers was initially used to identify couples without children. After interviewing such couples, snowball techniques were then used to identify other couples with infertility in that community. The study considered a married union as a unit of analysis irrespective of the number of partners involved. Hence, fifteen units of childless marriages were interviewed. However, three of the childless male partners in the study were married to two wives. This therefore increased the number of childless individuals in the study to thirty-three comprising of fifteen childless males and eighteen childless females. The age for the males' partners ranged between 35-63 years as against a female age range of 28-52 years. The couples were married for between 3-25 years and included couples with both primary and secondary infertility. Couples with children were also purposely selected and interviewed. The couples were recruited from both urban and rural settlements where childless couples lived. In all, forty-five married unions were recruited and interviewed within the neighbourhood where couples without children lived. However, four of the males were married to two wives whilst one was married to three wives increasing the number of women with children interviewed to fifty-one with ninety-six participants with children. The daily analysis of the transcripts made the research saturated with Forty-five married unions. Collecting enough data to the point of saturation adds to the credibility, dependability of the study and transferability of the study results to other studies on infertility and childbearing. --- Focus Group Discussion Three Focus Group Discussions were organized. One FGD for women without children as their male counterpart refused to take part in FGD. Their refusal to take part in the FGD was because infertility was perceived to be the inability of women to bear children and therefore not suitable for males to discuss. The researchers organized two FGDs for community members selected from where infertile couples were resident, one for males and one for females to elicit normative ideas on childbearing and childlessness in the communities. The discussants drawn from the communities were all married adults with children. The FGDs lasted for between 60-90 minutes and were conducted in the evenings as participants preferred to do that after the day's work. All Focus Group Discussions were conducted in Dagaare however; participants who wished to contribute in English at any point in the discussions were allowed to do that. Such contributions were translated immediately to Dagaare for participants who did not understand English. This was done to ensure that such contributions conformed to the normal practice in the community. All participants were allowed to give their view on any subject raised before progressing to another theme. --- Key Informant Interviews Eight Key Informants Interviews were conducted. Two Gynaecologists who take care of infertile couples were interviewed as Key Informants. An Islamic scholar was also interviewed to provide expert view on infertility and childbearing in the Islamic Community. A female Christian leader was also interviewed as she provided support for infertile females. Two traditional medical practitioners who also provided care for infertile couples were further interviewed. A manager of the National Health Insurance Scheme and a manager with Private Insurance Company were equally interviewed as Key Informants to provide information on the infertility related insurance policies in Ghana. --- Data Analysis The taped interviews were transcribed verbatim and the resulting texts analysed by using thematic analysis. An attempt was first made to extract broad themes from the transcripts and then progressed to identifying coded themes. In establishing themes, considerations were given to statements of meaning that were present in most of the relevant data. In an attempt to ensure, the credibility of the findings independent coders were used to verify or corroborate the themes extracted from the data. This allowed the researchers to progressively focus the interviews and observations, and to decide how to test the emerging conclusions. Individual and comparative analysis of the response of couples was carried out. The transcripts were entered into QSR Nvivo 8ß for analysis. We developed a codebook based on the major themes of the study. The major themes were transformed in tree nodes and free nodes. The authors based on the codebook developed and verified independently coded texts from the transcriptions. The emerging themes and sub-themes were identified and written out in the results. Quotes from respondents were used to support the emerging patterns of concepts from the data. --- Results --- Reasons for Preference for Children The pinnacle of reasons for preference for children was to maintain the family lineage and inheritances. Children are often given the family name of the father as the communities practice patrilineal mode of inheritance and they are supposed to marry in future and name their children after the family name. This according to respondents was done to ensure the perpetual existence of the family lineage. Examples were given of great families whose lineage had wiped out because they were unable to give birth to more children and majority gave birth to female children who got married and named their children after their husbands. Life without children was perceived not to be worth living as there will be nobody to inherit the properties of the deceased and not all efforts by such individuals are ever appreciated in the community. ''Children are supposed to maintain lineage and inherit your property…we are suffering on earth because of our children'' -. A second reason for procreation was for assistance at home and in the farm. The number of children most especially male children is an indication of the worth of the couple as it implies more hands in the farm. More hands in the farm means more revenue for the family as extra farm produce could be sold to generate income for the family. This further means more savings and security for future. This practice has become necessary especially as discussants stated, the break in the extended family system and increasing nucleation of families has made the quest for own children very significant as they provide security in old age. ''I use to plough only 5 acres of land but with my 5 children, I am now able to plough over 15 acres improving on the family income''-. The third reason for procreation, which was stated emphatically by the respondents, was to obey Gods words as the Bible and Koran enjoin Christians and Moslems to multiply and fill the earth. Respondents especially believed that failure to beget children was against the holy books and such a family will never receive the blessings of God. ''The bible says in genesis that we should multiply and fill the earth to ensure the continuous existence of the earth''-. Children are also source of joy, companionship and respect for community members. Children console their parents and are a symbol of achievement for couples. At old age, grandchildren act as companions for grandparents and this was believed to make them happy and prolong their life. ''Children make you happy, it is the greatest achievement on earth…we respect people with children in this community especially male children''-. Clearly from the respondents, one of the reasons for procreation is to have a befitting funeral, as there are significant difference in the performance of funeral rituals for people with children and those without children. Adult without children are more likely to be interred earlier in traditional communities and their funeral not attracting the needed attention it deserves as compared to adults with children. This is even more distinct for females than males without children. This is because of women traditional role to beget children to the family of the in-law. This reproductive role becomes the exclusive right of the husband following the acceptance of the bride wealth and not having biological children implies that the woman has failed in a fundamental way resulting in a loss to the family of the husband. In addition, childbirth is the culmination of a woman's rites of passage to adult womanhood. Without a live birth, she remains in an infinitesimal state where she is neither a man nor fully a woman. The rituals performed for a deceased adult without children were comparable to that of a child. ''Adults without children do not have funerals in this community as they are treated the same way as the death of a child''-. --- Knowledge on Causes of Infertility According to respondents, infertility was caused by both biological and social factors. The biological causes were more pronounced among the urban and educated residents whilst the rural communities attributed infertility principally to social factors. The highest biological factor that has been blamed for infertility among females was previous use of contraceptives. This was also directly attributed to past promiscuous lifestyle of the woman as the contraceptive were used to prevent unwanted pregnancies. ''Women who were prostitutes during their youthful ages, used contraceptives to prevent pregnancy…they are the people who by all means become infertile in future and worry their partners. I know of a friend who used those family planning methods when we were growing up and is now hopping from hospital to hospital looking for a child''-. The believe in contraceptives as a cause of infertility was unanimous for both male and female participants and couples strongly believed the use of contraceptive could result in them becoming infertile. Some infertile women even attributed and strongly believed that they were currently having difficulty in becoming pregnant because they had used contraceptives in the past to prevent unwanted pregnancies since abortion was illegal in the country. Males in FGD alluded to this belief and indicated that contraceptives were a major enemy in the community because it use encourages promiscuous lifestyles for the youth and give trouble to their male partners in future. ''My wife made a mistake and took those drugs , and it took her several years to become pregnant again. This almost resulted in a divorce but I have since warned her never to take those drugs again''-. Sexually transmitted infections , blocked fallopian tubes and uterine fibroid were also mentioned as female factors that can cause infertility. Knowledge on the relationship between STIs and blocked tubes was high as it was unanimously agreed in FGD and cited in in-depth interviews. Participant mentioned gonorrhoea, syphilis and Chlamydia infections as common causes of infertility in the community. ''My wife was told that her tubes were blocked and that was why we could not get a child, when I heard this I immediately suspected she has had gonorrhoea before''-. Abortions in all forms are also believed to cause infertility but most especially those conducted by unqualified individuals. To community members, all women are born with a fixed number of children to conceive and when these children are aborted, and then the consequential effect is infertility. However, males are believed not have such fixed number of children and are therefore capable of producing uncountable number of children. Safe abortion in this context refers to an abortion that has been carried out by a qualified person using appropriate equipment and in a place designed for such procedures. Unsafe abortion, which they believe, is what is performed by quacks in the community and the use of herbs. ''Some women throw all the children they were supposed to give birth into the gutters through abortions''-. There was a paradox concerning Female Genital Mutilation and infertility. As FGM was mentioned as capable of causing female infertility, an ''extra-germination'' of the clitoris was also believed to cause infertility. As to which size of the clitoris was considered normal, there was no consensus but it was generally believed that the gods give direction when consulted by the healers. Therefore, the gods will often determine the ''extragermination'' in which case it has to be excised by the traditional healer after which herbs are applied. ''Extra-germination of the clitoris can cause infertility and in such a case you cut off that extra germination''-. ''….My wife bled highly when a part of her clitoris was excised by a traditional medical practitioner as treatment for infertility''-. Contrary to the female factors that were spontaneously mentioned in IDIs and FGDs, male factors were not mentioned until when prompted by the researcher. Males especially attributed infertility to lifestyles such as intake of excessive alcohol and smoking, though taking of alcohol was perceived to be socially acceptable, smoking was generally believed to be alien to the culture and described as Western culture. ''Too much intake of alcohol and smoking can cause infertility….It leads to weak penis and impotence''-. There were varying views on which alcoholic beverages and what amounts could be described as too much. As some participants stated that all alcoholic beverages could cause infertility when not taken in moderation, other believed that bottled drinks were more accountable whilst the local alcoholic beverage pito was even considered to increase both sperm production and potency and could be used to treat infertility. ''Pito can increase sperm production and that is why it is sometimes required that you add pito to the herbs in the management of infertility''-. Another school of thought was that bottled alcoholic beverages that are believed to enhance sex and increase appetite could cause infertility. Calls were even made to ban the production and advertisement of all such alcoholic beverages as they have become so common and easily accessible even by minors, participant emphasised. ''This bottled drinks that can increase erection can cause low sperm production and cause infertility. Any man who relies on drinks to perform will not be able to impregnate his wife or give birth to weak children''-. ''You see, our ancestors did not have a problem with childbearing, because they took things that were indigenous, but what do we see today, different varieties of foreign foods in our markets that we are blindly eating….These are responsible for some of the funny conditions we are seeing today''-. Female participants however mentioned watery sperm and inability of a male to sustain an erection as common causes of male infertility. References were also made to the use of aphrodisiac preparations as they can render a man impotent. ''Some males produce watery sperms and their penis is usually not strong enough during sex to impregnate a woman that is why we describe such people as having a dead penis''-. The belief in supernatural causes of infertility was widespread and consistent. The belief that some women are witches, and that either curses can be placed on them or their associates was consistent. Such women are believed to live for longer years because they are capable of exchanging their death with the death of children in the community. Such old women are abandoned by their relatives and hooted at when seen in public and are not allowed to come close to children, as they are believed to be capable of bewitching them. Children are specially socialised to run away upon seeing such old women in the communities and some of them are beaten mercilessly sometimes when seen in public gatherings. ''A childless woman after killing all the children in her womb recently took the head of a child in my community and was beaten until she returned the head. She has subsequently been banished from the community''-. Though participants also agreed that some men could be wizards and curses invoked on them, the penalties for such infractions with the gods appear not to include infertility for men. However, there was consensus on wizards also bewitching children but it was a common phenomenon with females. ''Men who are wizards at old age use their witchcraft to protect their family''-. One of the main and common social causes was described as a pledge made to the gods by people to sacrifice their manhood or womanhood for wealth referred to as plumma or donpuli. This pledge can only be reversed by undergoing some rituals, which is often prescribed by the soothsayers or traditional medical practitioners. However, the fear of losing one's properties and becoming impoverish discourages people from reversing such a pledge. To be able to maintain the wealth and bear children, the individual has to perform several rituals which are always very expensive and not within the ability of infertile couples. Such people are reported to be exploited by healers. ''Some people exchange their children with wealth but after coming to earth to see people having children, they become interested in children''-. ''I have sacrificed several animals in an attempt to reverse a pledge I was told I made to the gods to be rich yet the traditional healer is still demanding more because he has seen that am desperate to have a child''-. Breaking the codes of marriage was also widely held belief as a cause of infertility. The gods and ancestors were believed to be ''supernatural policemen'' who are capable of rewarding couples with children and punishing those who break the codes of marriage or visiting them with infertility. ''The gods and ancestors who are the custodians of this land can show their disapproval of the conduct of the marriage couple by making them infertile''-. Male couples who are currently not seeking biomedical management tended to attribute their infertility to female factors whilst their females' counterparts also blamed the males for their inability to make them pregnant. This was a basis for one to prove his or her fertility by engaging in multiple sex with other partners with the hope of either becoming pregnant or making another woman pregnant. In a response to a question on why it was difficult for them . The female respondent in an individual interview stated: ''My husband is unable to make me pregnant, for me am fertile; my mother gave birth to eight of us so how can I be infertile''-. The male partner of this childless female also stated in an IDI, ''I think my wife is barren because I have made a girl pregnant before when we were growing up''. Another supernatural cause of infertility, which was widely reported and justified by Christians was masturbation, which was described as ''male abortion''. Masturbation was perceived to be sexual immorality and a sin. Respondents generally believed that masturbation could attract a punishment of infertility as it is deemed an immoral act. Participants believed that masturbation was a form of abortion as sperms were believed to be pre-formed babies that are put in woman's womb to incubate until the woman brings it out during delivery. There was therefore no distinction between a woman aborting a fertilized ovum and a man masturbating and discharging spermatozoa. ''Sperm are human beings and should not be masturbated and discarded…in fact it is male abortion and God is against abortion''-. --- Infertility and Sex Preference Concerning the definition of infertility, three main themes emerged in interviews and FGDs each without time limit. One related to the medical definition and the others related to the desire to have many children and sex preferences. To the participants, infertility is not only defined as the inability of a couple to beget children. It also includes inability to beget male children or inability to comply with society's norm of having many children. The ideal numbers being a function of the desire of the couples but in many cases about five is preferred. ''Infertility is the inability to give birth to the number of children that you prefer and most especially male children''-. Generally, it is believed that it is the responsibility of the male to maintain the lineage of the family. Hence, male partners will go to all extent to have male children including marrying multiple wives or continue to give birth until such a time that a male child is born. The significance of male children is so intense that couples without male children are treated in the same way as those who do have children at all. ''I am married to three wives because I thought my first two wives were unable to give me male children. When my third wife gave birth recently, a friend came to congratulate me, after which he asked whether this time my wife had given birth to a human being. With my so many female children, I will be treated like someone who has not given birth at all when I die''-. However, the three wives of the man believed that they were only incubators and therefore only brought out what the man has put in there during delivery and could not be blamed for their inability to give birth to male children. ''I told my husband that he cannot give birth to male children but he refused and went in for a second and third wife, but see we are all giving birth to females'' -. This is because it is the responsibility of deceased's male child to go to the bush to harvest a special stick known as Kpiendaa, which is to be kept in a special room reserved for ancestor to symbolize the deceased acceptance in the ancestral world. --- Help-Seeking Behaviour of Infertile Couples A consensus reached by the individual interviews and focus group participants was that the treatment of infertility in the community is usually directed specifically at women and that most people use three treatment outlets: churches , traditional healers and hospitals . However, there was no agreement between and within the groups on which of the three methods that people prefer most. Nonetheless, there was a strong sense that people often use the three treatment methods in combination and in sequence. The first method chosen is often determined by the perception of the couple regarding the causes of the infertility. As most people are deeply convinced of the supernatural causes of infertility, it is not surprising that infertile people often patronize traditional healers and religious leaders very early. Orthodox medical practitioners are often consulted later when religious and traditional methods have failed to provide a solution to the infertility. ''I am using both traditional and orthodox medicine, but I first went to the herbalist''-. The traditional remedies ranges from making sacrifices of animals to pacify the gods or ancestors for wrongdoing to taking of local concoctions prepared from herbs. They may also be requested to perform some rituals at a place where two, three or four roads cross or at where ants live and this was the pronounced ritual. Another ritual that is often practiced is the wearing of prescribed costume mostly by the female partner, which is removed at the entrance to their residents. Another treatment, which was mentioned both in individual interviews and in focus group discussions was the excision of some part of the clitoris, which is believed to cause infertility in women. It is belief that an 'extra-germination' of the clitoris was capable of causing infertility in women. The gods are believed to give direction when consulted by the healers. ''….My wife bled highly when a portion of her clitoris was excised by a traditional medical practitioner as treatment for infertility''-. Churches also prescribed number of days of fasting and prayers and making of some special offerings to the church. References were also made of washing of women genitalia with holy water and anointing oil prepared and blessed by the faith healers. ''I was given some anointing oil to smear on my private part before having sex with my husband''-. Another important consideration in the choice of a practitioner is the issue of privacy. As infertility is considered a very sensitive issue in the community, people often seek out practitioners who will be able to keep their infertile status a secret. Both individual interviews and Focus-group participants had different views about which practitioners; orthodox, traditional or spiritual, would maintain the most confidentiality. Although it was generally agreed that traditional healers are capable of assuring the most confidentiality, nevertheless, the view was expressed that traditional practitioners often exploit women consequently, either financially or sexually. This did not however reduce the power of the treatment provided by the traditional healers, but was considered ''a good price to pay'' to get a child to make you happy and save you from ridicule and your marriage. In general, the modus operandi of traditional medical practitioners was believed to maintain better privacy than that of biomedical practitioners. Their facilities are often situated at the outskirt of the community where community members hardly go except for those with problems who have gone to consult. ''Traditional medical practitioners get their powers from divine sources and are therefore compelled to maintain their practice in secret otherwise they loss the powers and this is different from the hospital where nobody cares about privacy''-. Many participants also recognized the importance of going to the hospital for tests and firmly believed that the doctors can often determine the exact cause of the infertility, and prescribe drugs to treat it. However, the medical approach is often not used immediately since biological factors are not acknowledged as prominent causes of infertility in the community. Many people believe that Western medical treatment can only help if there is a biological cause of infertility. If people believe that the infertility was caused by a curse or spell, or by gods and ancestors, they will seek an appropriate solution, which may not include the biomedical practitioners. They may seek help from traditional doctors or spiritualists first, and come to the medical doctor later. This will be done for additional help or to treat a problem caused by the herbal treatments. The social stigma attached to infertility problems means that people are cautious of revealing their problems, and the hospital environment may be too open to accommodate such secrecy, participant's claim. Participants also described assigning special days and clinics for infertility in hospitals as broadcasting to the whole world about the infertility status of the clients and therefore not confidential enough. ''You go to hospital, and they will announce that those with infertility should go this consulting room telling everybody of your problem. The next thing you see is that people are pointing fingers at you in town''-. Interestingly, either couples with infertility sought remedies individually or the women were the only people who reported to be seeking treatment especially in biomedical facilities. Males also tended to seek for treatment alone at the traditional medical practitioners and only invited their wives when both are to perform some rituals. It was also very interesting to note that when it is even established that the aetiological factors had to do with male factors, traditional medical practitioners still focused on women and they are still compelled to lead the rituals. ''It is we, the female who are always concerned about childlessness, because of the pressures we get from our in-laws, so we are the ones who look for treatment. Your husband may even refuse when invited by the medical practitioners, even when the cause is from your husband you are still supposed to do something as a ritual''-. --- Prevention of Infertility The prevention of infertility was also unanimous and education was seen as the core preventive strategy especially for the younger generation as many of the perceived causes are consequences of behaviours during youth. People commonly suggested that drugs used for abortion should be banned even though there were other herbs that were believed to be capable of inducing abortion, which are believed to be relatively safer than medical abortificients. The drugs used to cause abortion are sold around by individual in the community but not in the drug stores-. In several cases, the participants suggested that the use of dangerous drugs should be stopped. It was suggested in some cases that the ''doctors'' themselves were selling these drugs. When prompted on the people described as ''doctors'' it was discovered that the term was used to refer to people that sell drugs in the community and various workers in biomedical health facilities. Many of the participants expressed the wish that all abortions should be eliminated. When the issue of prevention turned to the use of contraceptives, respondents were split on the value of contraceptives: while often, people believed that contraceptives could prevent pregnancy, and thus prevent illegal abortion and the infertility that might follow; others felt that contraceptives themselves caused infertility, and that their use should be stopped. Some of these same people felt that women should be encouraged to bear the children if they became pregnant because of the fixed number of children every woman is destined to conceive and give birth. ''Those drugs that prevent pregnancy should be stopped, they also cause infertility and in this community, many women have complained that they are unable to conceive again after using those drugs''-. Another preventive strategy of future childlessness was education of the youth to desist from unhealthy lifestyles like excessive alcohol intake, use of sex enhancing drugs and smoking. A call was made on government to place a ban on the importation of such products. Supernatural causes are regarded as not within the control of the individuals hence difficult to prevent. However, respecting the norms of society and adhering to codes of conduct for marriage couples were central in preventing the social causes of infertility. ''Respect for societal norms will prevent infraction with the gods and ancestors who are capable of visiting infertility on couples''-. --- Discussion There was little diversity in the participants' definition of infertility. Many defined it simply as a woman's inability to bear children or inability of couples to have children. Focus-group participants generally recognized the concepts of primary and secondary infertility in addition to another form of infertility, which will be described as ''tertiary infertility''. The surname is very important in the culture of the people of Northern Ghana and since women who get married are, supposed to name their children after their husband, or use the husband's surname, male children symbolize the worth of the man [25,26]. Couples without male children can therefore be described as suffering from ''tertiary infertility'' as they are treated the same way as those without children. Primary infertility has been described as the inability of couples to conceive after two years of unprotected regular sex and secondary infertility as the inability of couples who have conceived before to beget children [9]. The desire to have male children poses a threat to family planning services as well as STI control programmes as males generally experiment with other women in an attempt to have male children [25]. The use of children as farm helps also deny children of formal education and a challenge to achieving universal primary education to achieve Millennium Development Goal 2. Knowledge on the biological causes of infertility was generally high and unanimous. The role of STI as a common cause of infertility for both males and females was undisputed. However, couple infertility was more attributed to female factors. Previous use of contraceptive was the mostly mentioned biological causes of infertility for females. Northern Ghana is a poverty endemic region in the country and therefore efforts are tailored towards reducing the fertility rate of couples. The perception that the use of contraceptives can cause infertility in females therefore poses a challenge to approach adopted by the Ghana Health Service in her reproductive health policy to prevent unwanted pregnancies and ensure good child and maternal health aimed at accelerating progress towards achieving the Millennium Development Goals, 4 and 5 respectively. Family planning providers must also be concerned about sexually transmitted diseases, induced abortion and infertility and they should be able to find ways to incorporate such concerns into their programs. An integrated reproductive health policy that encompasses family planning, prevention of STIs and infertility will give a holistic approach to problem solving in Ghana. Vertical and disease specific approach may undermine and neglect some equally significant social problems in community context. The belief that Female Genital Mutilation can cause infertility in females is of medical importance. FGM is a negative cultural practice that is commonly practiced among people from the northern part of Ghana [25,27]. The perception that FGM can cause infertility therefore implies that parents will resist the excision of the clitoris of their female children making the cultural practices outdated. At the opposite side of this is the belief that an extra-germination of the clitoris could cause infertility. Female genital mutilation may therefore be embedded in this belief and therefore a major challenge to the prevention of this unhealthy traditional practice. The worldview that violating of marriage code of conduct could result in infertility had some positive effects on marriage unions. Because of the supernatural supervision of marriage by ancestors and sanctioning of couples who violate the codes, married couples live within the societal prescription of the conduct of marriage couples. This creates harmony and prevents extramarital activities especially for female. This may therefore reduce incidence of STIs among couples. Cigarettes contain nicotine, which has received much attention for its interference with normal endocrine function, it has been shown to cause testicular atrophy, gonadal dysfunction, and male factor infertility by triggering testicular cytotoxicity [28] and this was cited as one of the causes of infertility in males. However, of medical interest is the role that masturbation places in infertility. Though it is unclear how masturbation can cause infertility, the community believed that masturbation was a form of abortion. The concept of male abortion to the best of the knowledge of the authors has never been reported previously, and this serves as a social factor that plays a very important role among communities in Northern Ghana. The study revealed that couples used three main medical outlets: spiritual, traditional medical practitioners and biomedical. Couples used these facilities concurrently or in sequence. This finding agrees with similar studies in Nigeria and South Africa. Several women with infertility consulted herbalist, apart from a witch doctor. Others visit churches for treatment because they perceive infertility as a misfortune, which God can redress [29]. Traditional health care was also identified as an important alternative source of understanding, coping and medication for health problems, including infertility in the Gambia [30]. The traditional medical practitioners were the most utilized outlet of care as infertility was given a more spiritual cause than medical and they were deemed to provide better privacy than biomedical facilities. Contrary to this finding, medical treatment was the preferred choice for many couples in an earlier study [31]. Confidentiality and ability to keep infertile status of couples as a secret determined the help seeking behaviour. Since many respondents believed that the environment and mode of operation of the biomedical health facilities was not private enough, it was generally not conceived as first line treatment. Traditional medical practitioners in this study are very instrumental in the diagnosing and management of infertility in Northern Ghana. It is therefore very relevant to integrate traditional medical practitioners into the primary health care system because of the multiple advantages it presents. A good collaboration among the traditional and orthodox medical practitioners will provide an opportunity for the training of traditional medical practitioners in current scientific knowledge and enhance inter-practitioner referral system, which can cater for both biological and spiritual aetiologic factors. However, the use of specialized clinics or designation of some days within a general hospital for some disease conditions is widely used in Ghana, the findings of this study point to the negative effects and perception among community members especially for disease conditions with stigma. The general believe in the community is that such clinics literally broadcast the disease conditions of clients and hence additional measures should be put in place to ensure privacy and confidentiality of clients. Both the use of donated egg and sperm were generally unacceptable among participants in this study. Adoption was viewed as service to God but not a remedy for infertility. A study also found negative feeling associated with adoption in a study of Hindu couples because it was perceived as highly visible indicator of infertility [32]. Donated eggs seemed to be marginally more acceptable than the use of donated sperm, which is seen as sexual in a way [30]. Some of the couples, who believed that their infertility was due to witchcraft, curse or any cause other than physical are those who sought for spiritual help. Some of the couples had sought help from spiritual churches and prayer camps. The study revealed that traditional medical practitioners prescribed rituals including wearing of some special costumes. In Nigeria, herbalists prescribe certain ritual or actions, such as the women bathing at night at a place where roads meet, or making sacrifices of food to evil spirits that may be causing the problem and leaving the items at the crossroad [4]. The findings of this study are therefore similar to the study in Nigeria. Of medical interest is also excision of some portions of the clitoris by some traditional medical practitioners. Excessive bleeding, scarring and difficulty during labour could result from such procedures. Since the equipment used for such procedures are not usually sterile, this predisposes the victims to infections. Insertion of objects and concoctions into the vagina of women as treatment for infertility can also result in infections that may aggravate the situation. In is clear from the study that the concept of health have social undertones. Societal perception of a disease directly affects the behaviour of the community to people suffering from that disease and it intends affect the help-seeking behaviour of clients. In a typical pronatalist community like Northern Ghana, the concept of reproductive health can only be meaningful, when opportunities are available for individuals to beget the number of children they prefer. This can only be achieved through an integrated approached to reproductive health. Infertility and childbearing should be given a priority and the possibility of fertility insurance should be explored. ---

Background: Infertility is a major medical condition that affects many married couples in sub-Saharan African and as such associated with several social meanings. This study therefore explored community's perception of childbearing and childlessness in Northern Ghana using the Upper West Region as a case study.The study was exploratory and qualitative using in-depth and key informant interviews and focus group discussions. Fifteen marriage unions with infertility (childless), forty-five couples with children, and eight key informants were purposively sampled and interviewed using a semi-structured interview guides. Three focus group discussions were also carried out, one for childless women, one for women with children and one with men with children. The data collected were transcribed, coded, arranged, and analyzed for categories and themes and finally triangulated.The study revealed that infertility was caused by both social and biological factors. Socially couples could become infertile through supernatural causes such as bewitchment, and disobediences of social norms. Abortion, masturbation and use of contraceptives were also identified as causes of infertility. Most childless couples seek treatment from spiritualist, traditional healers and hospital. These sources of treatment are used simultaneously.Childbearing is highly valued in the community and Childlessness is highly engendered, and stigmatised in this community with manifold social consequences. In such a community therefore, the concept of reproductive choice must encompass policies that make it possible for couples to aspire to have the number of children they wish.

Community-engaged medical education , a concept used in medicine and other health professions, underscores the relationship between the community served and education, which is both interdependent and reciprocal. Our objective was to characterise how SHE could be made relevant to the local context and the health of the community. Methods: Qualitative content analysis was used to explore the perspectives of faculty members, community experts and health professions students on the relationship between SHE and CEME. Semi-structured interviews were conducted with 51 participants between March 2016 and May 2017. Results: Participants highlighted that it was vital for students to become familiar with real-world, locally relevant issues by collaborating with community members and identifying opportunities for engagement. For optimal learning, CEME experiences should be --- The integration of SHE through CEME will draw the community voice into the curriculum and will compel students to connect CEME SHE learning to their continuing education accompanied by relevant curricular changes. Educational partnerships with local organisations, societies and activists provide continuing opportunities for dialogue about, and integration of, SHE. Discussion: The integration of SHE through CEME will draw the community voice into the curriculum and will compel students to connect CEME SHE learning to their continuing education. The credible and respected role of the health professionals and health professions institutions provide a foundation for the development of community partnerships and the promotion of SHE. --- … community involvement would offer interdisciplinary opportunities for students and the community to learn from each other INTRODUCTION T he impact of humans extends over a vast majority of the planet, increasing the pressure placed on limited resources and affecting climate, ecosystems and disease. 1 Health care professionals must acknowledge the connections between climate and ecosystem change, sustainability and health to drive the culture of health care towards ecological responsibility and, subsequently, improve patient and population health. Sustainable health care education develops students' knowledge, skills and attitudes about the interdependence of climate, ecosystems and health, the health sector's impact on the environment, and provide sustainable solutions for both ecosystems and human health. [2][3][4] Currently, little SHE content exists within health professions curricula. 2 Both medicine and nursing have developed SHE learning objectives, 2,5 and recent work has provided guiding frameworks and examples for SHE curricula. 3,4 In light of imminent global environmental changes, it is both important and urgent to implement SHE. Importantly, in order for learners to comprehend the impact of these global environmental changes on patient care, connecting these changes to local health and well-being is paramount. Yet little is understood about the methods by which to facilitate SHE within the local or community context. Community-engaged medical education underscores a relationship between the community served and medical school that is both interdependent and reciprocal. 6 Students learn about the community experientially and the community is mutually engaged in education to provide solutions to the community's health problems. CEME has been used worldwide in medicine as well as other health professions to engage the community in education and to engage the learners in addressing the needs of the communities. 6,7 Curricular time constraints and the proliferation of knowledge that students are required to master pose barriers to the integration of SHE in health professions education. CEME allows health professions educators to provide learners with an applied and immediately germane framework for understanding the impact of climate and environmental change on health within their own local lived context. Hence, the purpose of our study was to characterise how SHE could be made relevant to the local context and community. The findings from this study guide health professions educators to design SHE experiences that are applicable, effective and timely, and improve the health of the local community. --- METHODS --- Design We used conventional qualitative content analysis to explore faculty members', community-based experts' and health professions students' perspectives on the connection between SHE and CEME. 8 Semi-structured interviews were conducted between March 2016 and May 2017. The Institutional Review Board of the University of California, San Francisco, approved the study. --- --- Instrument The semi-structured interview protocol was developed based on the CEME framework by two investigators . The first draft of interview questions was piloted with two participants and then refined. Interviews were recorded and transcribed. --- Analysis Interview transcripts were analysed by two investigators to create the final codebook. Through application of the final codebook, each transcript was coded by two investigators , discussed and reconciled into final codes for each transcript. Through a member checking process, the final results were shared with the participants to ensure that the results represented their perspectives accurately. Data were analysed using dedoose TM . --- RESULTS A total of 51 participants agreed to be interviewed out of the 76 participants identified, who were SHE faculty members and community-based experts or learners . Participants represented the professions of dentistry , medicine , nursing and pharmacy . Most participants were from the USA , and the remaining participants were from the Netherlands, Germany and Switzerland . The interviews lasted 15-60 minutes. Participants described how CEME for SHE should be taught through community involvement, curricular change and the formation of partnerships. These themes are described below and selected quotes representing each theme are listed in Table 1. --- Community involvement Participants highlighted that it was vital for students to become familiar with real-world, locally relevant issues by collaborating with community members and identifying opportunities for engagement. Participants described projects such as pulmonary function testing for individuals living near the highway, or dental exams for students at local schools that simultaneously evaluate the environmental impact of dental treatment. One student participant described how an educational opportunity to visit local low-income and under-resourced areas served as an eye-opening experience to learn where patients came from and the issues that they experienced. Similarly, participants noted that community involvement would offer interdisciplinary opportunities for students and the community to learn from each other. As one participant described, health care providers should engage with their communities on issues of climate change and health. Health professionals are perceived as highly credible and respected members of society and should use that credibility to guide and advocate for communities affected by climate and environmental change. --- Curricular changes Participants noted that for optimal learning, CEME experiences should be accompanied by curricular changes that involve the community, such as inviting community leaders and patients to be Interview questions* • How do you think sustainable health care education can be achieved for learners in a way that helps them connect sustainable health care education to the local or community context? • How can we make sustainable healthcare education relevant to the community learners live in and community they serve? • How can partnerships be developed with community organisations and/ or other groups, such as community leaders to educate learners about sustainable health care education? *Participants were provided the definition of SHE in the recruitment e-mail • A great thing to do would be to go with a public health nurse on a home visit, and give them also some guidance … about observing environmental health things as well as the other things they might be focused on in most home visits • Our medical society has an environmental health committee. That was a great way for us to get students involved, because we did talk about air quality, water quality, we talked about agricultural runoff … and that was a great way to also get medical students and residents involved … We [also] had many community members that we invited on a very regular basis to work with us • I like to see medical students go teach grade school, junior high, and high school children, because if you don't figure out how to make it relevant when you go talk to schoolchildren, they're not going to listen to you. I think that's a way to sort of force the student, in a way, to figure out how it is relevant to the community • Our school of pharmacy is involved in health fairs where they do diabetes screening, hypertension, hyperlipidemia, and I think that would also be a forum where pharmacy students could provide this information about how medications can be disposed properly, and about these take back programs…and even just educating around trying not to take medications that may be expired and dispose of them appropriately Curricular changes • [I found] the interdisciplinary programs really useful because it was so beneficial for the dental students to discuss oral health issues the patient might have with the medical students who were focusing on the medical issues, and yet they are all combined. It's all one person, and what kind of drugs and the impact of the drugs on the individuals. I think that way of teaching is exceptionally important • I think some of the most memorable moments of first year, are patients who have come in and showed how something we're learning about has really affected their life, because you can read about a list of side effects, or ways that someone might suffer with a given condition, but when you actually see that, and you hear someone talk about how it affected their life, I think that's really powerful. To make it relevant to the local context, if there was some sort of site visit or if there was a patient who came in to talk about how their local environment is affecting them, I could see that as being powerful • I've actually had this idea of having an elective class called 'Welcome to the area' bringing people from the community and talking about what they see in their environment and how it plays a role in people's lives and bringing it to the medical and nursing student. Then that way they can really see that it's just something that's really big, that it's actually happening in our backyards. People who deal with those issues, coming to [our institution] and talk about what they're seeing from a community perspective • Think to the extent that it [the curriculum] can be connected to real examples in where perhaps even students would be going to do their clinical rotation, I think that's very powerful for building not just more confident clinicians, but also for building people who will be advocates and advocate for their state legislature around these issues to which patients trust the health care system and their health care providers, particularly physicians, varies by country and time period. 12 Thus, as institutions consider how to collaborate with the community on SHE, future research must address the extent to which the community trusts the profession and how that trust impacts the inclusion of SHE in CEME. The limitations of our study included that the participants were primarily from the USA and represented the profession of medicine more than other health profession. --- CONCLUSION The implementation of SHE for health professional learners should be accomplished in collaboration with the local community in order to understand real-world community issues affecting health. Integration of SHE within CEME will draw the community voice into the curriculum and compel students to connect CEME SHE learning with their continuing education. Community partnerships are key to creating educational SHE opportunities as well as change. Finally, the credible and respected role of the health professionals and health professions institutions provide a foundation for the development of community partnerships and the promotion of SHE.

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ?If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Introduction Today, the importance of taking evolutionary perspectives in understanding physical and physiological traits of humans is well-recognized , and there is little doubt that humans have experienced selection in relatively recent human evolutionary history . Indeed, many human physical and physiological trait variations have been celebrated as evidence of natural selection or gene-culture coevolution . Increasingly, researchers in the social sciences are paying more attention to evolutionary explanations for universality in human behavioral and cognitive traits . Nonetheless, to this day the majority of evolutionary studies on homo sapiens have focused on the past, using evolutionary theories to explain evolved physical, cognitive, and behavioral traits , with the notable exception of human behavioral ecology, which primarily focuses on human phenotypic plasticity in small scale societies within the fitness-maximization framework . However, recent advances in genomic sciences and statistical techniques provide us with a better understanding of the genetic architecture of quantitative traits in humans, which, combined with observational studies on how selective pressure is operating in contemporary human populations, makes it possible to detect on-going selection and to potentially predict the future evolutionary trajectory of specific traits . Of course, special attention needs to be paid when behavioral traits are under consideration, as the transmission of human culture could systematically and significantly affect the evolutionary dynamics . Educational attainment has long interested social scientists. It is strongly, negatively associated with fertility and extensive research in behavioral genetics estimates significant heritability . Large scale Genome Wide Association Studies have identified specific loci that contribute to EA , and two recent studies show that in certain western populations there is on-going negative selection against EA , where the genetic component of EA as measured by polygenic score has been declining for the past century in Iceland and the US respectively. On the other hand, the phenotype of EA, along with other cognitive measures such as the IQ score, has drastically increased in western populations . To explore the unusual evolutionary dynamics of EA, I construct an agent based simulation model to track the relative change in both genotypic value and phenotypic value of EA assuming that 1) EA is negatively associated with fitness; 2) the phenotype of EA has both a genetic and a certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint the genetic component is vertically transmitted from parents to offspring and the cultural component is obliquely transmitted from the parental to the offspring generation and 4) the cultural component is transmitted with a prestige-based bias, which refers to the psychological tendency of humans to copy the behaviors of the more prestigious individuals in a community . Alternatively, the tendency to copy from more educated people can be viewed as payoff biased transmission if education itself is considered as a desirable outcome. The present simulation model differs from traditional evolutionary models in two ways. First, this model is not aiming at identifying evolutionarily stable strategies ; it is rather an attempt to combine standard evolutionary logic and cultural evolutionary theories to explain and potentially predict genetic and phenotypic changes of a specific trait over time. Second, this model explicitly specifies the phenotype function as having a gene-culture interaction term in addition to the additive components of gene and culture. Orthodox gene-culture coevolution theory explicitly recognizes the importance of both factors, but typically models the phenotypic contribution of gene and culture in an additive fashion , presumably for analytical tractability. In a recent proposal, Day and Bonduriansky provide a theoretical framework to account for both genetic and non-genetic inheritance under a generalized from of the Price equation . However, to quantitatively track evolutionary dynamics they also assume that the phenotype is additively determined by a genetic component and a non-genetic component. As will be seen, the inclusion of a gene-culture interaction term significantly alters the evolutionary trajectories of EA. --- Methods --- I employ the common assumptions of asexual reproduction and non-overlapping generations in theoretical evolution . Each individual is represented by a vector [g, z] where g stands for the genotypic value, and p stands for the phenotypic value . In all simulation runs, the population is initially seeded with individuals drawn from the uniform distribution . "Cultural fitness", or the relative likelihood that a particular individual is picked as a model, is defined as zi φ , where φ is a population level parameter which roughly represents the idea of a "cultural reproductive skew". When φ>1, individuals with high cultural fitness have disproportionally larger probabilities to produce cultural offspring by certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not this version posted April 17, 2019. ; https://doi.org/10.1101/605311 doi: bioRxiv preprint serving as models for culturally naïve individuals, and when φ=0, all individuals in the population are equally likely to be picked as a model. Starting from generation 2, for any given individual i, its phenotype of interest is defined to be zi = giβ + zmodel + gizmodel βinter where giβ represents the genetic contribution from the focal individual, zmodel represents the cultural contribution from the cultural model that the focal individual picked from the previous generation based on the model's cultural fitness, and gi zmodel βinter represents the interaction between the genetic component and the cultural component. I constrain the β coefficients for the first two terms to sum to 1 as it reduces the total number of parameters and I am more interested in the relative contribution from gene and culture and their magnitude compared to βinter. For any given individual i, its genetic fitness wi is defined to be wi = wb + γzi + ε where wb represents the baseline fitness, γzi represents the fitness contribution from the phenotype of interest , and ε is a uniformly distributed error term. Importantly, genetic fitness is forced to be non-negative. In cases where the numeric value of an individual's genetic fitness becomes negative during simulation runs, it is automatically set to be 0. In the beginning of the life cycle, an individual is born with a genotypic value inherited from his parent, and will sample the whole parent generation and pick an individual as his model based on cultural fitness. His own phenotypic value is then determined by the phenotype function . Once all individuals in the focal generation have acquired a phenotypic value, they become the new parental generation and produce offspring based on their genetic fitness as determined by . At the end of each life cycle, both the genotypic value and phenotypic value can change with some probability in frequency and magnitude. To further explore the evolutionary dynamics in relevant contemporary contexts, I added a "demographic shock" condition by having EA uncorrelated with fitness up to a fixed generation number and then changed the correlation into negative. This is to mimic the process certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint in a relatively short amount of time. --- Results --- General result The main result from my simulation model is that phenotypic value of EA can increase as the genotypic value of EA decreases , at least for a short period of time . There are two opposing forces at work here: the negative correlation between phenotype and genetic fitness which constantly drives down genotypic value, and the prestige-biased transmission which increases population mean phenotypic value. After running a large number of simulations I identified two key parameters that contribute to this pattern of initial divergence of genotypic value and phenotypic value. First, increase in phenotype depends on the presence of variation in phenotype. In the model, the main source of phenotypic variation is supplied by random "cultural mutation" at the end of each life cycle. This result is largely intuitive: if every individual in the population has roughly the same phenotypic value, then there is less potential room for biased transmission to increase population mean. Second, the interaction term contributes to mean phenotypic value increase by magnifying the effect of biased transmission, as it places extra weight on the model's phenotypic value in constructing the focal individual's phenotype. When βinter is set to be 0, the initial phenotypic increase was much milder . As genetic fitness is negatively correlated with phenotype , genotypic value always monotonically decreases from the beginning until variation is exhausted in all simulation runs. The initial increase under certain parameter settings does not last indefinitely and we quickly see a subsequent drop over time because the variance in both genotypic and phenotypic values decreases due to stabilizing selection . Eventually, the increase in phenotypic value due to picking a model with higher EA becomes insufficient to compensate for the overall population level decrease in phenotypic value due to natural selection acting against the phenotype. In the vast majority of simulation runs genotypic value stabilizes after ~ 120 generations. This phenomenon is entirely due to the exhaustion of genetic variance. In the initial population, genotypic values and phenotypic values are randomly seeded, and I assume a very small certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint . Additionally, the simulation only looks at a relatively short evolutionary time period . As a result, once natural selection has used up all existing genetic variation no further genetic evolution can occur. In cases where variation in environmental contribution is large, all observed phenotypic variation would be due to environmental differences ; in cases where there is no variation in environmental contribution, there would be no phenotypic variation nor genotypic variation . --- Adding demographic shock In the "demographic shock" condition, I model educational attainment as first contributing nothing to fitness , then after several generations EA suddenly becomes negatively associated with fitness . As shown in Figure 2, the demographic shock does not immediately drive down average phenotypic value; rather, there is a lag in the phenotype response. Genotypic value, on the other hand, remains constant for the "preshock" period and starts to decrease immediately after the shock. Unlike the "non-shock" condition where the contribution of EA to fitness, γ, is always negative, the first 20 generations for the "shock" condition does not incur genetic evolution as genetic fitness is uncorrelated with phenotype. As expected, we observe no change in genotypic value up to generation 20. However, in the "pre-shock" period, cultural evolution in the form of prestige-biased copying happens which significantly boosts the phenotype of EA. This cultural evolutionary process reduces phenotypic variation in the absence of genetic change, and consequently the diverging pattern only lasts a relatively short period of time , compared to the non-shock condition . --- Effect of prestige-biased transmission and other parameters To systematically examine the effect of other parameters, I thus used the end point phenotypic and genotypic values to represent the quasi long term evolutionary outcomes within this short time window . As expected, most genotypic values are significantly lower at the end of the simulation run than at the beginning due to natural selection. In accordance with the logic of evolution, larger β and certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint increases the strength of natural selection on genes, and when both β and γ are small we barely observe any decline in genotypic value . Most end-point variation in phenotypic values, on the other hand, is primarily due to the varying degree of prestige-biased transmission. For example, when the ego's phenotype depends more on the phenotypic value of a cultural model , the magnitude of prestige-biased transmission φ is positively correlated with the end point phenotypic values. In other words, it "helps" drive up the population mean phenotypic value. Interestingly, an increase in end-point phenotypic value due to larger cultural reproductive skew is accompanied by a smaller end-point genotypic value under certain parameter settings . This is because the same conditions that leads to large phenotypic variance also leads to large genotypic variance, thus as cultural evolution drives up mean phenotypic value, genetic evolution pushes down mean genotypic value. Expectedly, these conditions are also where we observe the initial diverging patterns, except when φ = 0 . --- Discussion --- Simulation as a way of studying gene-culture coevolution It is no news that human cognitive and behavioral traits are significantly influenced by human culture , and much effort has been put into modeling the gene-culture coevolutionary dynamics analytically . However, most models are designed for analyzing long term allele frequencies and evolutionarily stable scenarios for discrete traits. These models are not ideal for examining short term evolutionary changes, especially when models are becoming more complex from incorporating additional assumptions and parameters. The evolution of educational attainment serves a case in point. Although it is not difficult to envision a situation where phenotype and genotype undergo different selective pressures, the complexity of real-world biological systems often makes it difficult for humans to evaluate not only the effect of individual assumptions and parameter settings but also the short-term/long term evolutionary outcomes. For example, the steady increase in educational attainment over the past century can be plausibly attributed to income increase or "top-down" government policies aiming at increasing literacy , but it can be difficult to know the genetic consequences of these environmental effects or to examine how certain cultural evolutionary mechanisms-such as prestige or payoff biased transmission influence the evolutionary certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint and provide validation to analytic predictions , yet there has been very little simulation attempt to track the evolutionary change of genetic and cultural component of quantitative traits simultaneously. In the present case, the non-intuitive hill-shape phenotypic value change provides important insight for both understanding the role of specific transmission biases in shaping genetic evolution and potentially predicting the short term evolutionary change in the trait of interest . --- Gene and culture evolving towards opposite directions: How should we interpret the result? There can be many environmental factors that contribute to the phenotypic change of any human trait. In particular, the technological progress and vastly improved living conditions since the industrial revolution has drastically altered many human phenotypes; for example, adult human height has substantially increased throughout the world over the past century due to improved nutrition and reduction in childhood disease . What is especially interesting about EA is its consistent negative association with fertility. In general, more educated women delay the onset of childbearing and have fewer children overall compared to less educated women . This pattern is very robust in both developed and developing countries , and various theories have been proposed to explicate the intrinsic, potentially causal relationship between education level and fertility . This negative association between EA and fertility, which is a reasonable proxy for fitness, has important evolutionary consequences, as both theoretical predictions and empirical evidence suggest that the genetic component of EA is undergoing a temporal decline . The phenotype of EA, on the other hand, may experience a very different type of selective pressure. A large literature in sociology shows that educational attainment and socio-economic status are associated , and cultural evolutionary theory predicts that humans readily copy the behaviors of the those who are perceived as more prestigious or successful . Although educational attainment cannot be "copied" in a literal sense, individuals who pick models with high EA are likely to be more motivated in learning and committed to their academic studies, and as a result become more likely to obtain higher EA themselves. Thus, genetic fitness and cultural fitness of the same trait invites selection in opposite directions, and what is shown certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not this version posted April 17, 2019. ; https://doi.org/10.1101/605311 doi: bioRxiv preprint in the simulation model is that biased transmission alone can lead to a short term increase in phenotype. In reality, the massive increase in EA and over the past century is almost certainly a combination of various social, economic, and cultural factors, and the educational expansion has been described as an "environmental override" . The substantial increase in the phenotype of EA indicates that whatever the mechanism is, the decline of genetic component of EA has been vastly overcompensated by enhanced environmental inputs. However, the simulation results reveal a crucial insight: short term change in phenotypic value does not guarantee long-term phenotypic change in the same direction. A natural question to ask is whether there is some ceiling effect of environmental input: that is, whether the environmental factors can drive up the phenotype of EA indefinitely. Unfortunately, the exact relationship between any behavioral genotypes and phenotypes are poorly understood . Yet it seems reasonable to assume some upper limit of environmental influences, which means at some point genetics will become the limiting factor in determining the phenotype of EA, which may eventually decline as a result of natural selection. It has been suggested that because humans constantly modify their environment, the association between education and fertility may not remain negative long enough to have noticeable effects . This is certainly a possibility, but at this point we simply do not have enough information to accurately predict future cultural trends. Though it is not difficult to imagine a situation where highly educated, successful people also have more offspring, there is a genuine trade-off between the time one invests in education and in raising offspring. Alternatively, education could be dissociated from status or prestige, yet this possibility does not seem very probable given the great emphasis on education in modern economies . A disappearance or reversal in the selection on EA is likely to involve large scale, top-down intervention. Even in the short term, the decline of the genetic component of EA may have non-trivial consequences. Although people can reasonably question whether educational attainment is the trait under selection , it is not crucially important what the precise trait is. What matters in this evolutionary process is that genetic variants associated with EA are being selected against. We know that these genetic variants are primarily involved in brain development processes and certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint , which accords with previous research showing that the same set of SNPs is associated with both educational attainment and cognitive function . Thus, it seems that at least part of the genetic component of EA comes in the form of cognitive ability, which is important for not only EA but also other aspects of social outcomes such as income and job performance . --- Are contemporary humans fitness maximizers? A long-standing question in evolutionary biology is whether organisms behave as if they are maximizing fitness . Humans, as an evolved primate, must be subject to the logic of evolution via natural selection ; yet on the other hand, due to the presence of a second, cultural inheritance system , human behavioral evolution seems to require a different type of analysis. Richerson and Boyd offer a plausible account of how cultural transmission could result in maladaptions -when genetic fitness is not perfectly aligned with cultural fitness for any particular trait. Stated broadly, our capacity for culture evolved so that we could acquire information socially, which conferred significant fitness advantage in the evolutionary past; yet the very same psychological mechanism allows for maladaptations at the same time. In the modern environment, copying the most prestigious individuals may not be the most adaptive strategy . However, this view does not go unchallenged by evolutionary biologists. El Mouden et al. present a narrative of why genetic fitness should always align with cultural fitness by arguing that "genetic selection will shape psychological mechanisms to avoid cultural traits that bear a genetic fitness cost". This line of thinking assumes an infinite strategy space, in particular the possible psychological mechanisms that can detect genetic fitness consequences in novel environments. Even if we ignore all constraints on the potential evolution of human cognition, it remains the case that novel mutations reaching evolutionary equilibrium typically require some significant amount of time . What is very notable about contemporary humans is that the rate of environmental change can be very rapid. For example. substantially reduced infant mortality , dissociation of sexual activity from reproduction , and the reversed association between status and fertility happened only in the past few hundred years, and it is unrealistic for genetic evolution to generate new psychological mechanisms to keep up with the cultural change. Thus, certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not this version posted April 17, 2019. ; https://doi.org/10.1101/605311 doi: bioRxiv preprint to understand the short-term evolutionary dynamics it would be better to focus on what is known about the selective pressure and the heritability of the target of selection in contemporary societies. --- Conclusion This paper demonstrates the potential utility of a type of simulation model that specifies both the phenotype function and the fitness function in tracking short term evolutionary change in human behavior traits. The results indicate that payoff/prestige biased transmission can significantly affect the evolutionary trajectory of educational attainment in contemporary societies, and that a short term increase due to increase in environmental input towards phenotype may not last indefinitely. The long-term evolutionary trajectory of EA depends on a few key parameters, and researchers can gain valuable insight from analyzing the conditions under which the genotypic value and phenotypic value change as a result of both genetic and cultural selection. certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure legends

There has been substantial increase in education attainment (EA) in both developing and developed countries over the past century. I present a simulation model to examine the potential evolutionary trajectories of EA under current selective pressure in western populations. With the assumption that EA is negatively correlated with fitness and has both a genetic component and a cultural component, I show that when prestige-biased transmission of the EA (i.e. people with more education are more likely to be copied) is present, the phenotype of EA is likely to keep increasing in the short term, yet the genetic component of EA may be undergoing a constant decline and eventually lead to an overall decrease in the phenotype.

Introduction More than 20% of the burden of disease among women of reproductive age is connected with sex and reproduction. In the developing world, a woman's lifetime risk of death from maternal causes is 33 times that of her counterparts in developed countries [1]. It is also recognized that women suffer silently from a large number of reproductive illnesses, which were termed the silent emergency. This understanding leads to women's health researchers and activists focusing more on women's health, especially in the field of reproductive health [2]. Women of reproductive age have health problems like endometriosis, uterine fibroids, gynecologic cancer, HIV AIDS, interstitial cystitis, polycystic ovary syndrome , menstrual irregularities, complications related to contraceptive use, mid-cycle pain, mid-cycle spotting, dysmenorrhea [3]. The major problem of reproductive health that is faced by three fourth of the world's women, i.e., menstrual irregularities, result from various disorders including endometriosis, fibroids, PCOS, hormonal imbalance, increased stress levels, and improper use of contraceptives [4]. A study showed that malnutrition in females, e.g., due to anorexia nervosa and bulimia nervosa, which leads to reduced energy levels affects up to 5% of women of reproductive age causing amenorrhea, infertility, and increased likelihood of miscarriage [5]. An estimated 105 million married women in the developing world face an unmet need for contraception [6]. The fear of weight gain affects the uptake and continuation of hormonal contraceptives, although existing trials indicate that any such effects are small. For all methods of hormonal contraception, weight above 70 kg is associated with increased failure rates [7]. Evidence suggests that excess insulin levels lead to high androgens level, which suppresses ovulation and causes excessive hair growth and acne, two important signs of PCOS [8]. Evidence indicates that depression is closely linked with disproportionate exposure to risk factors, stressful life events, and adverse life experiences that are more common for women and that also affect their reproductive health [9]. Mental disorders including depression, which is the most important mental health condition for women worldwide, make a significant contribution to the global burden of disease. Women suffer more often than men from the common disorders of depression and anxiety, both on their own and as comorbidities [10]. According to a study, 61.6% of women globally are suffering from major mental disorders including depression and anxiety, leading to increased reproductive health problems [11]. An estimate suggests that 25% of women of reproductive age have menstrual irregularities including heavy bleeding in Turkey; this is 20% in China and 15% in India. Infertility, defined as not getting pregnant within 12 months of having unprotected sex with the same partner, affected 5.8% of married or cohabiting women aged 15-44 years in 2008-2012 in the United States [12]. Worldwide, in 2015, 12% of women are estimated to have had an unmet need for family planning; that is, they wanted to stop or delay childbearing but were not using any method of contraception [13]. --- Materials And Methods --- Study design, period, and area This is a community-based cross-sectional study conducted for a period of seven months from February 1, 2022, to August 30, 2022, in three villages around Dijikot in the Punjab province of Pakistan. --- Inclusion and exclusion criteria Mentally and physically healthy women between the ages of 18-48 years. Women who were under or above the aforementioned age range or/and had congenital deformities, intellectual disabilities, a history of major accidents or trauma, and a family history of motor neuron disease were excluded from this study. --- Study population The total population of villages covered by the study is estimated to be 105,052, of which 54,627 are males and 50,425 are female. From this female population, those who are less than 18 years old and more than 48 years old were excluded, and 38,109 females under the reproductive age group were listed. --- Sample size and sampling technique A multistage random sampling technique was used to select villages, and each village is considered a cluster. The population proportion to size method was used in selecting the samples from each cluster. With a 95% confidence interval, 4.88% margin of error, a population size of 38,109, and 50% response distribution for each question, the sample size calculated using the formula n =N x/E2 + x) was 400. --- Study tool A structured questionnaire consisting of three sections was used. The first section covered the demographic data of the woman, including her current age, educational status, socioeconomic status based on family joint income, type of family, marital status, and religion. The second part consisted of the age of menarche start, duration of the menstrual cycle, number of days of menstrual bleeding, number of painful bleeding days, irregular bleeding episodes in a menstrual cycle, age of first sexual encounter, number of pregnancies and parity, number of sexual partners, and age of marriage. The third section had 11 questions related to awareness regarding reproductive health and hygiene practices. Socioeconomic classes are defined by different income ranges per month. Monthly income levels of Rs 4,000 to Rs 20,000 was defined as the cut-off for the lower class, and Rs 20,000 to Rs 40,000 for the upper lower class. For the middle class, the cutoff was Rs 40,000 to Rs 80,000 and for the upper middle class, the cutoff was Rs 80,000 to Rs 100,000. People earning above Rs 100,000 were categorized as belonging to the upper class. --- Statistical analysis The data collected were entered in Microsoft Excel , and statistical analysis was done using IBM SPSS Statistics for Windows, Version 28.0 . The Chi-square test was applied, and p-value < 0.05 was considered significant. --- Ethics statement The study protocol was approved by the Ethics Committee of Faisalabad Medical University, Faisalabad, Pakistan . All patients were informed about the objectives of the studies. They provided their consent and confidentiality was assured among participants regarding the information they give for this research article. The study was conducted in line with the ethical principles of the Declaration of Helsinki. --- Results In this study, the majority of the respondents were between 22-28 years of age , with education mostly up to eighth grade , unmarried , from the lower class , having a nuclear type of family , and belonged to the Muslim religion . --- Category Number Percentage Age of respondent Thirty-three percent of women aged 22-28 years never used any method of contraception and combined oral contraceptive pills usage was found to be 31.7% in the same age group . It was found that 10.8% of participants with education up to fifth grade and 70% of participants with education up to eighth grade have never used any method of contraception. In lower class and upper lower class, 33.3% and 41.7%, respectively, never used any form of contraceptive. These results are found to be significant with a p-value < 0.05 . --- Method of contraception used --- TABLE 5: Comparison of methods of contraception used in different demographies of participants The significant P-value is 0.05 The use of sanitary pads during menstrual bleeding is higher in educated women and married women with 68.8% of those who used pads being educated to the intermediate level and 62.5% being married . The frequency of using cloth pieces was more prevalent in less educated and unmarried women, which is 50% and 87.5%, respectively. A similar prevalence is seen for cotton use, which is 64% and 50 % in women who studied up to eighth grade and women from the low socioeconomic class, respectively. Table 7 shows these results are found to be significant in chi-square analysis. The significant P-value is 0.05 Table 12 shows that 37.5% of women with intermediate education accepted knowing about screening for cervical cancer. The prevalence is found to be about 50% in married women and 60% in the upper middle class. Of women with education up to eighth grade, 65% answered with No, and the prevalence is 50% for women from the lower socioeconomic class. Two hundred and forty women from nuclear-type families answered No. These results are found to be significant in chi-square analysis as seen in Table 13. The significant P-value is 0.05 Do When asked about how often the participants had pap test screening so far, 50% of women with up to eighthgrade education, 144 unmarried women , and 41.7% of lower middle class answered never . Ninety percent of married women, 60% of the upper middle class, and 90% of women with three-generation families mentioned having it regularly. These results are significant in chi-square analysis as seen in Table 15. How often you had pap test screening so far? The significant P-value is 0.05 Table 16 shows that 41.7% of married women, 33.3% of women in the upper class, and 50% of women in the upper middle class mentioned that they know about the human papillomavirus vaccine, while 68.4% of women with education up to eighth grade and 47.4% of lower class women did not know about it. These results are significant in chi-square analysis as seen in Table 17. The significant P-value is 0.05 --- Do you know about the --- Discussion The prevalence of knowledge about the vaccination of HPV was found to be less in women from low socioeconomic status , as compared to women from upper socioeconomic status, which was 33.3%. This can be attributed to poor knowledge, which translates into poor practices. A review article highlighted the finding that bad odor contributed to fear, embarrassment, and distress among school-going girls; it hampered freedom to participate in daily activities during menstruation and is an insignia of proper menstrual practice [14]. Fortunately, as per our study, the condition in Pakistan is still better than in other Asian countries. For instance, in Nepal, very strict ritual seclusion of "Chaupaudi" is still practiced where women are ostracized during the entire period of menstruation [14]. Limited availability of soap and water deters some women from low socioeconomic backgrounds from regular washing of the groin during menses [15]. It has been suggested that women with dysmenorrhea or other menstrual disorders were often hesitant to discuss matters pertaining to sexual health owing to their warped cultural values and many found the healthcare providers to be unsupportive [15] A systematic review was conducted to assess the effectiveness of "hardware intervention", that is, the provision of absorbing materials to address the material deprivations and access to water, sanitation, and hygiene facilities [16]. A moderate non-significant effect was observed when reusable homemade and disposable sanitary pads were provided. Nevertheless, it is still believed that the institutional availability of pads can benefit young girls hailing from a low socioeconomic background, as a study in Ghana revealed that school attendance rose by 9% after five months of provision of disposable sanitary pads [16]. The same is the case in our study in which only 31.3% of women from low education backgrounds use sanitary pads and the prevalence is even less in lower socioeconomic classes. Menses leave policy has been implemented in the United Kindom, India, and Australia where menstruators are exempted from working while they are experiencing severe pain or discomfort [14]. This can empower women as it is an acknowledgment of their physiological process. Such policies can also be introduced in Pakistan to facilitate the healthcare workers and the general working force. In India, the majority of women are excluded from religious gatherings and, in rural areas, women are restricted from even entering the kitchen during menstrual bleeding days [17]. The superstition of the association of menstruation with evil spirits is particularly prevalent in Asia. A menstruating woman deemed impure is more vulnerable to getting possessed by demons and, hence, some women bury the clothes used during menstruation [17]. Akbarzadeh et al. reported a significant association between age at menarche and dysmenorrhea onset [18]. On the other hand, Kural et al. could not find such a correlation [19]. The different results may be attributed to the differences between nutritional habits, public health, geographic location, and cultural factors in the studies. According to the literature, dysmenorrhea usually begins within one to two years after menarche [20]. This indicates the importance of educating adolescent girls at the age of menarche about dysmenorrhea. In our study, dysmenorrhea prevalence is found to be 60.0% for at least two to three days in a regular menstrual cycle in participants. This finding is consistent with the results previously reported in the literature. In a study by Chen and Chen from the United States, adolescents were observed to largely have moderate-tosevere menstrual cramps [20]. In a study by Gun et al., dysmenorrhea began with menstruation in 39.9%, one to two hours before menstruation in 37.2%, and a few days prior to menstruation in 22.9% of the participants [21]. In our study, 60% had menarche at the age of 12 years with 77% of participants having a duration of the menstrual cycle of 25 to 28 days, 70.5% having four to five days of menstrual bleeding, and 14% having at least one episode of irregular bleeding in a menstrual cycle. Most developing counties, however, have been unable to implement comprehensive Pap smear screeningbased programs. In countries where Pap smear screening is available, it often is accessible to only a small proportion of women through private healthcare providers, or it is offered primarily to young women through maternal or child health clinics or family planning clinics where the population being screened generally is not at high risk [22]. These approaches have had little effect on morbidity and mortality and generally are not as cost-effective as centrally organized screening programs implemented by the public sector [23]. STIs have a great impact on the health of populations worldwide. These may be contracted by people of any age, race, or social standing, and their early diagnosis and treatment are necessary to avoid propagation. Sexual education is fundamental to STI prevention [24]. The high number of participants with a lack of knowledge about healthy reproductive practices in our study can be explained by the fact that the majority of Pakistani women consult family elders, local hakeem , or daies for their decision about reproductive health. People also sometimes turn to pharmacies or traditional healers instead of healthcare facilities, and self-medication or alternative therapies can make STDs worse or better. Many people seek medical advice when the pain is unbearable. Therefore, the differences in STI prevalence between upper and lower classes, as well as under and well-educated classes can be attributed to various factors like family norms, social beliefs, and level of awareness of complications between the study populations. The factors that determine health behaviors in Pakistan can be seen in various physical, socio-economic, cultural, and political contexts. Religious and social ethics discourage open discussion of sexual matters. Women's low social status limits their economic opportunities, and women can trade sex for money or other forms of support. Poor health services provide little for the prevention and treatment of STDs. Various factors, including proximity, affordability, availability, family pressure, and strong community opinion, lead to self-care and consultation with traditional healers, hakeems, or even quacks [25]. Long-held misconceptions continue to contribute to the nationwide neglect of treatment and prevention of STIs. Immediate STD detection, prevention, and STD-related counseling in clinics for vulnerable groups, as well as raising awareness, should be the basic pillars of the health policy of the public and private health sectors in Pakistan. --- Limitations of the study It is important to note that this should not be considered an accurate predictor of knowledge, attitude, and practice related to menstrual hygiene in the female population of the whole country. Second, our study had a narrow coverage of the socioeconomic classes, with most participants belonging to the lower and middle classes. Third, using a specific type of contraceptive also depends on other factors like medical indication or contraindication, cultural, and religious beliefs, but in this study, the research mainly shows the impact of education and socioeconomic levels on knowledge about different types of contraceptives and their usage. The course should be more varied and conducted with a larger sample size in the future to more comprehensively assess women's practices in a geographic area. The questionnaire was self-designed and many commonly believed myths may have been overlooked, and recall bias may have occurred answering some questions in the survey. --- Conclusions Knowledge and practice about reproductive health as well as awareness about the HPV vaccine and Pap smear test were found to be low in the participants with low education levels. Moreover, in these rural areas, the reported use of contraceptives was low among women with education up to the eighth grade and reliable methods of contraception were more prevalent in women with an education of intermediate level or above. Women from the upper and upper middle class have more awareness and knowledge about STDs and their screening methods and cervical cancer screening and these results were found to be significant. The trend of having home deliveries is found to be more in women with less education. and hospital deliveries are more prevalent in women with education up to an intermediate level or above. To overcome all of these, early education regarding menstrual hygiene and regular screening of cervical cancer and STDs should be provided to women of reproductive age in these rural areas using various learning materials. It's also very important to get teenage women involved in health education, both in high school and college, to improve the situation. --- The significant P-value is 0.05 Table 20 shows that 92.5% of women with education up to eighth grade had one to two deliveries at home and 68.8% of women with education up to fifth grade had three to four deliveries at home. Fifty percent of both lower and upper lower class had one to two deliveries at home. Twelve women from the upper middle class had all of their deliveries at home, and 20 had five-six deliveries at home. These results are significant in chi-square analysis as seen in Table 21. The significant P-value is 0.05 --- Age Table 22 shows that 64.3% of women with education up to fifth grade and 28.6% of women with education up to eighth grade had all their deliveries at a hospital. This was also true of 22.9% of women from the upper class and 20% of the upper middle class. Thirty-three women with bachelor's education had one to two deliveries in the hospital. These results are found to be significant in chi-square analysis as seen in Table 23. The significant P-value is 0.05 --- Age On asking about follow-up visits after each delivery, 112 women with education up to eighth grade admitted to having it regularly and 96 women irregularly . Sixty women with education up to fifth grade never had a follow-up while 35.7% from lower middle and 59.5% from the upper lower class had it regularly. The prevalence of having regular follow-up visits is higher in nuclear families compared to joint and threegeneration families. These results are significant in chi square analysis as seen in Table 25. 2022

Background: In addition to physical welfare, reproductive health is also vital for psychological well-being. All stages of reproduction can take place safely if reproductive health is well cared for, and it ultimately leads to the formation of healthy new offspring. The aim of this study is to know about reproductive healthrelated knowledge and practices in women of reproductive age in an underdeveloped area in Pakistan and to identify the associated factors that give a meaningful impact on reproductive health. Methods: A cross-sectional study was carried out among women of childbearing age in underdeveloped areas in the province of Punjab, Pakistan. A sample of 400 was taken on a random basis. All the relevant data were collected from February 1, 2022, to August 30, 2022, with the help of a structured questionnaire, designed specifically for the study, informed consent was taken from all of the participants before data collection. Questions were asked about their menstrual cycles, use of contraceptives, knowledge about sexually transmitted diseases, screening of cervical cancer, pap test, human papillomavirus vaccine, and related to home or hospital deliveries. Socioeconomic classes were defined by different income ranges per month as lower class, upper lower class, middle class, upper middle class, and upper class. Results: Ten percent of participants with education up to fifth grade have never used any method of contraception while 70% of participants who studied up to eighth grade never used the same. In lower class and upper lower class, the prevalence is 33.3% and 41.7%, respectively. The prevalence of screening for cervical cancer is 50% in married women and 60% in the upper middle class. Of women with education up to eighth grade, 65% answered with No, and the prevalence is 50% for lower-class women. Regarding the human papillomavirus vaccine, 41.7% of married women, 33.3% of women in upper class, and 50% of women in the middle class mentioned that they know about it, while 68.4% of women have education up to eighth grade and 47.4% of lower-class women answered with No. Of women with education up to eighth grade, 92.5% had one to two deliveries at home, and 68.8% of women with education up to fifth grade had three to four deliveries at home. Fifty percent of women from both lower and upper lower classes had one to two deliveries at home. Twelve women from the upper middle class had all of their deliveries at home and 20 had five to six deliveries at home. Of women with education up to fifth grade and eighth grade, 64.3% and 28.6%, respectively, had their all deliveries at a hospital; 22.9% of women from the upper class and 20% of the upper middle class also had all deliveries at the hospital, Thirty-three women who graduated from college had one to two deliveries in the hospital. All of these results are found to be significant with a pvalue <0.05. Conclusion: Knowledge about reproductive health is less prevalent in women with low education and the same is for lower and lower middle socioeconomic class. The education level of women and their socioeconomic class is one of the major factors that have a meaningful impact on their reproductive health and practices.

Introduction The government's disregard for development in border regions has resulted in social gaps. Due to centralized development policies, social disparity in Indonesia's and its neighbouring country's border regions is connected with complex security circumstances such as illegal trade, border area conflicts, population distribution, limited natural resources, and unequal economic activity . His divide gives birth to a comparatively high level of socio-political complexity, including the issue of poverty, in the border region between Indonesia and Timor-Leste. Belu District in Figures reports that the impoverished population in East Nusa Tenggara and Timor Leste's border regions exceeds 10 percent, above the national average of 9 percent . Women whose population larger than men have not been a concern in the current poverty reduction policy-making . As a consequence, women suffer restricted access to the economy, education, and health . Muryani and Esquivias underlined that poverty reduction policies need a comprehensive approach. Brady reaffirmed the notion by noting that it is vital to include stakeholders from diverse structural and cultural interest groups in formulating poverty reduction policies. So far, the majority of research on inequality in poverty reduction policies in border regions has focused on three primary themes. First is dynamic interactions in border communities as the border communities are increasingly interconnected due to each community's urgent economic needs and possible intensive relationships as a result of the governments' limited intervention . The second strand of research focuses on looking at both governments' key roles in reducing the poverty gap . The third theme is that the prevalence of poverty in border regions as a result of internal and external variables, such as the geographical position of both regions and the roles of customs and traditions . Considering these three patterns, it appears that the most ideal economic intervention would be to highlight people's opinions by tabulating issues on a micro scale and backing it with government's policies at a national level. Ocusing on the community being accountable for the slow per capita development and opinion is not justifiable due to the fact that the community is positioned as a passive subject when voicing the problems encountered. This research intends to supplement the shortcomings of prior studies, which do not account for initiatives to eliminate poverty disparities in border areas between two countries. In addition, the purpose of this research is to assess the difficulties and efficacy of poverty reduction policies in the border regions of Indonesia and Timor-Leste, as well as to examine the types of inequality that exist in the border regions. The findings can offer a more comprehensive understanding of the types of obstacles and the effectiveness of poverty reduction policies in border areas, and the policy disparity in border areas. This study is built on the premise that poverty reduction in border regions depends on the construction of policymakers' perspectives that prioritize community needs. Poverty alleviation policies that are not aligned with community needs are deemed ineffective and a waste of resources. --- Literature Review --- Policy Inequality Government policies are systemic and measurable solution for poverty reduction . The World Bank evaluates four policy-related causes of inequality. First is opportunity inequality, which is shown in the plight of children from impoverished families impacted by the places where they were born or their parents' education. Second is labor market inequality, where highly competent people earn more, while others get limited opportunities to enhance their abilities. The third one is wealth concentration, where the the elite's ownership of financial assets, such as real estate and stocks, contribute to existing and future inequality. The f ourth one is inequality in facing crisis . The poverty alleviation programs are created t o prevent policy gaps, including target and data used to establish the targets, participation of local governments, general public, and program beneficiaries, as well as the program implementation at the government and community levels . The process of anticipating government policy inequalities in poverty reduction may be implemented in many phases: 1) Acquisition of excellent governance and Clean Governance; 2) Development of the people's economy; and 3) Synergistic and integrated interaction between the government and Non-Governmental Organizations in community training and mentoring programs . The following five steps must be taken to anticipate policy imbalances: Agenda Setting, Policy Formulation, Policy Adoption/Legacy, Policy Implementation, Policy Evaluation . --- Poverty Alleviation Policy The causes of structural and social poverty include disparate development results, institutional structures, and development strategies . Good government policies may alter the impoverished circumstances of its citizens, raise their earnings, and enhance communal fairness. Conversely, poor government policies will result in a regime's financial collapse and the spread of poverty . Policies aimed at alleviating poverty should be based on the ratio between the population and the average income. The community structure is also crucial to the effectiveness of the national poverty reduction program . The community's reaction to the policymaking process is an externality of the policymaking process, which may have both beneficial and detrimental effects. A policy is deemed beneficial if it addresses public issues and improves the welfare of the community. In contrast, a policy is deemed negative if it seeks to exclude a certain group while benefiting another group . There are two approaches to reduce poverty possible losses caused by associated parties. First is presenting the qualities and dimensions of building partnership, strategies for implementing the partnership, the individuals participating, the duration and location of the partnership, and the method for its implementation. The second model is an objective-based partnership framework for partnerships. The approach focuses on policy, public services, infrastructure, capacity building, and economic growth. Efforts to alleviate poverty may be characterized as economic development goals. Evaluate the effectiveness of partnerships, ones should utilize the dimensions of equality, social inclusion, and empowerment as metrics . Another model to enhance the welfare of the society is asymmetric policy model, which implementing distinct programs for different regencies/cities depending on their separate problems. Positive discrimination policy model to accomplish justice and equality, because distinct tribes, languages, geographies, characteristics, cultures, and backgrounds necessitate different policies to achieve the same objective . --- Border Area The border areas are fundamental manifestation of a country's territory. Problems in border regions tend to complicated dimensions . A border is the line that divides one country from another or, in the case of internal entities, one province or municipality from another. B orders function to contain individuals inside their own territory, restrict, control, and regulate their interactions . B order communities are both de facto members of one sovereign political entity and community members that reside on the territory of a neighboring country. Consequently, the link between people and border areas of the two nations is essential . Border communitites have an unique human milieu created by their physical distance from the center of nation and ongoing exposures to transnational processes . Disparities in race, class, culture, and the degree of cross-border and interethnic interaction exist in every border resident. Consequently, despite the fact that all borders have functional commonalities resulting from cross-border exchanges, the border residents nevertheless maintain different identities depending on their particular local contexts . --- Methodology This study focuses on poverty alleviation within the national borders between Indonesia and Democratic Republic of Timor-Leste. The policy comparison study must examine both areas' regional settings with their diverse policy creation and implementation processes. The comparative study of poverty alleviation measures gains its impetus when residents in both nations have similar sociocultural traits. This study employs a qualitative methodology using various approaches. Observations were conducted using a participatory approach, where the researcher extensively engaged in the field, the study process, and both countries' citizen forums. The selection of research subjects is based on the subject's authority over the program and comprehension as social activists targeted by the poverty alleviation program. Consequently, the researcher expected to collect data pertaining the comparative poverty reduction. The informants were selected individually and collectively on the ground of their direct individual or collective involvements in socialization and implementation of poverty reduction policies. The researcher conducted interviews with respondents and Focus Group Discussions with informants and essential stakeholders such as the Regent of Belu, the District Head, the Head of Social Service, the Head of Bappeda, local people, and NGOs in both countries. The informants from Belu District residents consist of the Mayor of Belu city, Head of Badan Pembangunan Daerah , Head of Dinas Sosial, some village chiefs, local dan social activists from NGO CIS Timor, while the informants from Bobonaro District including Municipiu Administrador , Xefe Do Suco , Director of Ministerio Solidaridade E Social , Planu Nasional Dezenvolvimento , local residents, and social activists. This study started with a literature review to identify problems related to poverty reduction policies, followed by field observation. The researcher traced the process of field observations using a snowball technique. Initially, the researcher visited prominent community members before turning to other actors. The researcher then solicited recommendations from important informants, who nominated additional informants deemed significant roles in influence and character. The snowball technique helps ensure that the acquired statistics adequately and comprehensively describe the inhabitants in border regions. The researcher then did a document study to gather textual data pertinent to this investigation. The researcher used books, journals focusing on community conditions and poverty within national borders, and materials related to policy experiences in national borders across the globe to support the findings. The data were analyzed by sorting particular data to derive generalizations. The unit of analysis in the study focused on the micro and meso levels. At the micro level, the researcher examined individual participants and interviewed them to understand their perspectives. Afterward, the researcher examined the poverty alleviation paradigm in each country's community at the meso level. --- Results In-depth field research reveals three interconnected and supportive pieces of evidence: a comparison of different state's policies in reducing poverty and the sociocultural transformations of border communities. --- Different Portraits of State Policy Inequality in Alleviating Poverty --- Different models and forms of management In Indonesia, the poverty reduction policies in important areas are administered centrally under the National Team for the Acceleration of Poverty Reduction , which integrates coordinating and relevant ministries' work programs, through cooperation with the Committee for the Acceleration of Priority Infrastructure Provision , the National Border Management Agency , the Central Statistics Agency , community, and business actors. Several national poverty reduction programs are still implemented in Belu district, including the National Social Security System , the Integrated Social Protection System, and the Family Hope Program , Healthy Indonesia Program , Smart Indonesia Program , Prosperous Rice , Non-Cash Food Assistance , Village Fund Program, Agrarian Reform and Social Forestry Program , as well as APBN-funded Pre-employment Program , 2021). The Indonesian government provided a health assistance for the community namely Health Operational Aid . The government through APBN and APBD also provided funding for building physical and health facilities using P hysical and Nonphysical DAK fund. Unfortunately, t he government's allocation of APBD and APBN funding, particularly in the health sector, is often uncertain. The annual proportion of healthcare funding varies from 10 to 30 percent. The provision of health fund is always available annually although the amount of available fund is quite tiny. The ratio of healthcare workers to the population is not yet proportional. The number of health facilitators is still quite low in terms of quantity. The majority of doctors on duty are general practitioners. There are around nine dentists in Belu district, which is a rather small amount. Belu district has Puskesmas facilities, 14 sub-districts. Thus, one or two sub-districts have more than one puskesmas. Kakuluk Mesak sub-district which have three health facilities. Each Puskesmas is equipped with two ambulances for transportation purposes. Each puskesmas employs one to two physicians and one midwife. The Health Office is directly responsible for coordinating the placement of doctors and other medical personnels . --- Timor Leste: Policy centered on regional districts The Bobonaro district of Timor Leste is among districts that adopt pre-deconcention policies in preparation for decentralization, therefore the municipality and village birocracies have implemented several programs and initiatives. Oth at the national and local levels. In addition, several NGOs, as well as international organizations or agencies help run the programs implementation help from other. Several schemes, including irrigation enhancement, warehouse construction, and seed distribution to farmer organizations, distribution of hand tractors to disadvantaged agricultural sector groups, as well as training for farmer groups to create mini-markets in various locations around the district have been organized. The Ministry of Social Affairs' subsidy program is one of the most useful initiatives for underprivileged populations in the area. This subsidy is for senior citizens and veterans. In addition, the subsidy is also allocated for the families of Timor Leste martyrs and heroes, orphans, and disabled groups, as well as scholarships for students. Moreover, health and education aid is provided in the form of ffree access to health services and education, as a political policy to facilitate low-income families with access to education and health. Government made a policy to make public schools and healthcare in Bobonaro area of Timor-Leste free of charge. Mothers, children, and toddlers are prioritized in health programs due to the government's plan to give better access to health care services for vulnerable RTM groups . The policies accomplish the intended goals using both centralized and decentralized approaches. Centralization and decentralization constitute a policy continuum. Centralization concentrates all power on a few managers or those at the top of an organization's hierarchy, while decentralization delegates responsibility, power, and resources from the central government to local governments . The execution of the aforementioned policies at the Regional Apparatus Operations level in Indonesia tends to be plagued by difficulties. The decentralization program intended to serve as a framework for local administration in Indonesia is not functioning as intended. The central government's attempts to regulate policies for national borders diminish the local Belu District government's roles and reduce the area's potential. On the other hand, the poverty reduction programs at the local level in East Timor have been accelerated through partnerships with national and international non-governmental organizations. Non-governmental organizations and international organizations seeking to alleviate poverty have attempted to establish programs such as saving and loan cooperatives . One example is Lanamona Credit Union, which has recruited 1,678 vulnerable groups and impoverished families in the Bobonaro District as saving and loan cooperative group members. Source: Union Credito Imprestimo Lanamona --- Socio-Cultural Changes That Hinder Development Between Maintaining Tradition and Continuing Life Communities within the national borders of Indonesia and Timor-Leste have adhered to conventions and traditions. The states' rules may divide communities based on their territories, but the communities within the borders may still hold similar traditional values and customs. Gotong royong , a jargon popularized by the Indonesian government, has become an integral part of people's lives in both areas. Family ties are one of the links that bind people within national borders together. Family bonds have been a linchpin of cooperative heritage, although these communities are divided by territory and territorial borders. Moreover, the inhabitants of the border territories stated that blood ties exist despite the physical division. Culturally and socially, the national border communities' strong blood ties eliminate the stigma associated with "citizens of the exodus from East Timor" so the residents can assimilate into the cultural unity of "Eastern people" of Melanesian ethnicity. The social solidarity of border communities in both countries has taken root and knows no state boundaries. Culture and tradition are reflected in "Belu" term. Our research informants stated that the communities within the national borders primarily work in farming and animal husbandry, which are practiced traditionally using belu and the spirit of togetherness. The communities still follow the customary laws, so initiatives approved by local elders tend to be successfully implemented . These communities' blood ties and traditional customs indirectly changed the communities' way of life. The national borders in the two regions are also still very underdeveloped and isolated, lacking infrastructure and facilities, except for the Motaain cross-border post in Silawan Village, Belu District. Access to border regions, such as in the border regions of Asulait and Fatumea, is difficult. Thie lack of accessibility hinders the development of the economy, which in turn causes the residents' low-income and poor households. To improve their economic situation, residents and refugees in Belu District and Bonobaro District often engaged in illegal activities, such as gasoline, food, cattle smuggling, and gambling. Smugglers frequently collaborate with security services to send prohibited products through mouse paths . --- The question of custom is a matter of money Belu and Bobonaro Districts are geographically separated by border, yet their cultures and traditions bind them together. Timor island residents C ustomary entities as relics of their ancestors, are holy "pamali" . Every customary activity, like marriage, land negotiations, and so on, requires that the tribes involved give the amount of money or property established by the Na'i . Another example is related to belis in marriage. Belis was formerly a plague for Timorese. The belis or dowry will be changed based on the bride's socioeconomic status. Bride grooms often had to provide belis of 40 cows to propose a bride. If the bridegroom is not able to pay the customary belis, he needs to borrow . Sociocultural changes in border areas influence the border communities' survival strategies. The community is eager to help each other things related to their social group and group's survival. Nevertheless, they may not fully help in customary items related to private matters, such as belis or responsibilities in a direct family. One informant suggested that neighbors may only provide assistance by lending money as belis. --- Inequality of Access in the Economy, Health, Education and Infrastructure sectors The World Bank and the Organization for Economic Co-operation and Development indicate that Indonesia's gross domestic product per capita between 2000-2016 increased by 329 percent while Timor-Leste's increased by 250 percent. . In 2000, Indonesia's GDP was $175.8 billion, with a per capita GDP of $830.58. In 2016, the gross domestic product was $932.26 billion, with a per capita GDP of $3,570.29. Following a vote in 1999 that led to East Timor's independence, the country's GDP in 2000 was US$350.39 million, with a GDP per capita of US$402. Timor-Leste's GDP in 2016 was $1.78 billion, with a per capita GDP of $1,405.39 . Tables 2, 3, 4, 5, 6 and 7 shows the communities' access encompassing the rights to health, education, and a reasonable standard of life in Belu District, Indonesia , 2021b): ''The poverty in Timor-Leste is multi-dimensional. Rural areas have a higher poverty rate of 70 percent, which is 55 percent greater than the urban rate of 29 percent. The categories with the greatest multi-dimensional poverty rate are children, elderly , and adolescent males aged 15 to 17 years old. It can be implied that adult women are more likely to endure hardship and poverty than males. This situation is comparable to what occurred in the municipalities of Ermera and Ainaro, which have the greatest number of impoverished people, with poverty rates of 75% and 68%, respectively shown in Tables 8,9, 10, 11. --- Discussion State policies on poverty reduction in bordering nations have spawned various management approaches and structures. The Timor-Leste government's policies differ from those of Indonesia because the policies are based on regional districts. Even though these two nations have distinct poverty alleviation strategies, the sociocultural issues in each border region are not nearly as concerning as the achievement of a poverty alleviation strategy. The national border communities' similar traditions and cultures may tie the cultural commonalities within the community, and each region's reactions to the applicable rules vary depending on the policies applied. This research demonstrates that centrally determined policies in a government system caused limited public involvement in decision-making. It turns out that within national borders, poverty reduction measures that engage the public will be able to satisfy their demands. Aspirations that result from direct community engagement stress the immediate satisfaction of all needs and solutions to challenges experienced. This tendency is consistent with Wong's perspective, which examined community poverty using two models: the needs and capacity approaches . Wong addressed poverty using the preventive aspect via legislation. This model may then be used to implement asymmetric policies , where distinct programs are implemented depending on the challenges of each area. Multiple program types in the Belu district tend to conform to a centralized policy structure, as shown by the number of program variations managed by the central government under the National Team for the Acceleration of Poverty Reduction . The lack of public engagement in policymaking has persistent and serious structural and cultural consequences. Thus, establishing the community as an engaged subject rather than a passive subject for obtaining social aid is seen as more significant. The comparative analysis of border area policies and other poverty reduction policy studies showed that the distinction resides in the effect of implementing policies and policy execution . Findings on the poverty reduction policies reveal a significant gap in asset ownership at the family level and difficult access to the state's assistance programs. Consequently, the impoverished circumstances in Bobonaro and Belu might be used as factual proof of the failure of government policies . However, poverty alleviation in Bobonaro is viewed as significantly more effective and collaborative due to the growth of citizen cooperatives driven by non-governmental organizations that can promote food security at the family level and use household funds more productively and constructively. This research demonstrates that externalization policies, both in the form of centralization and decentralization, affect sociocultural transformations. Issues pertaining to poverty reduction are becoming more pressing in light of restrictions on access to social assistance, disparities in the administration of state programs, and the historical problem of the state division of sovereignty. Poverty alleviation policies cannot be separated from the structure of society since they play a crucial role in the achievement of the national poverty reduction program . Nonetheless, the culture and traditions shared by these two nations have the same ancestry and origins, and at present, customs and culture are closely associated with financial spending. The marriage system, which demands a substantial amount of money, and the street gambling mentality, which prioritizes profit, are among the social and anthropological aspects contributing to inequality in the national border communities. This situation contradicts Foxley placed the responsibility for developing poverty alleviation programs only on the policy sector, while at the sociocultural level, it encourages families and people to act differently . Through a sociocultural lens, poverty in the border regions of Indonesia and the Democratic Republic of Timor-Leste is consistently influenced by prevalent practices and traditions . The findings of Siburian and Sudi and Rini , which suggested that culture has a substantial impact on poverty, are still applicable today . However, the issue of poverty cannot be seen just through the lens of social and cultural norms; there is also a structural component to the poverty alleviation process. The principles of "willingness to assist" and "gotong royong" among ethnic groups in social concerns explain the growth and development of social solidarity in border regions. Due to the inadequate utilization of the partnership dimension, however, the partnership that promotes equality, social inclusion, and empowerment tends not to cooperate with central and local government programs . --- Conclusions In summary, the governments' assumptions about implementing community assistance programs in border regions were incorrect. Centralized policymaking and planning diminish the necessary dynamism of the fundamental demands of the people within the national borders. However, decentralized policy planning in the district does not adequately address family-level community issues. It occurs due to the lack of community engagement in defining the beneficial policies. This research demonstrates that the ultimate objective of poverty reduction policies in border regions is contingent on the community's desire for change. Unfortunately, the central government imposes the necessities and the welfare of the community without considering the society's social and cultural evolution. This research underlines the need to recognize people's sociocultural changes in national borders owing to the restrictions of the family-scale economy, as well as the significance of relying on local wisdom when formulating policies to alleviate poverty. Inequality in border regions is not just a systemic issue but also a sociocultural one that policymakers must consider. Ultimately, the researcher acknowledges that this work is limited in several ways, including a small sample area. The researcher also has not been able to validate policy considerations in border regions at the central government level. In accordance with this, further research is required to accommodate the variety of societal issues in border regions and direct input from the central government in developing poverty reduction programs for national borders. --- Declaration of conflicting interests The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Reducing the poverty gap in border regions is proven to bring structural and cultural obstacles. The poverty problem permeates social life in the two border regions between Indonesia and Timor-Leste. It is essential to conduct a research on Indonesian and Timor Leste governments' policies in solving poverty problem in the border regions and the policies' influence on enhancing the social welfare of the community. The current study evaluates poverty reduction initiatives in the border region between Indonesia and Timor-Leste. The research employs qualitative methods to collect data, including observation, interviews, recording, and Focus Group Discussion (FGD) procedures. Observations are used to collect data on border region residents' behaviors. Several informants were interviewed after being picked by snowball sampling. Information on poverty rates and per capita income is obtained using documentation techniques, while FGD method was used to collect extensive data on the implementation of poverty reduction strategies and their effect on the community. The findings suggest that neither centralized nor decentralized poverty reduction strategies will be successful until the community is involved and the surrounding culture is altered. Incorporating cultural and local knowledge into poverty reduction initiatives is the best way to alter the social structure. Thus, this study should be used as a basis of poverty reduction policymaking.

Background In Germany, approximately 2000 children and adolescents under the age of 18 are diagnosed with cancer each year [1]. Leukemias are the most common malignancies in children and adolescents, accounting for approximately 33% of all cancers, followed by brain tumors . Other common malignancies in childhood are soft tissue sarcomas and bone tumors [1,2]. Because of more differentiated diagnostics and standardized treatment protocols, survival rates have increased significantly in the last several decades [3]. Across all cancers, the 5-year survival rate is 85%. The probability of living more than 10 years after diagnosis is only slightly below this value [4]. The prognosis of brain tumors and sarcomas is worse than that of leukemias and highly depends on localization, tumor size, pathology, and possibilities of tumor removal [2]. Due to this success of treatment, cancer in childhood has changed from an acute life-threatening to a curable illness. However, the price for this cure often lies in a not inconsiderable rate of long-term consequences to which not much is known yet. Thus, late effects addressed by patients have gained attention in pediatric oncology research. In addition to the consequences of chemotherapy and radiation , other effects may develop later, such as fertility disorders, metabolic disorders, secondary malignant tumors, cognitive impairments, and cardiac problems [5][6][7]. Some of these conditions even develop years after the end of treatment. Even young adults who have been treated for cancer in childhood still report neurocognitive impairments and reduced vitality and suffer from sleep disturbances and fatigue [8]. These issues require an adaption of one's lifestyle, for example, a reduced workload with respect to school, work, or even leisure activities [9,10]. The effects for children and adolescents are particularly serious, as they undergo important developmental phases during cancer treatment. Children's ability to participate in social activities can be considerably limited [11]. However, interacting with peers is a fundamental component of children's and adolescents' development of social skills and competencies [12]. To avoid disadvantages in psychosocial development, it is important to quickly reintegrate children and adolescents with cancer into social life. There are already a number of studies that focus on the impact of childhood cancer on quality of life [13,14]. Social dimensions of health, such as activity and participation, have rarely been investigated in pediatric oncology. Here, activity means the concrete execution of an action and participation the involvement in life situations. To date, no reliable results, neither international findings nor those for Germany, are available for use in estimating whether and to what extent the disease and treatment are associated with restrictions in social participation and the factors that mediate this effect. Furthermore, it is unclear to what extent social inequalities contribute to better or worse disease management by influencing personal and social factors. --- Study objectives The aim of the study is to investigate the influence of social determinants, particularly the socioeconomic position of the parents, on participation and activity in children and adolescents between 10 and 18 years with leukemia, brain tumors and sarcomas. Furthermore, personal and treatment-related factors and their effects on participation will be explored. Our study hypotheses are as follows : 1) Participation and activities in children and adolescents with leukemia, brain tumors and sarcomas vary during and after cancer treatment in relation to the socioeconomic position of the parents. 2) Personal, social, and treatment-related factors are associated with the participation and activity of children and adolescents with leukemia, brain tumors and sarcomas during and after cancer treatment. 3) The socioeconomic position of the parents influences personal, social, and treatment-related factors of children and adolescents with leukemia, brain tumors and sarcomas during and after cancer treatment and may therefore explain the findings for participation and activity. 4) Personal, social, and treatment-related factors as well as participation and activity are related to quality of life in children and adolescents with leukemia, brain tumors and sarcomas during different phases of cancer therapy and after treatment. 5) Predictors can be identified from personal, social, and treatment-related factors that can already be used to estimate a risk of low participation during cancer treatment. Primary endpoints are social participation, activity, and quality of life. Secondary endpoints are illness perceptions, self-concept, self-efficacy, sense of coherence, social support, coping, optimism, psychosocial problems and strengths, mental health, fatigue, psychosocial needs and evaluation of the treatment. --- Methods/design The study protocol complies with the STROBE guidelines for the reporting of observational studies. --- Study design The SUPATEEN study is designed as a multicenter, longitudinal, prospective, observational study in Germany. The patients and one of their parents will be interviewed during the first month after diagnosis , at the end of intensive treatment , and half a year after end of intensive treatment . At t1, paper-based data collection is conducted in the hospital. At t2 and t3, participants will have the choice to receive the questionnaire via mail or complete it online. The equivalence of written and electronic data collection is well-documented [15]. For the internet-based survey, the software CHES will be used. The software has already proven itself in practice and is supported by the European Organisation for Research and Treatment of Cancer [16]. --- Eligibility Patients aged 10-18 years of both sexes who are newly diagnosed with confirmed leukemia, brain tumors or sarcomas of all stages of cancer are eligible for this study. They must be admitted for the treatment of their disease in one of the participating study centers. Additionally, we include one parent of each patient. Parents must give written consent for the participation of their children if they are under the age of 18. The exclusion criteria are as follows: a) having a relapse or secondary tumor, b) insufficient command of German, c) profound cognitive and physical impairments, and d) no written informed consent. --- Recruitment Any patient meeting the inclusion criteria will be informed about the study within the first month after diagnosis by the responsible clinic staff at the study center. After explaining the content and objectives of the study as well as the voluntary nature of participation and adherence to the protection of data of children and adolescents and their parents, the clinic staff will distribute the questionnaires. Completed questionnaires are returned to the clinic staff in a sealed envelope. Patient enrollment will last for 18 months. Reminders, questionnaires, and login details for follow-ups will be sent from the study center in Halle. In cases of refusal, patients are asked to provide the following information: their age, gender, disease type and reasons for declining participation for responder analyses. --- Sample size The inclusion of many study centers and a large study population facilitate the recruitment of an adequately sized sample. The participating clinics treat approximately 470 children and adolescents between the ages of 10 to 18 years with leukemia, brain tumors, and sarcomas per year. With a recruitment time of one and a half years, approximately 700 patients are eligible for the study. A conservatively estimated response rate of 70% [17] and a further 30% loss to follow-up [18] will leave approximately 340 complete cases. With this sample size, we can test our hypotheses and include a total of 30 independent variables and confounders [19]. --- Instruments Data collected during each survey time point are detailed below and in the Table 1. --- Sociodemographic and clinical data Sociodemographic characteristics will be assessed via self-report with a standardized inventory. Clinical data will be ascertained from the medical records and include information about ICD-10 diagnosis, disease stage, current treatment, and comorbid conditions. --- Social participation and activity To evaluate social participation and activity, we will use the Child and Adolescent Scale of Participation [20,21]. This self-report questionnaire measures the extent to which children and adolescents participate in home, school, and community activities in comparison to their peers. The instrument consists of 20 items that form 4 subscales . --- Evaluation of the treatment The Questionnaires of the Evaluation of Treatment evaluate the therapy and quality of treatment from the children's and parents' perspectives [22]. The children's and parents' version consists of 20 items that can be summarized in a total score or scores on 3 subscales . --- Self-concept To assess self-concept, we use 3 dimensions from the short version of the Self-Description Questionnaire [23,24]. The questionnaire measures the self-concept of children in different domains via self-report. Every subscale consists of 3 items. --- Fatigue We will use the fatigue scale from the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire [25] to ascertain patients'fatigue symptoms. --- Social support Social support will be measured with the Social Support Scale [26,27]. The 8-item self-report instrument assesses support in terms of showing affection, listening, providing information, and engaging in activities together via self-report. --- Illness perception The revised version of the Illness Perception Questionnaire measures individual beliefs and feelings about an illness and is based on Leventhal's self-regulatory model [28]. We will use the following 3 scales: personal control, treatment control, and coherence. --- Self-efficacy The generalized self-efficacy scale consists of 10 items and assesses self-beliefs about coping with difficult demands in life [29]. --- Optimism We will use the 8-item scale of a positive attitude toward life from the Bern Subjective Well-Being Questionnaire for Adolescents to evaluate optimism [30]. This scale ascertains a generally positive attitude as well as one's personal conviction to lead a good life. --- Autonomy To evaluate children's and adolescents' autonomy, the 5item autonomy scale from the Kidscreen questionnaire will be used [31]. This scale measures the opportunities to create social and leisure time and will be summarized in a total score. --- Psychosocial problems and strengths To assess psychosocial problems and strengths, we will use the Strengths and Difficulties Questionnaire for the self-report of children and adolescents [32]. The instrument consists of 25 items equally divided across 5 scales . --- Familial burden Familial burden will be ascertained with the short form of the Impact on Family Scale [33]. The items are summarized in a total score and include questions about the general negative impact of parents, social relationships, and financial burden. --- Family resources A shortened version of the German adaption of the Family Environment Scale will be used to assess family resources [34,35]. The instrument contains 12 items that form 3 subscales and a total score. --- Psychosocial needs To evaluate the psychosocial needs of the parents, the short form of the Supportive Care Needs Survey will be used [36]. In this 34-item questionnaire, unmet needs will be reported in 5 domains . For this study, the sexuality scale was omitted. --- Satisfaction with life The Satisfaction With Life scale is a 5-item instrument that measures global life satisfaction and subjective wellbeing [37]. Answers will be summarized in a total score. --- Doctor-parent relationship The quality of the doctor-parent relationship will be assessed with the Patient Reactions Assessment [38,39]. The instrument contains 15 items that are equally divided across 3 scales . --- Quality of life The KINDL questionnaire will be used to ascertain children's and adolescents' health-related quality of life [40]. Parents' quality of life will be assessed with the 12-item Short Form Health Survey [41]. The instrument can be summarized in a total score for physical and mental health. --- Sense of coherence To assess sense of coherence in children and adolescents, the Children's Sense of Coherence Scale was used [42]. The instrument contains 12 items about children's sense of comprehensibility, manageability, and meaningfulness, which will be summarized in a total score. Parent's sense of coherence will be measured with a short form of Antonovsky's Sense of Coherence Scale [43]. --- Coping The coping strategies of children and adolescents were evaluated using the Coping with a Disease questionnaire [44]. The 28 items of the instrument form 6 scales and an overall rating of the disease management. The coping skills of parents will be assessed with the Coping Health Inventory for Parents [45]. This questionnaire measures parental coping with chronic childhood disease with 45 items that form 3 scales . --- Mental health The Children's Depression Screener will be used to assess depressive symptoms in children and adolescents [46]. The 8 items of the instrument are summarized in a total score. Parental mental health will be evaluated with the Hospital Anxiety and Depression Scale , which contains 14 items [47]. --- Statistical analysis All collected data will be checked for consistency, validity, and missing values. Descriptive statistics will be calculated for all study variables, and their correlations will be explored. Therefore, the strength of the association between the different independent variables and participation can be examined. Differences in social participation and activity, depending on the socioeconomic position of the parents, will be conducted using multiple linear regression and variance analyses. Both will be controlled for confounders such as age, sex, stage of disease, disease site, and treatment-related factors. In addition, multiple regression analysis will also be applied to identify which personal, social, and treatment-related factors are associated with participation and activity and whether the socioeconomic position of the parents is associated with these intermediate factors. Depending on the frequency and type of missing data, listwise deletion of patients with missing data or appropriate imputation techniques will be applied. --- Bias control To examine selection bias, we will compare responders and non-responders on sociodemographic and clinical characteristics. The loss to follow-up will be analyzed using more detailed personal and medical information. To keep missing responses to a minimum, patients with no response will receive questionnaires at most two times via mail and will be contacted by telephone and offered the opportunity to answer the questions in written form, online or by telephone. If the participant still declines further participation, he or she will be marked as an inactive participant, but the status can be changed back to active any time the participant changes his or her mind. Additionally, we will document the reasons for nonparticipation in the study. --- Ethical matters and data protection The study is conducted in accordance with the Declaration of Helsinki. Ethical approval was obtained from the Ethics Committee of the Medical Faculty at Martin Luther University Halle-Wittenberg and from the Institutional Review Boards of each participating institute. Written informed consent will be obtained from all patients and their parents before enrollment. All personal information is subject to professional discretion and data protection. Confidentiality is ensured by using pseudonyms with each questionnaire and case report form. Patient reported outcomes and clinical data will be stored separately from person-identifying information in a locker. The allocation list will be saved in a manner such that it is physically unlinked to the other data. The data will be accessible only to authorized study staff. The study has been registered at clinicaltrials.gov . --- Discussion For the first time, the study provides detailed results on the influence of social determinants on the social participation and activity of children and adolescents with leukemia, brain tumors, and sarcomas in Germany. Since these represent the three most common cancers in childhood and adolescents , meaningful insights can be gained for a large group of patients. Children and adolescents with impairments in social participation and quality of life can have health and emotional problems that should be recognized early. In addition, age-appropriate development can be hindered by limited interactions with peers. Thus, it is necessary to identify children and adolescents at risk of impaired social participation early in the course of treatment so that appropriate interventions can be initiated. In addition to the identification of risk groups, intermediary influencing factors are identified that can be used to explain inequalities in participation and activities. --- Trial status The trial is ongoing. Patient enrolment has not yet started. --- --- Abbreviations T: time point --- --- Funding The study is funded by the German Cancer Aid . The funding. source has no role in this manuscript and will not be involved in any stage of the research process. The funding body has peer-reviewed the study as part of the award process. --- --- Consent for publication Not applicable. --- Competing interests There are no competing interests, neither financial nor non-financial. ---

Background: About 2000 children and adolescents under the age of 18 are diagnosed with cancer each year in Germany. Because of current medical treatment methods, a high survival rate can be reached for many types of the disease. Nevertheless, patients face a number of long-term effects related to the treatment. As a result, physical and psychological consequences have increasingly become the focus of research in recent years. Social dimensions of health have received little attention in health services research in oncology so far. Yet, there are no robust results that allow an estimation of whether and to what extent the disease and treatment impair the participation of children and adolescents and which factors mediate this effect. Social participation is of great importance especially because interactions with peers and experiences in different areas of life are essential for the development of children and adolescents. Methods: Data are collected in a longitudinal, prospective, observational multicenter study. For this purpose, all patients and their parents who are being treated for cancer in one of the participating clinics throughout Germany will be interviewed within the first month after diagnosis (t1), after completion of intensive treatment (t2) and half a year after the end of intensive treatment (t3) using standardized questionnaires. Analysis will be done by descriptive and multivariate methods. Discussion: The results can be used to identify children and adolescents in high-risk situations at an early stage in order to be able to initiate interventions tailored to the needs. Such tailored interventions will finally reduce the risk of impairments in the participation of children and adolescents and increase quality of life. Trial registration: ClinicalTrials.gov: NCT04101123.

INTRODUCTION Over the last decade teen pregnancy and birth rates have declined in the United States overall, yet the U.S. sustains the highest teen birth rates among comparable countries [1]. Among adolescents in the United States, socioeconomic inequalities and poverty at the individual, household, and community levels are associated with early pregnancy and childbearing [2], [3]. Further, there are ethnic disparities: the birth rate for 15 to 19 year old females is highest among Latinas and is more than double the rate of non-Latina white teens [4]. In states with large Latino populations, teen birth disparities are particularly pronounced. In California, 74% of teen births are to Latinas, yet Latinas constitute 50% of the population [5]. Culturally-based childbearing norms, such as the importance placed on motherhood in Latino culture, may contribute to the difference in teen birth rates by ethnicity [6]. However, these norms, including childbearing expectations, evolve in neighborhood and social environments shaped by limited socioeconomic opportunities [7], [8]. There is a need to more fully understand avenues for addressing upstream, structural factors associated with pregnancy and birth rates in Latino youth. National studies suggest that Latinas living below the poverty line have double the rate of unintended pregnancy compared to non-Latina white women [9], [10]. Previous research has also documented the inverse relationship between educational aspirations and attainment and adolescent pregnancy and parenting, including adverse implications for future financial stability and wellbeing [11]- [14]. Compared to non-parenting youth, teen parents are almost 40% less likely to earn a high school diploma or GED, particularly Latina teen parents [15]. However, the relationship between economic disparities and adolescent pregnancy is multifaceted. Findings from national data suggest teens living in neighborhoods with high levels of income inequality may be more likely to become teen parents, but that teen parenting does not "cause" poverty, rather economic depravity gives rise to social conditions that persist and differ between youth who do and do not become teen parents [16]. There is a growing demand for comprehensive interventions that incorporate approaches to tackle the socioeconomic factors associated with adolescent pregnancy, acknowledging the need to eliminate barriers to educational, job, and health resources that impact poor and immigrant youth [17], [18]. Adolescent pregnancy and parenting among Latinos requires countering structural inequities and continue to present a complex public health challenge. Promoting protective social network ties is one asset-based approach to encouraging healthy sexual behavior among young adults living in marginalized communities [19], [20]. This approach draws on a resilience framework, an exploration of how positive outcomes prevail despite threats, such as abuse or poverty, to healthy development [21]. Protective factors are a key aspect of resilience and can include elements that can help youth avoid and/or moderate the negative effects of risks and cope successfully with adversity [21]- [23]. Protective factors may be individual assets and resources in the social environment [22], [24]. The role of social network ties, including among partners, peers, and family members, has been explored both as a phenomenon of normative behavior as well as a target for leveraging intervention effectiveness [25]. Christakis and Fowler suggest that the interaction of social network ties is greater than the sum of its parts and can affect individual behavior related to, for example, happiness, weight gain, and partner choice. Previous research has found social norms, shared behavioral expectations among network members , to be associated with adolescent pregnancy among Latino youth [26]. However, relationship ties do not develop in isolation: the physical and social environments in which they form shape network norms, including those associated with adolescent sexual behavior [26]- [28]. A strengths-based and contextual level approach to intervention design fosters community building and offers a departure from a conventional public health focus on deficits and individual risk behaviors, which has been shown to fall short in terms of promoting change in youths' sexual behaviors [29], [30]. To examine the relationship between protective social network ties and norms, neighborhood features, and individual childbearing expectations, we conducted a qualitative study with Latino youth in partnerships from a single neighborhood. A qualitative approach can compliment quantitative studies and bridge some of the challenges of assessing network and neighborhood norms, including their interaction, using aggregate survey and census tractlevel data [26], [27]. We chose to explore childbearing expectations rather than focus on more proximal behavioral correlates with adolescent pregnancy, such as contraceptive use, to assess a constellation of potential contextual and upstream social environment factors. This paper poses two research questions. First, how do Latino youth articulate childbearing expectations in terms of timing, goals, and future orientation? Second, are individuals' childbearing expectations aligned with those that dominate their social environment? To assess social environment and individual childbearing expectations, we examined partner and family expectations as well as neighborhood norms that may facilitate or impede attainment of pre-parenting goals. --- METHODS --- Study Design and Community Mi Cuento was a qualitative study conducted with young Latino women and men in San Francisco, California. This study was part of a joint academic and community partner research program initiated in 2001 in the Mission District to examine social environment influences on adolescent sexual health. Like other urban neighborhoods, the Mission District has been undergoing gentrification for decades and subsequent displacement of local businesses and low-income families leading to an increase in health and socioeconomic disparities, and, in particular, the social exclusion of non-white youth [31]- [33]. Home to the largest Latino community in San Francisco, with dozens of active youth agencies, the Mission continues to serve as a cultural center for families with ties predominately to Mexico and Central America [34]. The majority of underperforming schools in the city are also located in the Mission, where students are prohibited from wearing blue or red clothing to discourage visible markers of gang affiliation [35]. Recruitment-Within this context, and in collaboration with our community partners, youth were recruited from street venues as well as through referrals from youth agencies. Youth were approached and screened by trained, bilingual study staff. We stopped recruiting youth when a preliminary review of transcripts indicated we had reached saturation of key concepts of interest. Of the 33 youth who participated, 18 were recruited from venues near a neighborhood high school, and 15 were recruited from community agencies via referrals and study presentations to youth groups. Inclusion criteria included to self-identify as Latino, to be16-22 years old, to spend at least four days a week in the Mission, to have a parent born outside the U.S., and at least one romantic partner of the opposite sex. Being sexually active was not required. We chose to recruit youth between 16-22 years old, as romantic partnerships become more common during mid-to lateadolescence [36]. To consider our main research questions within the context of immigrant generation, we recruited equal numbers of foreign-and U.S.-born youth. Youth provided verbal consent as part of the audio-recorded interview. The Institutional Review Board at RTI International approved the study and waived parental consent for minor participants. All youth were recruited and interviewed between June and November of 2010. In-depth interviews-The interview guide was designed to explore how migration, time in the U.S., local and transnational social ties, and neighborhood norms are associated with sexual health, gender role norms, and partnership and childbearing expectations. The primary social network members of interest included partner and family, though peer norms were also explored. The interview guide was piloted and modified following a set of initial interviews with youth who met study eligibility. We revised or removed questions or probes to enhance comprehension and alignment with the overarching areas of interest. The guide began with closed-ended questions to collect demographic data, followed by openended questions with probes. To develop rapport, the initial open-ended questions asked about how youth in the neighborhood typically meet partners followed by questions tied to relationship dynamics and community violence. Youth were also presented with a series of statements expressing neighborhood norms about pregnancy, gang-affiliation, and educational aspirations. These statements were developed based on previous research activities as part of the larger research program in the neighborhood [37], [38]. Youth were asked whether they agreed or disagreed with each norms statement and to provide rationale for their opinion. For this analysis, we assessed one norms statement tied to post-secondary educational aspirations that was most aligned with participant descriptions of childbearing expectations. Two bilingual interviewers, a Puerto Rican man with previous experience with neighborhood schools and community agencies and a white woman who was part of the community research team since 2004, conducted interviews in Spanish or English, per participant request. All interviews were conducted at a community partner site that was safe and convenient for the youth. Names used in this paper are pseudonyms. Professional services transcribed and translated all recordings. We reviewed audio-recordings and transcripts after completion to assure quality and identify necessary modifications to the interview or data collection process. --- Analysis Using a modified grounded approach, for the first stage of qualitative analysis we created a codebook reflecting interview guide questions and emergent themes that was applied in reviewing all transcripts in ATLAS.ti [39]- [41]. Two researchers coded transcripts and made comparisons for a subset to assess consistency in coding. Study staff completed written memos and held meetings to discuss coding and synthesis of findings. For this analysis, we returned to hard copy transcripts to ensure rigor in exploring the research questions. First transcripts were analyzed for dominant themes identified via repetition , indigenous typologies , metaphors and analogies , and linguistic connectors [42]. This process consisted of open coding to assess similarities and variations in the narrative, constant comparisons of text until concepts were theoretically saturated, and the creation of categories of concepts [41]. Second, axial coding consisted of examining relationships between the categories to explore dimensions of the categories [41]. This occurred within and across narratives. Finally, selective coding served to identify core categories that tied the various categories together to "tell the story" [41]. For the community norms analysis, responses were each examined as "norms categories" to allow for axial and selective coding within and across narratives. A single norms statement that was directly linked to individual and network childbearing expectations was selected for this analysis. Field notes accompanied each interview, which were reviewed in conjunction with transcripts, particularly to assess neighborhood and gangrelated exposures. Six participants were pregnant and/or parenting at the time of the interview . To best assess pre-parenting goals, parenting participant responses were included only in assessing the community norms statement, which reflect the larger social environment in which youth, regardless of parenting status, interact and, ultimately, take part in shaping. --- RESULTS The 33 youth that participated in this study had a median age of 17 years and 52% were young women . More than half of the youth were born outside the United States with most living with their mother. Nearly one third of the participants were in a romantic relationship with a foreign-born partner and 58% reported a relationship length of at least one year. For nearly all youth, "the right time" to start a family centered on the attainment of two goals: completion of college and financial stability. While the rationale for and pathways to achieve these goals differed by gender and nativity, attainment of post-high school opportunities was considered most important by all youth. Youth reported that partners and family members generally shared these expectations and offered verbal and, to a lesser extent, active support to help youth access opportunities. Though individual, family, and partner childbearing expectations were well aligned with one another, youth responses to community norms statements about college indicated that there were significant barriers to higher education, including poverty, community violence, and immigration status. --- Individual Childbearing Expectations College completion-Nearly all of the non-parenting or pregnant youth explicitly identified college completion as a desired prerequisite before starting a family. In turn, college was viewed as necessary to obtain a desirable career. As 17 year-old Carla, who came to the U.S. from Teculután, Guatemala at 13.5 years old stated, "I don't want to get pregnant until after I graduate [from college] because I want to be a lawyer…and have everything in order. Somewhere around 32 years old is when I want to have a baby." Similarly, Juan, 17 years old, migrated to the U.S. at 13 years old from San Salvador, El Salvador. He described his childbearing expectations, "I think after you've finished college and you have-a certain good level of education…I think you can get better work. You at least will have a degree and you could defend yourself in life." College, as Carla and Juan suggest, could serve not only as a means to, but also as protection for a better, more stable future, a pre-requisite to parenthood. In addition to a pathway to professionalism, U.S.-born youth also identified college as an opportunity for personal development and exploration before assuming the responsibility of raising a family. Sixteen year-old Elena explained that she hoped to go to college in Los Angeles to "try something different": "I'm trying to live my life first…I wanna go party…go to one of those college parties." Pedro, 16 years old, shared his desire to become a teacher, but stated he wanted to start a family "around the age of 30" primarily to allow time to "[d]o as many little adventures as I can possibly do." For some, like Pedro, the appeal of post-high school exploration was as valuable as pursuing a college pathway to obtain a desirable career. Further, most youth explicitly stated that having a baby as a teenager would be a burden and interfere with their post-high school plans. Twenty year-old Erica, a community college student, was born in Mexico City, Mexico and came to the U.S. at three years old. She explained, "I don't know if I want to do law school, but-it's kind of unspoken, we're not going to go to law school with a kid." Similarly, Julio, 16 years old who migrated from La Ciudad, Guatemala at 15 years old, stated, "[If we had a baby] now, I 'd have to get a job, and I wouldn't be able to continue my education. And that's no future." Youth also described how early childbearing posed a significant barrier to reaching education goals for family members and peers. Sixteen year-old, U.S.-born, Hector commented on Facebook friends' posts about teen parenting, "Yeah, you can have kids…but what you going to do? You can't bring your kid into the classroom. Who's going to watch your kid from 8 to 3:30? Are you going to stop your project and help your kid? It's not going to happen." Though youth, like Hector, described early childbearing norms within their peer groups, they also critically assessed the "real-life" challenges of teen parenting tied to reaching future goals. Acquisition of resources-Young women, both U.S. and foreign-born, more often than young men emphasized obtaining resources, such as a house, as an important prerequisite before starting a family. Assets could be purchased through the better paying job one could obtain with a college degree. Liliana, 16 years old, migrated from Acapulco, Mexico at two years old and listed the steps she planned to take before starting a family, "Get a job. Get a career. Get a place, a good house." Sandra, 16 years old and born in the U.S., despite barriers tied to missing school credits and gang affiliation, alluded to the time needed to acquire assets, "I think the right time [to start a family] is when you're done with college… Not right when you're done but like, be done with college then decide what to do…save up money to get…a car and, you know, a little apartment." Yadira, 16 years old who came at 13 years old from San Pedro Sula, Honduras, highlighted the relationship between, education, resource acquisition and emotional preparedness, "she needs to at least have a house, and a car…money for the child's medications, for food for everything…A child isn't a toy…And that's why you have to be very prepared and mature." Such resources and assets described by these young women emphasize the significance of livelihood and financial markers or end-points of completing the higher education process prior to parenthood. Coping with social environment barriers-Despite nearly uniform aspirations among participants to complete college before starting a family, preparedness to attain goals varied significantly due to a number of obstacles experienced by some participants. Lack of preparedness was largely attributed to social environment barriers including gang involvement and documentation status. Before parenthood, Hector hoped to complete "college, definitely my shots at football, baseball and acting." However, he acknowledged, "I'm not the best student…I used to be into all that gang stuff. I don't have a good academic record." Miguel, 16 years old and gang-involved, migrated to the U.S. at seven years old from Lima, Peru. He was working to improve his grades and considering becoming an engineer before starting a family, but believed that criminal justice involvement carries a stigma impacting academic success: "if you get caught by the police, it just messes with you a lot. You get paperwork…You can't go back to school and just start doing good 'cause they'll look at you different…You feel marked." Foreign-born youth also stated that documentation status impeded youth from making headway on attaining post-high school opportunities. Twenty year-old José, who migrated to the U.S. from La Ceiba, Honduras at 17 years old, insisted that finishing college was necessary for him prior to starting a family. However, because of state guidelines for aid and entry into a University of California institution for undocumented students, he remained in high school: "they say that to go to a U.C., it's four years [for me to apply]…I have three…last year I went to apply…I couldn't do it last year, which is why I'm here." Though different types of barriers, both documentation status and gang involvement can become pervasive systemic barriers for youth, despite individual efforts to improve their future opportunities. --- Social Network: Family Shared childbearing expectations-The majority of non-parenting youth had discussed their educational and career goals with family, including parents, siblings, and extended family members often sharing the same home. Youth indicated that their individual childbearing expectations tied to education goals were aligned with those expressed by their family members. Elena, who wanted to go to college in part to "live [her] life" before having children, explained, "[my mother tells me] you always have to be safe…If you need birth control, you know, you can tell me…you're still young. You have to go to college [and] like…do things like travel." Ideals expressed by family generally reflected a desire to encourage their children or younger family members to strive for "a better life" for themselves and for their future family. Julio stated his parents told him "the best way [to prevent pregnancy] is abstinence [aguantarse]" but if "we can't abstain, then to use condoms or pills." He then explained, "my mom only finished high school, and the same for my dad…since he lived in small town [and] since he was little they put him to work after school, and then he wasn't able to continue his education. Yes [I agree with my parents' ideas], because first I need to have a solid foundation and a basic plan for the future." Family members also encouraged youth to consider pursuing higher education prior to parenting, even when adolescent pregnancy was normative. As Sandra explained, "my whole family, like my siblings and stuff, it's normal to have babies at a young age, you now. So, I'm here 16, like I don't have babies, so my mom is like happy…She's just like finish school…Go to a university, college, you know. Whatever you want to be… Don't give up…She wants me to finish school and that's what I want too." Breaking the barriers cycle-Despite support for higher education, some youth were confronted with conflicting messages from family members about the pathway to take, suggestive of the barriers to opportunities parents had faced. U.S.-born Sara, 16 years old, stated her mother, currently unemployed, supported her "going away to a four-year college," while her father, a cab driver, thought she should go to City College. With aspirations to be a nurse practitioner, Sara explained, "my dad feels that …nobody should be too good for City…I think City College is fine…it's affordable, I guess…I just wanna do everything all at once at one school…seeing my parents like work extra hard…I wanna like work hard to get a career where I don't have to worry." Youth, like Sara, had to contend with navigating complex messaging regarding family support to achieve pre-parenting education goals. Partner selection advice-With family members, ideal childbearing timing was often connected with partner selection advice: find someone who is "gonna do something with their life." Compared to family of U.S.-born youth, family members of first-generation youth more often offered partner selection advice. Miguel, for example, reported that his mother preferred that he date a "good girl" and explained, "The girl has to go to school. [She] [d]oes her work. Doesn't drink or smoke…talks nice." In some cases, family advice also included seeking a partner with citizenship status. Overall, partner selection advice that emphasized dating peers with post-high school educational aspirations reinforced broader social network norms tied to pursuing education goals prior to starting a family. --- Social Network: Partners Goals and contraception-Most non-parenting youth had discussed their future goals with their partners. These conversations, in some cases, directly prompted discussions about contraception. Eighteen year-old Victor who migrated to the U.S. from Tegucigalpa, Honduras at 16 years old aspired to go to the Marines. He reported that he and his partner use both condoms and oral contraception and had gone together to the clinic, "We always talk about it [birth control]…I always ask her whether she's been taking her pills, or if there's been anything, she'll tell me…we're not in a good financial situation because first of all, neither of us has finished high school… That's why we're trying to prevent [pregnancy] and always use protection." Friends first-The connection between pregnancy prevention to protect educational goals was also present in narratives of youth who were not sexually active with their current partner. Seventeen year-old Damaris came to the U.S. from Guanajuato, Mexico at 10 years old, "talked for a year" with her boyfriend, now in college, before becoming romantic partners. She shared about their conversations, "if we want something serious that we need to wait, because it's too early to have kids…because I'm still in school, and I want things for me, …Like college…he also wants his college thing to get done…[but]…we're not that close yet…like, not having sex." Like Damaris, most youth described their relationship evolving from "friends first" to romantic partnerships, which created the temporal and emotional space to discuss future aspirations. --- Interplay of Partner and Family Ties Protective partner and family ties-The interplay of family and partner ties related to completing educational goals prior to starting a family may create a protective environment for youth who are considering early childbearing. This relationship is illustrated by 19 yearold, Brenda, born in the U.S., who explained, "you can say we [my partner and me] have the same goals, because we both want to better ourselves…we've discussed it… you need to have money and an education and a job before having children…When I was 14, all my sisters had babies; one when she was 14, another at age 18, and the other one at 17…So when I turned 14, I wanted to have a child too…I was working, but only occasionally at night cleaning an office…But then I really started to think about it, and my sisters have also really supported me. They say, 'Look if you have child now, what are you going to do? Where are you going to get money? You need to stay in school… If he really loves you, you will take precautions to avoid having children and you'll wait.'" Brenda and her partner have gone together to the neighborhood clinic: "When I got birth control and everything, he was there with me." Overall, sexually active youth in this study reported using condoms as the primary method to protect against pregnancy and sexually transmitted infections. Opportunity linkages-Despite verbal support to accomplish post-high school education and social development prior to parenthood, very few partners or family members provided linkages to educational opportunities tied to formal programs or services. In one case, the partner of 16 year-old, U.S.-born Abel linked him to an afterschool program designed to introduce youth to careers in medicine. Abel also identified his partner as his motivation to focus on school, "at that point, I didn't really care much for school… But then I got to know her, she kinda changed my life around into me actually wanting to go to college." For Erica, though her parents faced language barriers, they attended all parent-teacher conferences, which led to private school scholarships. She elaborated, "when I had to get a computer, like even though it was really hard for them to like, pool the money to get it, like they managed to do it." In addition to few network linkages to educational opportunities, participant perceptions of neighborhood education norms did not reflect the high aspirations shared by youth, their partners, and their families. --- Community Norms: Educational Aspirations and Violence All 33 youth participants were asked whether they agreed or disagreed with the statement, "Most Mission youth plan to go to college." Despite a range in responses, participants provided nearly uniform rationale for their responses centered on three structural barriers: poverty, immigration status and the negative demands of gang affiliation. In addition, though some youth described positive peer norms related to both post-high school aspirations and adolescent pregnancy prevention, among non-parenting youth, most females and males reported having at least one friend who had been pregnant. Poverty and documentation status-For foreign-born youth, a college degree without U.S. citizenship was regarded as providing little career advantage. Sixteen year-old Jessica, who came from Nogales, Mexico at 12 years old, disagreed with the college norms statement saying, "they can't go [to college] because they don't have the money, because they're undocumented…they don't see a future. If you are undocumented, why the hell would you go to college?" In addition, familial fragmentation due to migration depleted financial and emotional support for higher education. Julio also disagreed with the norms statement and said, "most of them would like to [go to college], but there are always barriers…some people don't live with their parents and everything and some are undocumented. So they say, 'It won't help me here because I won't be able to get a job in the field I'm interested in.'" José was undecided and believed that recently arrived youth were more motivated to pursue higher education than their U.S.-born peers: "the ones who plan to go [to college] are the 'wetbacks' as they're called…they [U.S.-born youth] say, 'I already know English…' graduate and then go work at McDonald's." José believed youth born in the U.S. might also be less likely to access academic support or be recruited by high school college counselors. Poverty and gang life-Gang affiliation offered strong peer ties, particularly for some males in this study. Such allegiance often, but not always, conflicted with education goals and aspirations. Twenty-two year-old Leo came to the U.S. from San Salvador, El Salvador at eight years old and had been, like his sister, gang involved, including transnational affiliations through family. He was undecided about the college norms statement: "I hope they do. I want all the youth to, to plan to go to college, but…I know a lot of Mission youth, they planning on shooting somebody or they're planning getting shot or they're planning how the f---they gonna eat….So going to school is not on the list, let alone going to college." Leo explained that when he was younger his "list" was to "get money, stay out of jail, stay out of the cemetery." Nonetheless, he acknowledged, that despite having "homies" that are dead or "doing life," "I got homies that are doing it. I got homies in college, I got homies like me working, you know, living cool." Leo's girlfriend was training to become a medical assistant. They agreed she should complete her training and gain work experience before having children, though these plans were contingent on Leo's stipulation, "unless I get killed." Pedro, whose brother was stabbed, was also undecided about the norms statement and highlighted the role of individual motivation: "if it's somebody that has more goals for the future then yeah, but if it's somebody that just wants to die in the hood…then probably not." Yet, like with Leo's friends in college, Pedro's peer network included older gang members teaching "game" to younger members with positive messaging: "I was out late and this O.G. dude… He's like, 'You got to get that paper.' I was like, 'What paper?' He's like, 'That diploma'…And then he started rapping, and at the end of all his sentences, 'Got to get that paper.'" Future uncertainty-For youth born in the U.S., their foreign-born family members exhibited the daily reality of limited access to opportunity. Family members employed in low-wage work with long hours coupled with the fatalism endemic to gang life led some youth to question the tangibility of their future goals and the odds of waiting "too long" to have children. Eighteen year-old Alberto, born in the U.S., had applied to "barber school" and planned to save money by working a side job as a truck driver for Pepsi Cola with his cousin. He had also experienced and witnessed substantial gang-related violence. Alberto reasoned, "You don't want to have [a baby] when you're 30, because then you're just too old to have a kid. You might be too tired to do anything. And when you're young, you're energetic and ready to go." If necessary, Alberto believed that he and his partner, who had also applied to college, could alternate day and night classes to care for a baby. Abel, who had lost friends to gang-related homicides and whose parents worked opposite shifts as janitors, also felt like waiting until his 30s to start family was "gonna be too far off." He explained, "it's just the fact that like maybe there won't be enough time, 'cause like anything could happen…you wouldn't want to leave the kid by himself…like if someone were to die." --- DISCUSSION Overall, the youth in the study expressed individual aspirations to complete higher education prior to starting a family. Family and partners generally supported these aspirations. However, neighborhood norms and structural barriers posed significant roadblocks to meeting childbearing expectations. These narratives illustrate the resilience of youth given the disconcerting paradox between individual childbearing expectations and the social environment in which youth attempt to pursue their goals. Most youth not only expressed the importance of fulfilling educational milestones as intrinsic to their childbearing expectations, they explicitly articulated early childbearing as an obstacle to goal attainment. Previous research has identified this connection as a distinguishing characteristic of youth with risk profiles for teen pregnancy and suggests that pregnancy prevention efforts shift the focus from addressing unprotected sexual behavior to encouraging youth to directly examine how early childbearing presents challenges to fulfilling long-term educational goals [43]. Future research should also further examine the nuances of the relationship between immigrant generation and childbearing expectations to include, for example, emphasis on also protecting time for personal discovery as voiced by U.S.-born youth in this study. Our findings also suggest the role family members can play in supporting youth to avoid pregnancy as a means to reach educational goals. Family support can contribute to resilience by buffering the effects of structural inequities in the social environment. Previous studies with Latino youth have found protective associations between parental monitoring, ethnic pride, and communication of clear sanctions against teenage childbearing and adolescent sexual intentions and behavior [44]- [49]. Protective family ties can be vital in offering support for youth at a crossroads in their future planning which may be exceptionally complicated in families with teen parents. Few studies have examined the potential of protective influences of partners. Non-parenting youth in this study reported having conversations about educational aspirations with partners, including how early childbearing would negatively impact their ability to reach goals, and indicated that these conversations were helpful in terms of avoiding pregnancy. Previous research has identified the role partners play in contraceptive decision-making and pregnancy timing, including protective aspects of joint communication about contraception [50], [51]. Our findings suggest that promoting partner discussion that clearly connects future goals with post-adolescent childbearing, particularly during the friendship phase of romantic partnerships, may be a promising component of unplanned pregnancy prevention efforts. These findings also contribute to the development of a healthy relationship framework to inform adolescent sexual health programming [52]. Verbal social support should not be underestimated. Yet, if partners and family members could also readily link youth to resources to improve their academic preparedness and realize their goals, the potential impact of protective social ties on sexual health may be more sustainable and farreaching. Such linkages were not common in the youth narratives presented here. Again this finding raises questions about the social environment and the barriers youth identified to reaching education goals. We encourage researchers, youth providers, and policymakers committed to preventing unplanned pregnancy among Latino youth to consider structural interventions that ultimately challenge the status quo and break the cycle of an inequitable distribution of access and opportunity. --- Limitations Youth in this study were recruited drawing from a decade of ethnographic and quantitativebased venue research with the Mission community to ensure participation of youth from distinct social groups, but are not representative of urban Latino youth across the country [53]. Though their narratives can inform local programming and intervention design, we cannot generalize these findings. Youth in this study generally reported high educational aspirations, which may reflect sampling or social desirability bias and may not reflect the experiences of youth with low aspirations. Of note, high educational aspirations are not uncommon among Latino immigrant youth and youth of immigrant parents, despite legal and economic barriers to educational attainment [54], [55]. To further explore the relationship between aspirations and childbearing, a comparison of parenting and nonparenting youth or a longitudinal qualitative study of non-parenting youth would be informative. This study did not have a sufficient number of parenting youth for this comparison. Though the diversity of youth in this study speaks to the diversity of Latino youth in the Bay Area, nuances tied to country of origin and discrete levels of gang involvement could not be assessed in detail. Peer norms were explored, including protective aspects, but not to the extent of partner and family norms. A closer exploration of peer influences on individual childbearing expectations could yield important findings for interventions targeting peer networks. --- Conclusion The President's Teen Pregnancy Prevention Initiative for 2010-2015 centers on reducing pregnancy and births among target communities through evidenced-based programing and community mobilization and sustainability including accessing resources and public and private collaboration [56]. This vision is promising for Latino youth in the U.S. First, a focus on mobilizing and building local communities respects the diversity in youth experience. While large numbers of Latino youth live in poverty, Latino youth also arrive to the U.S. at different times in their lives from different countries. Unique transnational ties to culture and life histories can impact adaptation to urban life in the U.S. Second, a multi-sectoral approach encourages schools, criminal justice, and community agencies to work together with families and youth to mount a response to adolescent pregnancy prevention. Specifically, such an approach could include mechanisms to support linkages to educational resources. In cities with significant gentrification, to increase the likelihood of successful mobilization, it will also mean creating spaces for dialogue between neighbors that may coexist but have not come together to create a collective vision for local youth development. Structural interventions during mid to late adolescence can also affect social cognitive maturation related to the weighing of risks and rewards, controlling impulses, and social relationships [57]. Thus, it is vital to consider opportunities to help insulate youth by building on assets -including positive family, peer, and partner relationships -to offset deficits created by poverty or community violence that may offer competing and potentially harmful normative behavior expectations. A social environment that fosters truly tangible future opportunities for urban, Latino youth can create pathways to attain higher education and meet post-adolescent childbearing expectations. To this end, neighborhoods will also need to organize beyond their communities to effectively demand inclusive local and national policy that facilitates sustained and comprehensive higher education access. In 2012, national enrollment rates of Latinos in higher education surpassed that of white high school graduates [58]. However, compared to white youth, Latino youth are less likely to enroll in a four-year college, be enrolled fulltime and obtain a bachelor's degree [58]. Furthermore, citizenship status impacts eligibility for higher education and vocational programs. The federal Development, Relief, and Education for Alien Minors Act introduced in 2001 would allow undocumented youth, who meet certain eligibility requirements, to apply for U.S. citizenship on a conditional basis, including enrollment in higher education [59]. Though at the federal level the DREAM Act has not passed into law, individual states have passed their own Dream Acts, which, in addition to offering a pathway to citizenship also include college financial aid benefits. A multi-sectoral approach at the local level coupled with national immigration reform comprises key aspects of community mobilization as a means to address teen pregnancy. In conclusion, we highlight implications for social action and strengthsbased community intervention to promote sexual health and prevent adolescent pregnancy among Latino youth in the United States by supporting youth to fully participate in educational attainment options and contribute to building the assets of the communities in which we live.

In the United States, adolescent childbearing is disproportionately higher among Latino youth, a growing population facing substantial social exclusion. Exploring the relationship between the social environment and sexual health outcomes among Latino youth may offer insights into the development of novel interventions. In this study, Latino youth in partnerships were recruited from neighborhood venues in San Francisco and completed in-depth interviews. Youth reported a desire to complete higher education goals prior to starting a family to improve future opportunities and further personal development. Youth stated that social network members, family and partners, were supportive of their individual childbearing expectations. Social environment barriers tied to poverty, immigration status, and gang violence hindered educational attainment. Some differences were noted by gender and immigrant generation. Building on protective social ties and creating avenues in poor, urban neighborhoods for Latino youth to fully access educational opportunities may counter early childbearing and improve sexual health.

Origins of PrEP Stigma 2012 marked a watershed moment in the fight to curb the spread of the Human Immunodeficiency Virus and Acquired Immunodeficiency Syndrome . The Food and Drug Administration 's approval of pre-exposure prophylaxis bolstered HIV prevention efforts and has since demonstrated efficacy with HIV transmission reduced by greater than 90% when used as prescribed. 1 Despite the success of PrEP therapy in clinical trials, there were mixed reactions and criticisms regarding the use and promotion of PrEP when first introduced contributing to lower than expected adoption rates. Many of the critics surprisingly were among members of the LGBTQ+ health advocacy community. 2 Concerns about HIV drug-resistance, potential long-term side effects for users, and whether this new innovation would undermine decades of safer sex HIV prevention education led to stigmatization of early adopters of PrEP as engaging in risk compensation or increased HIV risk behavior through condomless sex and complacency towards the possibility of increased exposure to other sexually transmittable infections. 3,4 These concerns and others were also shared at the time by many health professionals and well-known and influential community leaders, including AIDS Healthcare Foundation President Michael Weinstein. 5 A widely disseminated article at the time popularized the phrase, "Truvada Whore," 6 which would, in turn, be reappropriated soon after by PrEP advocates in their marketing campaigns. While not representative of the all critics, such stigmatizing language and imagery reflected the highly charged atmosphere surrounding the public unveiling of PrEP. 5 Stigma is born out of existing stereotypes, biases, prejudices, and various forms of oppressive and discriminatory attitudes towards individuals and/or communities of people. The insidious impact of stigma associated with HIV/AIDS has permeated throughout the LGBTQ+ community, healthcare system, society in general and to this day, continues to exacerbate structural and social determinants of health disparities amongst sexual and gender minorities. While the initial resistance to PrEP has abated over time and many of the early critics have come to embrace it as an effective prevention tool, 7 PrEP stigma continues to impede those at greatest risk from preventive care. The Centers for Disease Control estimates that the majority of adults considered to be at greatest risk for HIV infection and who would benefit the most from PrEP are not using it. 8 Among the many harms associated with the HIV epidemic was the prevalence of HIV stigma that negatively impacted the emotional wellbeing and mental health of persons living with HIV/AIDS as well as exacerbating fear, misunderstanding, and discriminatory attitudes and practices. Often times this prejudice extended to those persons assumed to have been exposed to and/or suspected of having HIV/AIDS based not on exhibited clinical signs and symptoms, but on assumptions regarding sexual orientation, race, socioeconomic status, substance use, sexual health behavior, and even geography. 9 Laws enacted to help protect the civil and human rights of persons living with HIV, health education campaigns, and shifts in societal perspectives have significantly diminished, but have not eliminated biases associated with HIV and prevention measures. Some of the deeply embedded tensions between clinical advances and moral judgments were brought to bear following a contentious 2018 op-ed in the New York Times titled, "The End of Safe Gay Sex?" 10 In recent years, the number of adult persons living with HIV in the United States has grown to over 1.1 million individuals with estimates of 14% of those persons not knowing their HIV status. 11 Since the height of the epidemic in the mid-1980s, the rate of HIV incidence in the United States has been reduced by greater than two-thirds. After years of declining new infections, the number of new infections started to plateau in 2013. 11 In 2018, there were 37,832 new HIV diagnoses in the United States and US territories. For the 7 years prior, HIV diagnoses had decreased overall by 11% among adults and adolescents, but increased among some patient populations, 12 which is why the primary goal for national HIV prevention continues to focus on reduction in new infections. 13 Between 2010 and 2016, rates of new HIV infection decreased by 6% with notable drops among heterosexuals, persons who inject drugs, and White men who have sex with men but increased rates of infection for Hispanic/ Latino MSM and rates of infection among African American MSM remained unchanged. 14 Despite the success of PrEP, the widening gap in HIV prevention affecting Hispanic/Latino and African American MSM warrants a deeper exploration of the stigma and barriers impeding access and uptake of PrEP. Currently, there are only two FDA-approved medications for PrEP. The first, emtricitabine/tenofovir disoproxil fumarate , is sold under the brand name Truvada ® is recommended for use all adults and adolescents at risk for HIV infection. The second, emtricitabine/ tenofovir alafenamide , sold under the brand name Descovy ® was approved by the FDA in 2019 for adults and adolescents, but not recommended for those persons at risk of HIV infection through receptive vaginal sex. PrEP, when taken as prescribed, is very effective in reducing the risk of transmission of HIV. It does not provide protection against other sexually transmittable infections and blood-borne illnesses such as hepatitis C, syphilis, chlamydia, and gonorrhea. Based on a review of 14 randomized clinical trials, 8 observational studies, and 7 studies of diagnostic accuracy demonstrating that PrEP was associated with decreased risk of infection, the United States Preventative Services Task Force issued a Grade A recommendation for HIV prevention and HIV screening in adults, adolescents, and pregnant women. 15 The USPSTF guidelines offer an important recognition of PrEP's efficacy and value as a prevention tool. Further, the USPSTF suggested health professionals should be mindful of the health disparities disproportionally affecting racial/ethnic minorities. From 2014 to 2017, PrEP awareness among MSM in 20 urban communities increased from 60% to 90%, and PrEP usage increased from 6% to 35%. PrEP usage increased in almost all demographic subgroups, but still remained lower among Black and Hispanic MSM. 16 Studies have shown that White persons meeting criteria for PrEP use were up to 6 times more likely to be prescribed PrEP than Black persons with similar criteria for use. 17 The reason for such discrepancies is multifactorial and not surprising given the historical disenfranchisement of persons of color in the United States. Stigma related to HIV and PrEP affects many persons, but the discrepancies in PrEP usage suggest that there may be different types of stigma and/or exploration of how stigma affects some individuals more substantially than other persons based on race, ethnicity, and cultural perspectives. In regards to PrEP use and promotion of PrEP use among health professionals, there remain knowledge gaps that pose significant deterrents and place a substantial burden on patients wanting to request treatment. 13 These knowledge gaps in conjunction with structural and social determinants of health may help to shed light on the reason for disparities in PrEP usage among MSM and other marginalized patient communities. --- Sources of PrEP Stigma For nearly two decades following the onset of HIV/AIDS gay men and injectable drug users were seen as the "face" of the epidemic with initial reports referencing the mysterious condition as Gay-Related Immune Deficiency under headlines such as, "New Homosexual Disorder Worries Health Officials." 18 Homophobia, which was not only considered socially acceptable but codified into discriminatory laws coupled with sensational reporting in the media reinforced "gay plague" stereotypes that linked the disease with a moral failing of persons infected. Rather than being treated with compassion and care, many persons living with HIV instead were recipients of victimblaming as though their infection was due in part to a failure to protect themselves and therefore somehow deserving of their declining state of health. Health professionals were also culpable as the moral authority of medicine has often contributed to health disparities experienced by patients and communities. In the early days of the epidemic, patients were not guaranteed access to health professionals due to fear of the disease and potential exposure. 4 Such attitudes among health professionals are exceptionally rare today, but the lack of health professionals trained in how to prescribe PrEP is still a barrier yet to be resolved. Limited knowledge about PrEP use and/or willingness to engage with patients about their sexual health practices and needs rather than fear are the challenges that must be addressed. What is not clear is whether these barriers with health professionals are based on stigma or some other health systems deterrent? Well-intentioned physician advocates for PrEP use may also be sources of harm due to HIV stigma. In a narrative reflecting on his personal experiences as a gay man and medical student in training, Samuel Dubin wrote how he felt fortunate to have a physician who prescribed PrEP but felt "slut-shamed and stigmatized by seeing PrEP as the only option for me because of the intolerable risk of my acquiring HIV." 19 Focusing on prevention-based approaches is important and necessary to promoting the well-being of patients, but there is a distinction between partnering with a patient to maximize their health promotion and disease prevention options and relegating the patient's autonomous identity to the sum of their health behaviors. Internalized shame and negative associations with sexual health behaviors among MSM may impede open dialogue with health professionals. 20 Stigma and shame are closely related to one another as they both seek to socially disqualify individuals and communities from acceptance and equality. Stigma is a social opportunistic disease that attaches to many illnesses and increases morbidity and mortality. 21 This limits the opportunity for engagement on the use of PrEP and other HIV prevention approaches. It may also skew individuals' perceptions of their eligibility for PrEP despite meeting criteria for those persons most at risk of exposure. 22 Patients initiating conversations with their health provider about PrEP use may be uncomfortable in explaining their interest if there is concern of moral judgment regarding risk compensation or nonadherence to condom usage. Patient interest in utilizing PrEP may be driven in a desire to decrease anxiety and concern about HIV transmission, as well as increasing their sexual pleasure and sensation. These joint motives may be received quite differently by health professionals and social peers. Expression of the latter being difficult for patients given attitudes towards such preferences have historically been framed as reckless, dangerous, and socially irresponsible. 23 Added to these concerns is the reality that among health professionals there remain stigmatized views of sexual and gender minorities that have nothing to do with HIV and/or use of PrEP, but are deeply ingrained biases. A 2014 Kaiser Family Foundation study revealed that 30% of gay men and 45% of bisexual men reported not feeling comfortable discussing sexual behaviors with health professionals. 24 These findings on interpersonal and communication barriers for MSM point to the --- DovePress difficulty with initiating a conversation with a health professional on the use of PrEP as the same study found that more than half of gay and bisexual men reported that a doctor had never recommended HIV screening. 23,25 Mitigating Stigma as Barrier to PrEP Stigma surrounding the use and promotion of PrEP creates a barrier to treatment and PrEP maintenance. Addressing externalized and internalized stigma regarding PrEP use will require patients and health professionals to reframe our perspective and conversations on its use as an HIV prevention tool. Clinical recommendations can be perceived as moral judgments when inquiring about a patient's sexual behaviors and preferences. Physician assessment of a patient's eligibility for PrEP use, concerns for adherence, and the likelihood of engaging in risk compensation may be influenced by social characteristics such as race contributing to health inequities. 26 Medical records of Veterans Health Administration patients revealed attitudinal barriers in addition to provider knowledge gaps regarding PrEP usage in a study by Skolnick et al. 13 A qualitative analysis of documented conversations with patients showed providers' preferences to focus on behavioral risk reduction strategies. Rather than prescribing, providers discouraged PrEP and would instead suggest limiting or even ceasing sexual activity. Other examples of stigmatizing attitudes included requiring patients to obtain prescriptions from their sexual partner's physician or submission of documentation verifying a partner's serostatus as a pre-requisite to receiving a prescription for PrEP, despite the VHA and CDC not requiring such steps. In a study by Quinn et al, cisgender Black MSM between the ages of 16 and 25 were asked to participate in focus groups to discuss experiences with racism and homonegativity and how this impacted PrEP uptake among participants. A key finding of the study was the shared experience among participants of feeling stereotyped by physicians. Disclosures about sexual health behaviors or sexual orientation were immediately met with concern for risk of HIV and other sexually transmitted infections. Some participants shared experiences of "being mistreated, talked down to, or feeling like they were perceived as 'nasty' by White physicians." 27,28 Such examples of interpersonal and communication microaggressions can often be attributed to implicit biases, but sometimes reflect explicit biases too. Regardless of the intention of the health professional, the ability to have meaningful discussions about PrEP becomes problematic. Discussing sexual health practices and recommendations for PrEP based on preventive health goals cannot be done without recognizing the historical impact of stigma in all its various forms and how it continues to affect personal behaviors and interactions with the healthcare system. Cultural and socioeconomic differences among patients within and across marginalized communities can also affect health outcomes and engagement with healthcare services. 29 In order to create community-based disease prevention strategies, health professionals and public health agencies must be sure to incorporate the perspectives and voice of the community members so that specific needs are prioritized and stigmatized views are counteracted in messaging and outreach efforts. HIV prevention campaigns typically target those persons or communities considered to be at greatest risk of exposure. This may include the use of advertisements within print media or television programming with overlapping consumer demographics, but has increasingly shifted to the use of social media platforms. While this can be an effective outreach tool in reaching a younger patient population it fails to help address larger community misconceptions. Social marketing to promote PrEP use and benefits intended solely for persons considered to be at risk may increase awareness within those specific subpopulations, but will have limited impact on increasing understanding and support among the larger racial/ethnic communities from which those most vulnerable have intersectional roots. Programs that normalize HIV prevention and treatment among at-risk and HIV-positive and their familial and social communities may further decrease HIV and/or PrEP stigma. 14 Among the concerns about PrEP stigma is its impact on historically underserved and marginalized communities, who represent some of the subpopulations at greatest risk of exposure and therefore have the greatest need for access and use of PrEP. Latino MSM represents 23% of new HIV infections in the US and the CDC reports that the rate of incidence has increased by 14% or more in recent years. 11 Black MSM represents 44% of new infections. Yet, they represent only 12% and 10%, respectively, of PrEP users. 21 The CDC's National HIV Behavioral Surveillance Study Group collected survey data from 20 urban areas in the United States in 2014 and then again in 2017. One of the goals was to examine the awareness and usage of PrEP. Although PrEP use among MSM increased by 500% there are differences in uptake for Hispanic and Black MSM and therefore prevention efforts must be tailored to address such disparities. 16 In response to these trends, the CDC has launched the Targeted Highly-Effective Interventions to Reverse the HIV Epidemic 30 program to support state and local health department demonstration projects developing community collaboratives that provide comprehensive HIV prevention and care services for Black and Hispanic MSM. Lessons learned from these community-based endeavors can better inform and guide the refinement of health guidelines and strategies that will have optimal outcomes among persons who benefit greatly from PrEP usage. --- Why Focus on Stigma Stigma was one of the main drivers of the HIV crisis in the 1980s and 1990s and obstacles to enacting effective health promotion and disease prevention. Addressing stigma in all its forms is fundamental to delivering high-quality health care As we witnessed over the course of the HIV/ AIDS epidemic, stigma affected more than just those persons living with the disease. The ripple effect throughout marginalized communities decimated biological families and chosen families. The moral distress experienced by health professionals and community health advocates as they witnessed countless deaths is a painful reminder as to how ignorance, fear, mistrust, and despair were among the most fatal opportunistic infections. --- Disclosure The author reports no conflicts of interest in this work.

Pre-exposure prophylaxis or PrEP is a Food and Drug Administration approved human immunodeficiency virus (HIV) prevention tool that reduces the risk of infection by greater than 90%. While it does not provide protection against other sexually transmittable infections and blood-borne illnesses such as hepatitis C, syphilis, chlamydia, and gonorrhea, it is a highly effective in reducing the risk of transmission of HIV among men who have sex with men. Despite the success of PrEP, there remain barriers to PrEP uptake rooted in stigmatized perspectives shared by health professionals, patients, and community members. The insidious impact of stigma associated with HIV/AIDS has permeated throughout the LGBTQ+ community, healthcare system, society in general and to this day, continues to exacerbate structural and social determinants of health disparities amongst sexual and gender minorities. While the initial resistance to PrEP has abated over time, stigmatized perspectives regarding PrEP continue to impede those at greatest risk from benefiting from effective preventive care.

Introduction From the origins of mankind, many practices, tools, equipment, technologies and even abstract ideas which have shaped the socio-economic, political, and psychological composition of the globe and its inhabitant have emanated from inventions and innovations. These inventions and innovations have been materialized in some factors including the quest to proffer solutions to complex human problems, the desire to understand the nature, origin, composition, and working of objects or behaviours of both abstract and concrete phenomena, and the need to adapt to global socio-environmental changes . Central to these inventions and innovations is knowledge which is gained from constant foresight, learning and research, and which further translates to creative discoveries. For instance, in the natural sciences parlance, the discovery of Nitrous Oxide as anaesthesia in the 1840s by Joseph Priestley and Humphry Davy was necessitated after a series of experimental research on the same gas which was only used as a preserving agent in the 1700s . In the technological milieu, the Industrial Revolution which first began in Britain in the 18th Century was induced by the transmission of tacit knowledge during apprenticeship in areas like Britain itself, Ireland, and Germany . In the academic/scholarship parlance, the evolution of disciplines like Public Administration in the late 19th Century was impelled by the development of ideas which differentiated administration from mainstream politics and the spread of such ideas by classical scholars like Wilson Woodrow, Frank Goodnow, L. D. White, and W.F. Willoughby, etc. . In the sociocultural ambience, the unified and mono-cultural state of Eastern Asian countries like Japan and South Korea which is a major contributing factor to the political stability of these countries since WW II ) is a factor of the cultural education which was instituted through the Juche policy by great leader Kim IL Sung in the mid-20th Century . From the foregoing, the importance of knowledge and its proliferation is all-encompassing. However, though there are many forms through which such knowledge could be transferred, it is observed that the most organized platform for such knowledge transfer is through formalized institutions such as schools, as a result of an increase in population and socio-economic complexities of human activities. While these formal institutions at inception were seen as objects of status by the elites in the society, the gap was bridged in the 20th Century as the rise in the demand for skilled and learned human capital accommodated the different strata of the society. Consequently, the state of formal education and the rate at which it expands qualitatively amongst citizens of a country now determines the literacy levels of such country, and which is also seen as a major driver of development and civilization in the world. Hence, the higher and qualitative the educational literacy levels, the higher the chances of orchestrating and sustaining development in respective countries of the world . Nigeria has remained a developing country since her return to democracy in 1999 owing to low human development capacity which is determined by factors including literacy rate , Per Capita Income , and Life Expectancy . Specifically, the levels of educational literacy in this regard are not just the number of educated persons, but also the average number of completed years of education of a country's population aged 25 years and older, excluding years spent repeating individual grades. Hence, this which is measured using the Mean Years of Schooling takes into account those with no level of education, those with incomplete and completed lower and higher secondary education, as well as those with incomplete and completed tertiary education. Thus, the MYS offers insight into both the rate and spread of educational literacy amongst the population about other factors that may affect the completion of education in specific countries. This to a large extent further indicates the quality of education in these countries . In Nigeria, available statistics show that MYS is at 7.2 which is lower than the global minimum average of 8.7, thus, indicating a low level of formal education in the country. In respect of the quality of education and best educational systems in the world, Nigeria ranked 74th in 2022 out of 140 countries, indicating deteriorating levels of labour competencies as well as quality described in terms of digital literacy, interpersonal skills, and the capacity to think critically and creatively . This is however different from the situation in China which was predominantly educationally backward before the onset of the 21st Century. Aside from having literacy spread percentage of over 99.83% as of 2021 , the country recorded an average MYS of 10.75 in 2020, indicating a 2.5 increase from its MYS of 8.25 in 2010. China's steady improvement in the educational sector is attributed to factors including investment in the development of professional technical human resources, the training of highly skilled talents, education infrastructural development through funding, and the direct supply of trained talents to enterprises . Additionally, the country ranked 22nd in the list of countries with the best educational systems in the world, a rise from its rank of 24th in 2020 . Essentially, Nigeria has remained backwards in the fight against educational illiteracy owing to challenges like inadequate funding, poor educational infrastructure, poor supervision, unqualified educational administrators, and inadequate professional teachers, overcrowding in classrooms, etc. . For instance, in the aspect of infrastructure, the Senate President of the National Association of University Students , Comrade Ibrahim Lawal lamented in 2021 that both students in some of the Federal and State-Owned institutions fight to secure venues for lectures and examinations because of acute shortage of lecture halls, while other institutions were faced with a paucity of equipment for practical experiments, and lecturers also struggling for office spaces . Having exposed the discrepancies in the state, quality and levels of educational literacy in both countries, the study is undertaken to compare the approaches to tackling the problem of educational illiteracy according to three variables including structure of administration, budgetary allocation, and pattern/content of learning. This comparison aims to explore probable recommendations for the dwindling state of education in Nigeria and other educationally backward developing countries. --- Conceptual Framework Concept of Educational Illiteracy For the holistic comprehension of educational illiteracy, it is necessary to first explore the meaning of illiteracy. The definition of illiteracy is multifaceted and specific to diverse organizations, cultures, contexts, and even professions. Though it is generally viewed as the inability to read and write , more recent and specific definitions have emerged over the years. Consequently, to effectively understand what illiteracy is, it is important to define literacy as a concept. According to UNESCO , literacy involves a continuum of learning in enabling individuals to achieve their goals, develop their knowledge and potential, and participate fully in their community and wider society. This definition points out that literacy entails the use of theoretical and practical forms of learning for personal and public cum holistic development. According to Pierre et al , literacy which is also referred to as cognitive skills is defined as the ability to understand complex ideas, to adapt effectively to the environment, to learn from experience, to engage in various forms of reasoning, and to overcome obstacles by taking thought. Here, emphasis is still placed on both theoretical and practical learning, as well as their use in observation, adaptation and reasoning, all engulfed in the major aim of achieving results. From both definitions above, one can understand and define holistically the concept of illiteracy. It is veritable to aver that the holistic definition of illiteracy must highlight the absence of theoretical and practical learning while also enhancing the inability to apply learning results in real-life situations. Hence, Encyclopaedia of Oxford University Press defines illiteracy as not just the inability to read or write, but the perceived state of being uneducated or insufficiently uneducated. For the institution, an illiterate cannot use reading, writing, and calculation skills for his/her development or society's development. Similarly, and more acceptably, the term is defined by UNESCO as a state where a person cannot, with understanding, read and write a simple statement on his everyday life; and a functionally illiterate is a person who cannot engage in all those activities in which literacy is required for effective functioning of his group and community and also for enabling him to continue to use reading, writing and calculation for his own and the community development. The definitions above emphasize on learning and the application of learning for personal and public development. Educational illiteracy, therefore, is the inability of individuals to read, write, spell, and use mathematical, scientific, and technical skills to interpret, react, articulate and express their thoughts in specific situations. Hence, educational illiteracy does not only concern itself with the educational institutions where these skills are learnt, but also the quality of the content being learnt, the deliverers of the content, and the environment through which these contents are delivered to develop and shape individuals' mind and knowledge as they face societal challenges . --- Theoretical Framework: The Structural-Functional Theory The structural-functional theory which is also an approach to the study of Comparative Public Administration originated from the works of classical Sociologists and Anthropologists like Herbert Spencer , Emile Durkheim , and Alfred Radcliffe-Brown in the late 19th and early 20th centuries, respectively. The theory was however modified by American Sociologists Talcott Parsons and Robert Merton between 1954and 1975. As a sociological theory at inception, structural-functionalism originally expressed that society is likened to the human body with various organs which work together for the functioning of the whole organism. Accordingly, these organs according to Spencer are the social institutions, patterns of beliefs, behaviours and cultures meant to meet the diverse social, economic, and political needs of the society. In essence, the sociological aspect of the theory explains that the social structure consists of institutions, norms, and values which play specific roles through the patternization of behaviours to ensure the stability and functioning of the society. To this effect, any dysfunction on the part of any of these social structures might result in the collapse of the society . Gabriel Almond ) and Bingham Powell particularized and transitioned the theory to the political enclaves . According to them, the political system consists of both structures and functions, which makes it easier to understand the political system and ensure its workability and stability. For them structure in a political system does not only refer to the traditional institutions like the executive, legislature, and the judiciary, but also the observable activities and behaviours which characterize the political system, and which also consist of regularities and patterns. Hence, structures in a political system include the characteristics of the political economy, the socio-political culture of citizens, and the administrative patterns of institutions of governance and agencies of the government . Accordingly, functions are actions that are performed by specific structures to maintain the stability of the system and to keep it relevant. In fact, according to Ezeanya , functions are the consequences of actions or behaviours by members or constituents of each structure. These functions are however those within the jurisdiction of the structure involved and which might be performed by different structures based on the stage of political evolution. Therefore, the theory emphasizes the specialization of functions by particular structures to avoid instability and overlap of functions in the political system. The functions required to ensure the stability of the political system are divided into two; the input functions and the output functions. The input functions also referred to as the non-governmental functions or political functions are those involving the exchanges in the political context. They are usually performed by Non-Governmental Organizations and socio-political groups such as pressure groups, interest groups, political parties, etc. These functions include political socialization/recruitment, interest articulation, interest aggregation, and political communication. The output functions are those involving the direct exchanges between the government and the people. They include rule-making, rule application, and rule adjudication. Summarily, Almond and Powell conclude that though these output functions are not uniformly performed by similar structures in all political systems, the understanding of the structures and functions in a political system would provide details on how stability in such a system can be maintained. This understanding would also help in identifying and comparing the type of structures and their commensurate roles in different countries and cultures of the world . The theory is relevant to the study because it establishes an inextricable relationship between the stability of a political system to the activities of the structures of such system, which further determines the state of affairs of the system in aspects like agriculture, education, health, etc. In extension, it is the role of the political structures such as the executive, legislature, judiciary, and bureaucracy to maintain stability through their roles in policy making, policy application, and policy adjudication. Hence, ensuring widespread educational literacy is the function of both state and federal governments. This means that it is pertinent that educational illiteracy being a challenge in the country is salvaged through specific approaches of the executive, legislature, and judiciary, as well as other educational agencies established to oversee activities in the sector. For instance, the legislature can make educational reforms through the Acts of Parliament and also request the increment of budgetary allocations to the educational sector through its financial oversight functions. Likewise, the institutions of the executive like the diverse supervisory boards of education in the country can supervise the activities of educational institutions to ascertain the quality of teaching and infrastructures of educational institutions and make reports to the Ministry of Education where necessary. --- Overview of Educational Illiteracy between Nigeria and China As earlier emphasized, illiteracy is a major factor hindering development in several climes of the world. Investment in education is a component for increasing Human Capital Development , thus, a country which invests properly and judiciously in education is likely to build strong, efficient and innovative manpower, which is an indelible prerequisite for development. The analysis of the literacy levels in both Nigeria and China is determined by two factors which are also used in measuring literacy rates in all countries of the world. They are the data of the Literacy Index as provided by international organizations like the UN through UNESCO and the Mean Years of Schooling . According to UNESCO, the Literacy Rate of each country is determined by the number of persons aged 15 and above who can read and write, and understand simple sentences and statements about themselves or other persons. Accordingly, the MYS is determined using data from several persons with no level of education, those with incomplete and completed lower and higher secondary education, as well as those with incomplete and completed tertiary education. In this fashion, the MYS takes into consideration, the rate and spread of educational literacy amongst the population as well as the factors that may affect the completion of education in specific countries. Hence, it measures both number of years and the quality of education in respective countries. It is usually calculated every 5 years. The lower the MYS, the lower the spread and quality of education in a country and vice versa. The global average is 8.7 years with the highest being 15 years. An MYS below 8.7 years is therefore below average and literacy as well as the quality of education considered low in such a country . Statistically, the literacy rates of both countries are at extreme ends. China recorded a massive literacy rate of 99.83% of its 1.4 billion population in 2021 indicating a 4.71% increase from 95.12% in 2010. Nigeria recorded only 77.62% of its 213.4 million population in the same year, though this indicated a 13.9% growth between 2010 and 2021 . In terms of the MYS, China recorded a 2.5 increase from 8.25 in 2010 to 10.75 in 2020, while Nigeria recorded a 1.1 increase from 6.8 in 2010 to 7.9 in 2020 . It is thus shown that while China's educational literacy level topples Nigeria's in both literacy rates and MYS, Nigeria has recorded an appreciable improvement in literacy rate since 2010 but with lackadaisical growth in the MYS which takes into consideration both rates and quality. As expected, the reasons for the poor level of literacy in the latter are numerous. According to Birabil & Ogeh , the backwardness of Nigeria in literacy development is contingent on factors like value placed on paper qualification, poor planning, poor funding, poor educational infrastructures, inadequate classrooms and teaching aids, paucity of quality teachers, polluted learning environment, and academic fraud. However, a comparison of the approaches adopted practically by these countries to curb illiteracy shall be made and conclusions drawn. --- Tripartite Comparison of Approaches to Educational Illiteracy Reduction in Nigeria and China. 1. Comparison According to Structure and Pattern of Educational Institution Administration. All-inclusive vs. Specific Approach The Nigerian system of administering tertiary education is quite holistic. Tertiary institutions are being established with no specific interest in a particular field, area or profession, rather a general academic institution is created which absorbs all available fields, areas and professions. This approach to university education makes learning seemingly intractable, skill acquisition inappropriate, and specialization incongruous especially for a developing country. The negative effect of this is overcrowding and understaffing which is one of the problems of tertiary education in Nigeria. For example, the University of Nigeria, Nsukka , has 15 Faculties and 106 academic departments, with over 35,000 students lectured by a meagre 1,700 academic staff. Statistically, this represents only 16 lecturers per department with multiple courses being handled by some of the lecturers. The University also offers 82 undergraduate programs and 211 post-graduate programmes . This holistic form of educational administration without specifications results in rather poor academic terrain, few teachers/lecturers handling many students, poor institutional care, admissions without due process and screening, and the production of unqualified graduates . However, this is not the case in the Chinese tertiary educational administration. Each ministry is allocated a university which concentrates on the ministry's area of specialization. For instance, Hohai University is directly administered by the Ministry of Water Conservancy, while Dalian Maritime University is directly administered by the Ministry of Communications. The General Aviation Industry Corporation of China and China Ship-Building Corporation also manage a few universities. Of the 350 universities in China, some like the Hohai University, Nanjing, Tsinghua University, Beijing, University of Science & Technology, Hefei, etc., are well recognized globally for academic excellence. For instance, the Tsinghua University of Beijing which is observed to be vaster in natural, biological, environmental sciences and engineering is the best in Asia and also ranked 23rd in the global ranking of best universities in the world . These universities deal directly or are well recognized in specific areas of specialization, hence enhancing academic propriety and ease in learning and skills adaptation. --- Comparison According to Budgetary Allocations to Education The high educational literacy rate is a function of massive investment in education. This investment is evident in the budgetary allocations and other financial provisions to the educational sector and institutions in a particular country. From the Nigerian space, illiteracy is exacerbated by the meagre allocation to the educational sector. The educational sector is deprived of funding in the yearly budget, despite its large GDP. For example, in the tertiary education parlance, Nigeria has over 170 universities and 120 Polytechnics and Colleges of Education inclusive; yet, the allocation to the educational sector remains insignificant , especially for a country with low indices of literacy. In the 2021/2022 budget, educational sector expenditure decreased, despite the impacts of the pandemic on the sector. Total allocation to the education ministry was just 5.8% of the total budget. This was less than the last 5-year average of 7%, indicating a step backwards in educational development in Nigeria, which further depletes the quality of education in the country. The table below shows the educational spending in Nigeria as a percentage of budgetary allocation from 2015-2023. 1 shows that education expenditure started declining from 2016 to 2020. However, gradual increases began in 2021 and hit a massive 3.41% increase in 2023 from a 0.01% increase in 2021. However, these allocations have not met the UNESCO recommendation of 15% to 20% of the respective country's public expenditure . This is adjudged as one of the reasons for the country recorded over 20 million out-of-school children and series of unresolved strike actions by educational unions such as the Academic Staff Union of Universities , the Senior Staff Association of Nigerian Universities , Non-Academic Staff Union of Educational and Associated Institutions , Academic Staff Union of Polytechnics , National Association of Academic Technologists , Colleges of Education Academic Staff Union , etc. These strike actions which were more pervasive in 2022 has overtime in truncated academic progression in tertiary institutions, thereby, depleting the quality of Nigeria's educational environment The Nigerian case as stated above is a spatial deviation from the situation in China. Since 1983, the Chinese government has attached great importance to the development of the educational sector with the percentage of education spending to the total budget not below 9%. Besides structural reforms, public spending on education increased considerably . The table below shows the educational spending in China as a percentage of budgetary allocation from 2015-2023. From Table 2, it is observed that the allocation to education in China's annual budget has not fallen short of 10% since 2015. Although the percentages from 2015-2021 did not satisfy the UNESCO recommendation, increases to 15% and 17% in 2022 and 2023 respectively indicate massive improvements, especially in the face of global digital and technological advancement. A bulk of the country's investment in education is directed towards digital and technical literacy through diverse policies like the Education Informatization 2.0 Action Plan, and which has seen the country emerge as one of the leading countries in digitalization and manufacturing . --- Comparison According to Pattern and Content of Learning Traditional/Theoretical Approach vs. Digital/Technical/Practical Approach The existential focus on theoretical as well as traditional approaches and methods of learning has deepened both the spread and quality of educational literacy in Nigeria. It is indeed undeniably conspicuous that the world is tilting towards digitalization and technicalization in the era of the Fourth Industrial Revolution which in the educational terrain is characterized by the utilization of new technologies like Artificial Intelligence , cloud computing, robotics, 3D printing, etc., to enhance the speed and quality at which new information and knowledge are generated and transferred . Hence, countries of the world have instituted diverse policies, plans and strategies to adapt to the changing system. However, the majority of educational institutions in Nigeria from the primary to the tertiary level have not adapted to the 21st Century's use of ICT in transferring knowledge . The deficiency in digital illiteracy is prevalent among both teachers and students. For instance, in 2019, only 11 federal universities could provide at least 2,000 computers for students ' and lecturers' use . The absence of a practical policy framework for the use of digital technologies in the nation's educational system is perhaps one of the reasons for more than 50% of the country's over 200 million population lacking digital skills. A similar case is pervasive in the aspect of technical education. For instance, Lawal noted that technical educational institutions in Nigeria are faced with challenges including dilapidated infrastructure, obsolete equipment, and unqualified teachers and instructors to meet up to the current industry skills, generally owing to poor policy framework, sectorial corruption, inadequate funding, and mismanagement of funds. Stemming from these issues, it is observed that normal workshop practice which forms over 60% of the standard of technical college curriculum as set by the National Board for Technical Education is fast deteriorating. In fact, between 2007 and 2017, over 45% of Nigeria's professionals including technical educators had migrated to other countries for better learning and working environment, owing to the government's inability to invest in practical digitalization and technicalization of education . China's approach to illiteracy reduction in this regard is more pro-digital and pro-technical. This is achievable through the institution of policies and strategies to ensure computer and technical skill acquisition by students and the general public. Though investment in digital and technical education had become vigorous in the late 20th Century, recent policies have proven to be more effective in the national coverage of f digital and technical literacy in the country. For instance, in 2021, the country launched the "Action Plan for Enhancing Digital Literacy and Skills of the People" which was also included in the country's 14th Five-Year Plan. Digital education has become a national strategy in China . Previously in 2018, the Ministry of Education in the country had issued the Education Informatization 2.0 Action Plan which proposed to take education informatization as a key driver of education reforms. The Plan aimed to provide teaching applications to all school-age students, and construction of a digital campus in all schools, ranging from primary to tertiary schools . This has resulted in numerous strides in digital education in China as schools have produced more than 300,000 high-quality digital teaching resources including micro-lessons, teaching plans, courseware and classroom records, and the coverage of teaching material resources reaching over 60% in cities like Yancheng . Learning clouds have also been created to tackle the disparities in rural areas. Owing to these developments, China has within 20 years risen to the ranks of countries with massive digital strength, resilience and responsiveness as measured by the Digital Skills Gap Index . Specifically, the country currently ranks 18th of 134 countries with a score of 6.7 while Nigeria ranks 103rd with a score of 3.6% . Particularly, the investment in the digitization of education has had positive reverberating socio-economic impacts on the citizens and the country at large. For instance, there is an increase in the number of women involved in the digital economy with over 23 million female online shop owners, with 3.92 million from rural areas. This has to an extent bridged the income gap between both genders in the country. Improvements in digital literacy have also lessened the urban-rural gap in the country. Furthermore, 8 of the world's largest digital education companies are based in China with the digital education market grossing over $44.56 billion in revenue in 2019, indicating a massive rise from $10.03 billion in 2012 . Likewise, China has improved massively in the area of technical education. The country has the world's largest vocational and technical education system, featuring over 11,300 schools which enrol over 30 million students and 10 million graduates per year . Particularly, it should be noted that technical education in China does not only focus on expansion in size, but quality improvement. One of the measures adopted to strengthen vocational and technical education is the promulgation and implementation of the Vocational Education Law which is aimed to establish close connections and cooperation with enterprises in respective areas to cultivate talents for certain vocations in respective enterprises . In addition to this, students can participate in internships from matching enterprises during their studies and have the opportunities to become full-time workers at their chosen enterprises after obtaining their degrees, hence, mitigating the problem of youth unemployment. Furthermore, China has formed over 1,500 education groups and alliances covering more than 45,000 member units. These have included enterprises, schools, industries and research institutions to conduct studies. Also, domestic vocational institutions have established over 24,000 internship bases with enterprises which are supported by incentives like land, loans, and tax reduction . The country has also practically reformed the Scientific and Technological Progress Law of the People's Republic of China which aims to improve the measures to guarantee efficiency and effectiveness in science and technology together with a high level of self-reliance. This is why China is ranked 1st in the rankings of countries according to the manufacturing industry. The country makes over 28% of the total global manufacturing output which adds over $4 trillion to the world economy annually . --- Conclusion The level and quality of education is one of the factors of development in the world which is necessary because inventions and innovations are products of knowledge about how it is generated and transferred. Hence, countries of the world have over time developed approaches to ensure the spread and quality of education to tackle its adverse, illiteracy. In the study, it is discovered that there is a huge disparity in the approaches to illiteracy reduction between Nigeria and China in terms of the structure and patterns of educational institution administration, budgetary allocations to the education sector, and patterns and contents of learning. Specifically, it is discovered that Nigeria adopts the all-inclusive and holistic pattern of administering educational institutions, especially at the tertiary level, while China's pattern of administration about tertiary education is more specialized with each institution having a specific area of specialization and ministerial administration. This ensures easy accountability and administration of each institution, as well as vast specialization, in specific areas. Additionally, Nigeria's budgetary allocations to education since 2015 have fallen short of the UNESCO recommendation of 15%-20% of the total budget expenditure. This has resulted in other problems including inadequate learning facilities, poor infrastructure, poor lecturers' welfare, and intermittent industrial strikes in the educational sector. In China, though the allocations between 2015 and 2021 had not met the UNESCO standard, it is appreciable that such was not less than 9% of the total budget. The UNESCO standard however was met in 2022 and 2023 respectively, and has contributed to the increase in the spread and quality of education in the country. In terms of patterns and content of learning, the Nigerian system has not adapted to the 21st-century investment in and use of digital technologies in learning as the concentration is still heavy on the traditional and theoretical methods of learning. This is different in China which has evolved over the years through the promulgation and strict implementation of policies for digital and technical educational development. Hence, the country has risen in the ranks of digital and technical literacy quantitatively and qualitatively, which has boosted its manufacturing sector. In essence, it is identified that the bulk of Nigeria's challenges towards illiteracy reduction generally lies in the poor policy formulation and implementation framework, coupled with others like inadequate funding, mismanagement of funds, public sector corruption, poor supervision of educational institutions, and unqualified or inadequate staff in educational institutions. --- Recommendations From the analysis above, the following recommendations are made; 1. Tertiary institutions should be established to focus on specific areas to ensure specialization, and expertise, amongst both instructors and students and also to reduce over-crowding in Nigerian tertiary institutions. For instance, tertiary institutions could be established in such manner as the University of Social Sciences; University of Natural and Biological Sciences; University of Engineering, etc. 2. Allocations from the national budget to the educational sector should be increased to at least 15% as recommended by UNESCO. 3. Provisions should be made for digital infrastructures like digital libraries, computer systems, projectors, and other gadgets in public primary, secondary and tertiary institutions. A board should also be established to supervise the usage, maintenance and management of this equipment. 4. Public technical and vocational colleges and institutions should be provided with the necessary infrastructure and tools for practical experimentation. Both federal and state governments should establish technical education boards to examine and supervise the usage, maintenance and management of this equipment. 5. Vocational education should be strengthened through the establishment of vocational institutions which should also issue certificates/degrees equivalent to conventional university degrees. 6. Government should ensure partnership between enterprises and vocational/technical institutions through a policy framework for the cultivation of talents and employment of qualified hands-on completion of studies. 7. The independence of anti-graft agencies should be strengthened to enable them to tackle corruption and mismanagement among public officeholders.

The study focused on approaches to educational illiteracy in Nigeria in China: A Tripartite Comparative Analysis. Holistic development in some countries of the world has been frustrated by the absence of digital awareness and technological creativity which are achieved through the spread of qualitative knowledge in a uniform manner. The poor spread of such knowledge, which has since been embodied into formal education in educational institutions results in educational illiteracy. Hence, countries of the world have instituted different strategies to curb the issue which in some climes like Nigeria and other developing countries of the world has been existential. The aim of the study is to compare the approaches to educational illiteracy between Nigeria and China through a tripartite model. The descriptive research design was adopted, while the sources of data were majorly secondary through the use of internet sources, textbooks, print publications, etc. The structuralfunctional theory was adopted for the study. It was discovered that there is a colossal disparity between Nigeria and China in terms of structure and pattern of educational institution(s) administration, budgetary allocations to the education sector, and pattern/content of learning, with China faring better than Nigeria. It was recommended that tertiary educational institutions should be established to focus on specific areas to ensure specialization, and expertise, amongst both instructors and students and also to reduce over-crowding in Nigerian tertiary institutions.

INTRODUCTION Hypertension is the main risk factor for cardiovascular disease , 1 the leading cause of death in Colombia accounting for 133 deaths per 100,000 population annually. 2,3 Evidence from high-income countries suggests that conditions in the social and physical environment influence population health and risk for CVD. [4][5][6] These environmental conditions are, in turn, influenced by the decisions of public officials, community leaders, and service providers. These social and political leaders play key roles in the social and economic patterning of health through the selection of policy alternatives, allocation of public resources, and other decisions that shape living conditions, including access to protective resources and exposure to risks. [7][8][9] Globally, there is increased attention to understanding the health implications of social and economic policies at municipal, state, and national levels. 10 However, in Colombia, as in other middle-and lowincome countries, there is a paucity of research on how local actors who shape these policies view the contribution of living conditions to health in their jurisdiction. Colombian municipalities have responsibilities for public and social services, building and maintaining infrastructure, and health care and public health services. 11 Decisions made by municipal leaders have implications for healthprotective resources, harmful exposures, and resident's vulnerability to conditions that undermine their health. 8,12,13 People experience many of the effects of policy choices and conditions that influence their health risks, as well as the opportunities and resources that protect them against adverse health outcomes, at the local or municipal level. 14 Examining the frames that local decision makers bring to explaining hypertension can help in understanding the decisions that they make about strategies for addressing this problem in their jurisdiction. Of particular interest in this study is the extent to which these local decision makers consider the health implications of policy decisions which are often considered outside the realm of Bhealth^, and which have increasingly been linked to health outcomes. 10 In the following pages, we examine key themes that were identified from interviews with decision makers focused primarily on links they observed between structural conditions in the city and hypertension risk, with particular attention to themes of displacement, limited economic opportunities, and municipal infrastructure. --- THEORETICAL FRAMEWORKS This study is informed by ecological models of public health and by framing theory. Ecological models of public health make explicit the role of multiple layers of contextual factors 15,16 in shaping population health. Moreover, ecological models as proposed Galea and colleagues 8 and Schulz and colleagues 14 suggest opportunities for local actors across multiple levels to improve health. However, little research has examined whether or how local actors in Colombia explicitly connect decisions over which they have influence to opportunities for addressing hypertension, and if so, in which arenas. Framing theorists suggest that the way a particular problem is framed will shape the options that are considered for action, the rationale for intervention, who is perceived to be responsible for addressing the problem, and ultimately, the course of action selected, as well as those that are discarded. 17,18 For instance, to the extent that hypertension is framed as a consequence of individual health-related behaviors, interventions may focus on lifestyle modification, with responsibility for behavior change substantially left to the individual. In contrast, framing hypertension as a consequence of living conditions opens a broader range of potential interventions including, for example, modifications of the social and physical contexts in which people live. The frameworks that local actors hold may inform their actions. Frame theory suggests that frames are socially constructed, both reflecting and shaping beliefs about potential courses of action and their implications. As a result, they may serve to increase, reduce, or legitimatize social disparities in health. 19 In sum, a frame approach allows for the exploration of the frameworks local actors currently bring to their understanding of hypertension and a deeper understanding of the potential to frame solutions across a full range of options suggested by an ecological model. The literature on the use of frame analysis to understand social movements 18 has proposed three core framing tasks. The diagnostic task refers to how the problem is defined. It attempts to identify who or what is to blame or who is responsible. This task is important because it focuses attention on one particular explanation and not others. The particular explanation offered will influence actors' perceptions of the potential solutions to that problem. 18 Identification of potential solutions is the focus of the prognostic task, as well as tactics or means for achieving them. 18 Finally, in public policy, the motivational framing task relates to the identification of those actors who should do something about the problem. 20 Each of these stages is informed by the previous stages, with the definition of the problem during the diagnostic task shaping the potential solutions that are identified as well as the understanding of who is responsible for enacting those solutions. Given the importance of the diagnostic task in shaping this process, it is our focus in this paper. Specifically, the analysis presented below addresses the question BIn what ways, if any, do local actors' frames regarding hypertension risk in their municipality include the role of living conditions?^The intent is to expand our understanding of the frameworks of local actors in Quibdó, a Colombian municipality characterized by a high poverty level and a growing multiethnic urban population, with a particular focus on whether and how those frames encompass associations between living conditions and hypertension risk. Our overarching goal is to inform policy interventions aimed at reducing hypertension risk in Quibdó and similar contexts in Colombia and Latin America. In particular, this study provided the foundation for a larger public health initiative that has emerged from the initial interviews; results of which are reported here. It has informed the establishment of a community-academic-governmental partnership that continues to work together to address social determinants of hypertension in Quibdó. The results presented below suggested that municipal determinants of hypertension in Quibdó involved forced displacement and its implications for social relationships in Quibdó; unplanned urban space and implications for the city's geographic and social structure; and unemployment and economic conditions in Quibdó. These results continue to guide the work of the ongoing community-academic-governmental partnership to address social determinants of cardiovascular health in the city. --- METHODS Using a critical case sampling strategy, 21 we selected Quibdó to conduct a qualitative single-case study 22 addressing the above question. This selection considered demographic, social, and epidemiological criteria. Quibdó is the capital and largest municipality of the department of Chocó, Colombia. In 2012, Quibdó's had 115,052 residents, 25 % of the population of Chocó. Ninety-two percent of Quibdó's population are blacks, and 89.5 % live below the poverty line, according to the Unmet Basic Needs Index. 23 Quibdó is the only capital city located in, and the third largest municipality of, the Colombian Pacific Coast, the sub-region with the highest prevalence of hypertension in the country . 24 In Quibdó, coronary heart disease and stroke, the two most important types of CVD, are the leading causes of mortality in the population aged 45 years and over . 23 Although the prevalence of hypertension in Quibdó is unknown, the high rate of mortality for CVD and high prevalence of hypertension in the sub-region where this municipality is located suggest that hypertension is an important contributor to morbidity, disability, and mortality. This study was reviewed by and granted exemption from IRB at the University of Michigan for Protection of Human Subjects in 2011. Participant Recruitment and Data Collection. We used a stratified purposive sampling design 21 to capture information on political, social and economic characteristics, resources, and potential interventions from a variety of perspectives in Quibdó. Participants were selected from four strata, each with different roles in the local policy-making process: community leaders , appointed municipal officials, elected policymakers , and health practitioners. The first author, a local health researcher, and a fieldwork assistant compiled a list of potential participants within each of these strata, including municipal officials and policymakers appointed or elected during the periods 2004-2007, 2008-2011, and 2012-2015; health practitioners; and community leaders. Participants were then selected based on the potential richness of information they might provide about the research question through semi-structured interviews. Among those who accepted the invitation to participate, interviews were scheduled based on their availability. The first author conducted three rounds of interviews in 2012, using a topic guide informed by the diagnostic task in framing theory . Interviews were conducted in Spanish, the native language of Quibdó residents as well as the first author. Interviews were completed with 13 of 18 of participants invited to participate based on the identification of people from each stratum. Five interviews were not conducted due to participants' time constraints or travel schedules. Interviews lasted 60-80 min and were audio taped, transcribed verbatim, and checked by the first author for errors. Examples of the types of topics included in the topic guide used for the interviews are shown in Table 1. Data Analysis. We analyzed the data following the five steps of the framework approach: familiarization; identification of a thematic framework; indexing; charting and mapping; and interpretation. 25 Familiarization involved immersion in the data by listening to tapes, reading field notes and transcripts of each interview, and listing key ideas and themes. In the second step, a thematic framework was developed based on the diagnostic framing tasks and the themes derived from the data. In the third step, indexing, the thematic framework was applied to all the data gathered. Charting, the fourth step, involved abstracting and synthesizing the data before rearranging them to create charts. The original text was referenced in order to trace the source. At this stage, the abstracted and synthesized themes were translated into English. Translation was not done earlier in the process to reduce potential distortions of meaning during the initial analytic stages. In the final analytic step, charts were used to map the nature of the phenomenon under study and identify relationships between themes to provide sufficient description and explanations for the findings. 25 Nvivo10 was used for managing the data including the indexing and charting. Findings were discussed with 5 out of the 13 individuals who participated in the interviews . Three participants excused themselves from attending the meeting due to travel or work constraints. The reasons why the remaining five participants did not attend the meeting were unknown. The meeting with participants included a summary of preliminary results and discussion on completeness, relevance, and clarity of the findings. Results from this process have been incorporated into the findings reported below. --- RESULTS The final sample included four community leaders, four appointed municipal officials, three elected policymakers, and two health practitioners. Six participants were female and seven were male. All but three participants had post-secondary education. Seven were between 40 and 50 years old, four in the 30-40 age category, with two of undetermined age. Based on analysis of the qualitative data, three interrelated municipal determinants were identified related to the diagnostic framing task. Specifically, participants described the following as associated with hypertension risk and disparities in Quibdó: forced displacement and its implications for social relationships in Quibdó, unplanned urban space and implications for the city's geographic and social structure, and unemployment and economic conditions in the city . The role of these three forces as described in participant interviews is presented in the following paragraphs, followed by connections participants made between these conditions and proximal risks factor for hypertension. Forced Displacement and Social Relationships: Living Together and Apart in a Place We Own with Others Participants described changes in the city's population-both size and composition-and the implications of these changes for the social composition of the city. Among the drivers of these changes was migration, primarily the forced displacement of indigenous and black communities from rural areas. Displaced people arrived in Quibdó after escaping the armed conflict in rural and isolated areas in Chocó. Many have experienced trauma due to the loss of loved ones, lands, social connections, and culture. After arriving, displaced people lived in poor conditions in crowded public spaces, relying on public assistance or donations to partially provide basic needs. In addition, they also often encountered rejection and stigmatization by residents of the city, as described by this policymaker: The traditional people of Quibdó, like Colombians in general, have considered that these persons [displaced population] are like a plague. Study participants often described the displaced population as unskilled, delinquents, or street sellers, and as people of rural customs who have difficulties adapting to life in the city. Some participants described members of the displaced population as perpetrators of violence, while others described them as victims. Some noted that displaced persons experienced violence both in the areas from which they had fled and in the city to which they had arrived. Displaced people were stigmatized. They were seen as thieves, so people began to reject them. Without understanding that the displaced were victims, they were victimized [themselves]. In addition to the immigration of forcibly displaced people described above, participants also described the impact of voluntary migration into Quibdó on contemporary social relationships within the city. Voluntary migrants included people moving from rural areas or smaller cities in Chocó, as well as people from neighboring departments , into the city. Of particular concern were mestizos who came to the city looking for economic opportunities through, for example, business ownership. These complex and shifting processes of in-and out-migration of various groups were linked to dynamic relationships between social segregation and stratification of racial/ethnic and socioeconomic groups in the city. Forcible displacement of persons from rural communities and their immigration into the city contributed to shifts in relationships between ethnic and racial groups. Dichotomized relationships between white and black racial groups, previously the norm, became an increasingly complex social and economic hierarchy that encompassed racial/ethnic background, place of origin, and possession of material resources. Participants noted that mestizos occupied the top of this pyramid, followed by long-term residents of the city , displaced blacks, and indigenous residents . In the following section, we examine the ways that the processes of displacement and immigration, and the social hierarchies created through these dynamic processes, are reflected in the city's geographic and social structure, and the meanings ascribed to those social and geographic locations. Unplanned Urban Space and Geographic and Social Structure: a Central and Peripheral City in the Periphery of a Centralized Country Interviewees described Quibdó as a city that is both geographically and politically isolated, historically receiving little attention from the national government. Many shared a perception that the policies of the national government, located in the Andean region of the country, were informed by, and more relevant for, that region than for the department of Choco, located on the Pacific coast. As one participant noted: …the [national] government's policies have always been for the Andean region… because those who work for the [national] government are from that region and know only the dynamics of that region of the country. The perception of being located on the periphery of political power was linked to a sense that the department's needs and interests were not always well served by national policies, including decisions related to the distribution of resources, services, and infrastructure necessary to support quality of life and economic development. This sense of marginalization was reflected as participants recounted histories of struggle that have characterized the city's efforts to gain greater recognition and access to resources distributed through the national government: …the little we have achieved here in Quibdó and in the [department of] Chocó has been through a people's struggle. We got electrical power because there was a strike, we got telephone services because there was a strike…we got a university because of a strike, this is all evidence that we have been in complete oblivion… Similarly, interviewees described processes of geographic and social centralization and marginalization as these were reproduced within the city itself. Historically, Quibdó grew from the Atrato River to the periphery. This geographic orientation has been, and continues to be, recognized as a marker of social stratification among city residents, with higher social status associated with locations near to the river. Migration dynamics, particularly forced displacement, and the absence of urban planning policies have reinforced processes of stratification and marginalization as the city continues to grow. In particular, displaced immigrants tend to locate in new neighborhoods with more limited infrastructure located on the periphery of Quibdó, while longer term residents, and new migrants with greater access to resources, may locate in neighborhoods with better infrastructure and access to municipal and economic resources. In the mental maps of the city described by interviewees, Quibdó has a center and a periphery. Clear social differences were ascribed to these geographic spaces, as in the following excerpt from an interview with a policy maker: Currently and for many years there have been new neighborhoods [in the periphery]. There are people living downtown who do not know those neighborhoods. And there are people that would not go there [into the new, peripheral neighborhoods] even if they were paid in Euros. While the geographic boundaries between areas are not clear-cut, and are complicated by geographic and socioeconomic similarities across Quibdó, the social differences ascribed to geographic locations throughout the city were clearly part of the mental maps held by participants in this study. These carried over to opportunities for employment and economic conditions across areas of the city, as described in the following section. --- Unemployment and Economic Conditions: Without Opportunities for Living with Dignity As participants in this study discussed relationships between living conditions and hypertension, they almost invariably noted the importance of unemployment and its associated insecurities. For example: …another thing that has exacerbated the problem of hypertension is that we are number one in Colombia in the problem of unemployment, also in violence and insecurity. At the time these interviews were conducted, unemployment in Quibdó stood at 17.7 %, substantially higher than the 10.4 % unemployment rate experienced by Colombia as a whole. 26 While participants noted the effects of unemployment across all social groups, they described the particularly profound effects experienced by those with low socioeconomic position and by displaced populations. Displaced populations, in particular, were described as having few skills that match the available labor market. Participants repeatedly described the economic circumstances and the dearth of employment offering living wages as crucial explanations for the poor quality of life and the daily economic struggles for many. Reflecting these desperate economic circumstances, one participant noted that BHere, we survive off miracles.Ĥ igh unemployment was also linked to the absence of public services. Participants noted that the lack of public services and the huge investment this infrastructure requires restricted the establishment of businesses and created challenges in attracting companies that might generate employment opportunities in the city. For example, the following quote from a participant describes the need for basic infrastructure currently unavailable to support economic development: Who is going to provide water to you here? [If you are an investor] you have to develop a big water supply plan in order to provide water to all people that visit a mall. In addition to the need for basic infrastructure for economic development, participants described the absence of national and local policies to promote job creation and employment opportunities. As a result, many jobs-particularly those available to members of the displaced populations-were informal or temporary positions . In contrast, formal jobs in the city were described as almost exclusively public sector positions. Such public sector jobs were often described as under the control of politicians and frequently offered in exchange for political support. In such a system, several participants described the pressures experienced by those holding public jobs to provide support to the politicians that helped them get jobs. They also commonly feared losing their jobs with shifts in political power. Within the context of high unemployment and a political patronage system for public jobs, interviewees described women's particular vulnerability to sexual exploitation in the labor market. One female respondent described her experience with sexual harassment in finding and keeping jobs: I came in to present my resume and the first thing they take note of is your physique from head to toe, and excuse me for my language, they take note of your tits and your ass to see if you are good. They call you to offer you an opportunity if things happen as they'd like, and not all women are going to accept those conditions. This is also part of what happens here in Chocó, as they say very rudely you give it [sex] up to me and I'll give you the job then, and it stays like that, if you gave it up already, you need to keep giving it to keep the job. Under such conditions, women might be forced to choose between employment and income, continued sexual harassment and exploitation, and unemployment. Thus the themes of forced displacement and migration combined with political, social, and geographic isolation were linked with the dynamics of poor employment prospects. Displaced persons forced into the city by violence and conflict, limited infrastructure to support economic development, and labor force discrimination were clearly linked to a context in which economic vulnerability, underemployment, and unemployment are pervasive, with those most marginalized along racial/ethnic, and gender dimensions, among the most vulnerable. Participants in this study explicitly linked these conditions to hypertension risk. These linkages are described in the following section and presented visually in Fig. 1. --- Living Conditions: Shaping Behaviors and Getting under the Skin Participants highlighted the role of living conditions in shaping dietary practices and physical activity, two behaviors for which there is good evidence of associations with hypertension. 27 Furthermore, they noted the role of the forces described above in shaping stress, another factor they associated with the risk of hypertension among residents of Quibdó. We discuss each of these themes below. Living Conditions and Dietary Practices. Participants mentioned that people in Quibdó eat Ba lot, bad, salty, and fatty.^They tied structural conditions in Quibdó to the availability, accessibility, and storage of staple foods. Historically, food in Quibdó has come from other regions of the country. Due to the poor road conditions and long distances from places where food products are harvested or produced, products, particularly fresh fruits and vegetables, are often of poor quality when they arrive, and prices are high and unaffordable for most local residents. Participants noted that this situation is exacerbated by the lack of a supply center or large supermarkets in Quibdó, which contributed to difficulties in ensuring a regular supply of products. Similarly, historical deficits of electricity, which remain an issue in some rural communities, have contributed to contemporary dietary preferences. Salted cheese, produced in the northern part of Colombia, is widely available and accessible as its high concentration of salt preserves it on the long trip to Quibdó. Fish, an available and essential food in the region, has traditionally been cured with salt. The fish is salted and then it is put to be dried by sun, and also the meat, and then they are put it in the refrigerator. High dietary intakes of sodium are well established as a key factor in shaping risk of hypertension. 27 Thus, the absence of reliable sources of fresh produce, concomitant with ready access to, and dietary preferences for, salted cheeses and salt-cured fish, together may contribute to high risk of hypertension. Living Conditions and Physical Activity. Walking, exercising at the gym, and participating in public aerobic classes are the main types of physical activity participants mentioned as available in the city. For each of these activities, there were also barriers and facilitating factors linked to geography and social structure, displacement, and uneven employment opportunities. Participants linked the inadequate and uneven nature of infrastructure development across neighborhoods in Quibdó to physical inactivity among residents. Neighborhood playground areas were more likely to be present in the planned areas of the city near to its center. Neighborhoods located in the periphery lacked both recreational facilities and access to transportation systems that might enable them to access playgrounds or recreational facilities. In a climate with temperatures oscillating between 25-30 °C and where frequent and intense rains occur throughout most months of the year, the absence of covered facilities is an important deterrent to physical activity. Recognizing the importance of infrastructure to support physical activity in Quibdó, participants also described the constraints for creating such infrastructure. Particularly in the periphery, areas for building recreational and sport facilities have not been set aside and the town hall faces economic restrictions for acquiring houses or lots with the purpose of building these facilities. Some participants in this study who were responsible for making or enforcing urban planning policies expressed concerns that the acquisition of land for building public spaces in neighborhoods occupied predominantly by displaced people might interfere with the availability of these spaces. Failing to protect housing for populations in critical need might lead to legal problems for the local government or personal accusations of violations of human rights. Thus, obtaining land for recreational facilities in communities with a high demand for basic shelter offered an important challenge to efforts to improve access to these facilities in those communities. Walking, an important form of physical activity, is constrained by the poor conditions of sidewalks and the occupation of those spaces by informal sellers. These conditions are exacerbated by the lack of traffic lights and street signs, the large number of motorcycles in the city, and frequent infractions of transit rules by drivers, with resulting safety concerns for pedestrians. In addition to contributing to traffic congestion, participants described the public and private use of motorcycles as replacing walking as a major means of transport even for relatively short distances. For example: …you get a motorcycle to go the governorship building, get a motorcycle to go from the market to the church, but from the market to the church there are just four blocks. So, you have to pay one thousand pesos [less than half US dollar] for four blocks. Indoor recreational facilities offer an alternative to walking and other outdoor activities. However, although some gyms have been opened in the city, lack of economic resources is a relevant barrier for paying gym fees. As a result, access to gyms is limited for most parts of the population: Gyms? There are gyms, but who goes there is who has the money to pay; who has the exercise habit and the money to pay… Thus, residents of Quibdó confront substantial challenges to efforts to maintain an active lifestyle. As with other resources, both the distribution of indoor recreational spaces and their cost prevent many members of the population, particularly those with restricted incomes, from accessing spaces that support physical activity. Stress and Social Hierarchy. In addition to the conditions that influence both dietary practices and physical activity, participants described stressful life conditions as a factor contributing to hypertension: So yeah, I think my stress led to me having hypertension. While participants described stress as affecting everyone, everywhere, they also noted differences in the sources of stress and stressful experiences across social groups. Participants noted that the lack of job opportunities is the most important and common stressor associated with hypertension for the population of Quibdó, and may be particularly important for those who are most marginalized. Other stressors participants cited included the limited opportunities for having a good quality of life, defined by one participant as the possibility that: …each person has a decent house, education, and basic public services. Food insecurity was also cited by participants as an ongoing source of stress for poor families. People of Quibdó, can be stressed for many reasons: the lack of a decent job, about how to get the daily bread for his/her family. That is stressful. Additional stressors mentioned by the participants were the high levels of noise, traffic congestion, and safety concerns in some neighborhoods. Poor public services were also stressful, as residents experienced challenges associated with access to resources necessary for daily living: …after three days without rains, it is a terrible problem for the general population. It is stressful; people get desperate because there is no rainwater. For displaced people, the trauma of the displacement, memories of lost land, family members and friends, disruption of family and community, reception and treatment received after arriving to Quibdó, stigmatization and marginalization, and difficulties adapting to the city all contribute to profound stress, taking a substantial physical as well as emotional toll. The experiences faced by displaced persons from indigenous communities-a subset of those who have been displaced-are profound. The failure of nonindigenous people to recognize these stressors may be linked to cultural differences in the expression of stress, coping mechanisms that emphasize spiritual or other less visible forms of coping, and by language barriers that constrain the communication of stress. These failures are exemplified by an excerpt from a local leader, who stated B…indigenous people do not have many concerns.R egardless of their socioeconomic condition, participants noted that black women face additional stressors associated with the intersection of race and gender. For instance, black women participants noted the multiple roles they play as mothers, housewives, and workers . Although no comments were made about indigenous women, this may offer another example of the failure to recognize stressful living conditions in this population. Alcohol consumption was considered by some participants to be a common and Bnormal^behavior among the adult population of the city. Participants did not recognize socioeconomic differences in alcohol consumption, but rather it was seen by many as an alternative to the lack of recreational or cultural opportunities and a coping mechanism for stress particularly for informal workers: People go to work and after work the only fun is dancing and drinking. In Chocó there is no more fun. In sum, participants in these interviews linked the structural conditions experienced in Quibdó, including displacement, limited economic opportunities, and the infrastructure of the city, to hypertension risk through multiple pathways, including behavioral risk factors for hypertension and both behavioral and physiologic responses to stress. Further, they described ways in which these conditions were socially patterned, varying across socioeconomic, racial/ethnic, and gender hierarchies. Although several adverse conditions are widely distributed in the city's population, social processes of marginalization and stratification create additional circumstances for disadvantaging those already at the bottom of the social hierarchy . --- DISCUSSION In the public health discourse, hypertension is considered a chronic condition due to its long latency period and prolonged course of illness. 27 Paradoxically, the chronicity of the social circumstances that may lead to hypertension at the population level has received less attention in Colombia. Participants in this study offered explanations-diagnoses, to use the language of frame analyses-that included multiple and interconnected processes, many of which have unfolded over long periods of time, influencing excess risk and the unequal distribution of hypertension in Quibdó. Forced displacement, unplanned urban space, and unemployment that shape living conditions in Quibdó were important aspects of the diagnostic frame that local actors identified. These conditions, in turn, influence risks for and disparities in hypertension in this population. These explanations demonstrate the framework for an analysis of hypertension risk that encompasses clear linkages to social determinants of health in this jurisdiction. There is substantial evidence, discussed in more depth below, to support the connections that these participants are making, linking structural factors and living conditions to population health outcomes. 10,28 As described above, these factors and conditions may act through different pathways and across levels of an ecological model . Forced Displacement and Hypertension. Forced displacement has multiple negative social, economic, and health consequences. [29][30][31][32] Generally speaking, people flee toward areas where they expect better conditions when their integrity is threatened. For many, the displacement is the continuation or even the accentuation of an already poor and vulnerable condition. 29 This fact was clearly pointed out by participants as they suggested that most of the internally displaced population in Quibdó experienced limitations for meeting basic survival necessities, stigmatization, and a variety of difficulties for accommodating to their new life in the city. Although literature has focused more on acute and mental health consequences of exposure to conflict and displacement, 33 those who survive may experience long-term effects that similarly threaten their health and lives, Displacement results in socio-cultural, psychosocial, and physiological stressors. 32,34 These stressors can set in motion a series of physiological changes that have been clearly linked to hypertension [35][36][37][38] as well as other chronic health risks. Displacement may also contribute to the erosion of social ties. Community integration and active involvement in a social network have been found to be a protective factor for hypertension. 39 As a consequence of displacement, people tend to lose their social networks and associated social supports, which in the stressbuffering model is associated with the modulation of stressful events. 40,41 Furthermore, integration into new contexts is a difficult process, particularly when members of the host society are hostile to and wary of internally displaced persons. As noted in this study, feelings of hostility are common in host societies that are asked to accommodate new groups and can be exacerbated in communities that fear competition for already scarce resources. 32 These mistrustful social relationships are an important source of stress for all groups in society and have been associated with a higher level of systolic blood pressure. 42 Second, unfair treatment and stigmatization have also been associated with increased risk of high blood pressure. 43,44 As people who have been displaced move into the city and encounter the types of unfair treatment and stigmatization described in these interviews, they may experience social isolation and chronic stress that, ultimately, contribute to increased risk of hypertension. 45 Finally, displacement may lead to hypertension through the adoption of damaging coping behaviors , and the lack of economic resources for healthrelated behaviors such as healthy diets, as was evidenced in this study. 33,45 Unplanned Urban Space and Hypertension. Participants attributed a key role to displacement as a driver of the urbanization process of Quibdó. Urbanization, regardless of its causes, has been widely cited as an important contextual factor associated with hypertension in low-and middle-income countries . 39,46 However, living in an urban context is not always negative and what may matter for the risk of hypertension among urban dwellers in LMICs might be variations in the features of the urban context. 47,48 Urban areas, in comparison with rural, may be more beneficial for health as cities are generally better equipped with health care and public health infrastructure, as well as other features that have a lasting benefit on health. 49 In contrast, risks associated with urban spaces in LMICs include exposure such as heavy traffic by motor vehicles, limited green space, and dependency of food supply. 28 Some urban areas of Quibdó meet criteria to operationally define a slum. These criteria are as follows: inadequate access to safe water; inadequate access to sanitation and other infrastructure; poor structural quality of housing; overcrowding; and insecure residential status. Two additional attributes that characterize slums are poverty and social exclusion, which are considered to be both causes and consequences of slum conditions. 50 Despite the fact that 1 billion of the world's population currently live under slum conditions or in other marginalized urban settlements in poorer countries, 51 research directly tying poor urban living conditions to hypertension in LMICs is still limited. [52][53][54][55] Studies from African cities, some of which may share urban and sociodemographic characteristics with Quibdó, have found an increase in blood pressure in urban compared with rural populations as a result of current urban residence and/or lifetime exposure to urban environments. 39,56,57 Most of these studies suggest that changes in well-established risk factors for hypertension, such as physical inactivity and stress, are involved in this increase. From a life course perspective, it is possible to recognize multiple predisposing exposures that would increase the risk of hypertension in disadvantaged urban areas. These exposures include, but are not limited to, under-nutrition early in life, low socioeconomic status across the lifespan, crowded spaces that restrict building recreational and social spaces for physical activity and gathering, poor food environments, and multiple social, physical, and psychosocial stressors. 28,52 Economic Circumstances and Hypertension. Policies and urban conditions that hinder job creation and foster unemployment are also expressions of the inability or lack of political will to address social determinants of hypertension in Quibdó. Despite limited and conflicting research considering how unemployment may contribute to the etiology of hypertension, 58 the interconnected mechanisms suggested by participants in this study are plausible, even if not consistently empirically established. The lack of job opportunities is an important stressor for the whole population of the city, and particularly for those in the bottom of the social hierarchy. In fact, stress is one of the most commonly proposed mechanisms in the literature linking unemployment to hypertension. 58,59 In addition, high levels of unemployment in Quibdó have contributed to the proliferation of informal jobs, with some of these employment opportunities linked to well-established risk factors for hypertension. For instance, reduction in walking was in part explained by the excessive number of both informal motorcycles taxi drivers and sellers on the street. In addition, exposure to noise from combusted fuel vehicles as motorcycles has been linked to chronic stress and increased blood pressure. 60,61 Strengths and Limitations. Two strengths of this study should be highlighted. First, it builds on the unique contribution of qualitative research to mapping contexts and answering Bwhat is^and Bhow^questions that are needed to start understanding the social phenomenon in a particular context. 62 In this regard, this study extends existing research on hypertension in Latin American that has particularly focused on individual-level explanations using quantitative approaches. [63][64][65] Second, we engaged local actors in examining the contributions to hypertension across levels of an ecological model. Findings such as these are crucial in public health as they represent the particular social world of participants and extend the few studies aimed at directly exploring lay understandings of the causes of health risks and disparities, 66 as well as the role of place in framing a particular public health problem. Research that actively engages decision makers in thinking and talking about the multilevel causes of excess health risks in their city provides an opportunity to initiate a conversation about the policy-making process and potential initiatives that might be considered for addressing hypertension in Quibdó. Engaging decision makers in such conversations is also an important opportunity to consider specific features of the city that may influence the distribution of hypertension across residents, and to consider important issues of equity as they relate to urban space. This study also has limitations. Because we collected the information in Spanish and conducted the last step of the analysis in English, it is possible that the Englishlanguage description of the findings may not fully capture participants' meanings, feelings, and understandings. In addition, despite the efforts to involve additional key local actors, particularly from the government, it was not always possible due to their time constraints. These actors may have provided additional or different perspectives regarding the diagnostic task reported in this study. It is not clear if the perspective of those who did not participate in the study would have helped to make distinctions clearer. However, the recurrence of major topics across the interviews that were conducted suggests that the study captured critical aspects of this complex phenomenon. --- Conclusion. The social and political actors who participated in this study clearly described social determinants of hypertension in Quibdó. They articulated multiple pathways through which these social, economic, and spatial conditions might contribute to excess risk of hypertension among residents, with particular risk accruing to some of the most vulnerable populations . As described in the literature, 28,66 the analysis presented here suggests that the non-random distribution of social determinants of health emerges from a confluence of factors. These factors and their devastating manifestations and consequences are products of the combination of political and social decisions and processes, as well as the accumulation of vulnerabilities in the population. 28,67 That the participants in this study were able to articulate the role of the characteristics of the place where they live in shaping hypertension risks and disparities, as well as the role of historical and structural factors in this process, suggests the potential for policies and practices across multiple levels of an ecological approach that might be implemented to address these challenges. This approach implies a need for public policies that have a meaningful impact on the social determinants of health. Participants suggested a set of public policies that might be adopted to address some of the structural factors they mentioned. Not surprisingly, they prioritized policies to increase economic opportunities and job creation, regulate urban planning, improve access to public services and physical infrastructure, and allow the population access to decent housing. Undoubtedly, the modification of the factors that foster forced displacement will have a great impact on the social conditions of the population of Quibdó. Even if the local government increases its current capacity for planning and providing services, the presence of forced displacement and a continued influx of displaced persons with few resources would remain as an important factor in shaping the health and well-being of urban residents and social inequalities would persist. Indeed, under these conditions, a successful group of urban residents may improve their living standards, while those left behind will be joined by the newly displaced. Efforts to address the complex and multifaceted challenges associated with the social and economic determinants of hypertension in Quibdó will require concerted attention by policy and decision makers, working collaboratively across multiple units of government . Given that many of the participants of this study are policy and decision makers, it is most promising that they identified these broader social and structural factors as key factors influencing health status. Translating these findings into population health interventions and policy change will be a key step in improving health equity in the years ahead.

Conditions in the social and physical environment influence population health and risk for CVD, including hypertension. These environmental conditions are influenced by the decisions of public officials, community leaders, and service providers. Examining the frames that local decision makers bring to understanding hypertension can provide important insights into the decisions that they make about strategies for addressing this problem in their jurisdiction. The goal of this study was to examine the frames that local decision makers in Quibdó, Colombia, bring to understanding hypertension risk, and in particular, whether and how they use frames that encompass associations between living conditions and hypertension risk. Data for this qualitative study were collected using a stratified sampling strategy. Semi-structured interviews were conducted in 2012 with 13 local decision makers and analyzed using a framework approach. Participants linked the structural conditions experienced in Quibdó, including displacement, limited economic opportunities, and the infrastructure of the city, to hypertension risk through multiple pathways, including behavioral risk factors for hypertension and physiologic responses to stress. They described the social patterning of these factors across socioeconomic, racial/ethnic, and gender hierarchies. Although several conditions associated with hypertension risk are widely distributed in the city's population, social processes of marginalization and stratification create additional disadvantages for those on the lower rungs of the social hierarchy.

Introduction There is a war in the online world for gender and development mainstreaming because of the massive prevalence of gender-based humor. These jokes are powerful messages which can pierce the campaign for mainstreaming because there are message characteristics of these jokes which may slow down the success of GAD campaigns. This study employed discourse analysis as the approach for utilizing social components of communication and the ways people use language to achieve certain effects required in changing what has been perceived in reality and in this case, to breakdown message patterns into language features of the gender-based humor online which can play as a contrarian mainstreamer for GAD. Gender mainstreaming is aimed to help government officials, field personnel, and customers of various organizations . There are GAD capacity-building programs available to assist officials and bureaucrats in their efforts to mainstream gender . Gender and development mainstreaming is the gold standard for achieving gender equality and protection. The community is dedicated to supporting gender equality, recognizing gender differences, and allowing for varieties determines this . Upon learning the discourse language patterns of messages within these online humors, the key players of the GAD advocacy and the language educators can fine-tune campaigns on how to deconstruct language mainstreaming campaigns in social media. The purpose of this study was to examine the language message patterns that are hidden in the coats of humors specifically on language-promoting discriminatory practices, language reproduction of gender bias, gender-stereotyping in language use, and disempowering tool to personal images. This research attempted to uncover the contexts of these messages by mapping out the themes and purposes of gender-based humors. Conventionally, this study does cover the humors and its message patterns beyond the scope of culture and social classes. Sexist jokes, for example, have been used to discriminate against people based on their gender orientation, identity, and roles, and they are sometimes disguised as comedies or jokes that others may overlook. Language choices in regular speech that are sexist perpetuate gender stereotypes of men as capable and women as communal . Because of the nature of social media, people's attitudes have evolved, allowing for discrimination in comments and dialogues across the mechanisms of virtual engagement . In specific portrayals of humors online, males are depicted as capable, whereas women are portrayed as communal. People's opinions have developed because of the nature of social media, allowing for discrimination in statements and debates through virtual community . As this study moves to unveil the characteristics of the discourses in humors online, it wants to proceed to identify the different sources of these types as well as languages and how their sources of contexts related to the traits of the humors they form and use in conversations and comments. --- Research objectives --- Theoretical underpinnings of gender-based humor Humor also functions as an escape from situations that are stressful. Moreover, definitions of comedy frequently focus on the intended message of the speaker or audience. Humor is "a particular sort of communication that establishes an incongruent relationship or meaning and is delivered in a way that causes laughing" . Undoubtedly, more research is required in assessing humor and gender over the years . This is being done to determine if there are any connections or distinctions between the gender humors from previous studies and the most recent humor phenomenon. However, the rise of humor in social media-imposed concerns about its social implications. In particular, hate speech is targeted at certain people or groups who are deemed inferior based on fundamental identity characteristics or demographic indicators , promoting the stigmatization of an individual or people . It constitutes an intrusion of human integrity and establishes attacks on specific individuals or groups . To frame this study, social dominance theory was established. Essentially, the SDT individual and institutional aspects that lead to several kinds of group-based oppression are emphasized. All common forms of communal oppression are viewed as special manifestations of a broader human tendency to build and sustain group-based hierarchies . --- Humors as discriminatory practices In the context of discriminatory practices, gender-based humor that is derogatory towards women can be seen to reinforce the notion that women are inferior to men. As described by social dominance theory, discriminatory acts are carried out by individuals with particular behavioral dispositions, subgroup allegiances, and social identities, within certain social settings, frequently in relation to the conduct of social institutions and roles in society and rooted within cultures with specific social views and structural ties . Instead of being a movement in psychological reductionism , the social dominance theory seeks to shed light on how psychological perspectives, communal identities, social circumstances, societal structures, and ideologies of culture all interact to generate and perpetuate social inequality based on groups . This leads to the initial assumption of the study that discriminatory practices of people in social media through gender-based humors have specific connotations from hierarchy and social strata. Gender role differentiation and group segregation also result in unequal distribution of privileges and responsibilities across arbitrarily determined groups and gender . This does not imply that individual behaviors are determined by their status in society. In contrast, a hierarchical structure implies that it is easier to perform actions that maintain or increase inequality than actions that diminish the hierarchy. There are strong differences among people with the same social class in terms of what groups their actions favor, the extent to which they discriminate, their level of group prejudice, and how firmly they choose to discriminate compared to egalitarian policies . This is an indication that the behaviors of individuals are not solely influenced by their position in the social structure. --- Humors as gender bias Gender-based humor that reinforces traditional gender roles and stereotypes can perpetuate gender bias by reinforcing these attitudes and behaviors in individuals. This can lead to the reproduction of gender bias in language use and other aspects of social interaction. In relation to the social dominance theory, persistent inequality between groups is preserved in part by the inappropriate use of power towards groups that are subordinated . One significant type of confluence involves the categorization of individuals into different hierarchies-attenuating and hierarchy-enhancing social roles and social institutions. Males score higher than females on social dominance orientation; they are significantly disproportionately represented in hierarchy-enhancing roles, e.g., law, finance, and military, while females are overrepresented in hierarchy-attenuating roles, e.g., charity works, and social work . Empirical evidence demonstrates that this is the result of "self-selection, institutional discrimination in employment, ideological socialization, and differential feedback and attrition" . Similar selection processes are applied to hierarchy-enhancing roles within groups perceived to be dominant, who also score stronger on social dominance orientation and uphold hierarchy-enhancing legitimizing beliefs compared to inferior members of the group. People of dominant social groups typically receive an imbalance of desirable and symbolic resources, including political influence, wealth, physical security, leisure activities, and educational opportunities. The negative social value can be disproportionately imposed upon or left to people from subordinate groups in the form of disproportionate punishment, vilification, and stigmatization . Even though the extent, seriousness, and defining foundations of group-based hierarchical structure vary throughout communities and over period within their community, group-based hierarchical structures appear to have become a human universal . --- Gender-stereotyping in language use Gender-based humor that relies on stereotypes can reinforce these stereotypes and lead to genderstereotyping in language use . This can lead to a range of negative consequences, including discrimination and the perpetuation of gender inequality. Stahlberg et al. found that masculine generics elicit a male bias in mental images and cause listeners and readers to consider male exemplars of an individual categorization more often than male exemplars. Women are not given authority because they appear incapable of upholding it "… as demonstrated by their linguistic behavior…" . In line with the theory of dominance , women are viewed as oppressed bodies whose language is dominated by males. Several factors contribute to the development of linguistic differences between males and females. It was implied that masculine language was more abrasive mature, and forthright or pointed . These expressions include mhm, yes, and right. These words are frequently employed by males to assert dominance. Men frequently use the words gimme, gotta, and going to when giving instructions and orders, particularly when speaking with other men . Third, it was commonly believed that males used more profanity than women . --- Disempowering tool to personal images When ideologies become standard, they can exert a significant impact on behavior and on individual standards for the behavior of others. Social interpretations have constantly recognized the significance of social norms for human behaviors, and for recognizing discrimination and coordinated action in relationships between groups, social norms are crucial . Gramsci was mindful of the significance of ideological norms for preserving the authority of the dominant group. To normalize the social status of dominant groups, he argued that dominant individuals controlled the strongest normative ideologies as well as the language that expands these views. By establishing dominant group authority and making their social status appear innate, unavoidable, and positive, dominant groups could prevent subordinate groups from compromising the unequal society and preventing any resistance from arising from subordinate groups . In the case of cyberbullying, adolescents report feeling angry, sad, and hurt . Also of concern, they are poor concentration and subpar academic achievement . Gender-based humor that objectifies women can be seen as a disempowering tool that reduces women to their physical appearance and reinforces the notion that women are primarily humorous objects. Gray and Ford argued that only 20% of respondents considered sexist remarks and crude language to be harassment. This further extends the assumption of this study that humors can be a channel for disempowerment tool to personal images. --- Methodology --- Research design The research design is based on the exploratory research by Saunders et al. and Singh which deeply examines the narratives of the participants from the different orientations. It aimed to go deeper into research issues which may be useful for further analysis from the pathways of discourse analysis. The process of using discourse analysis will uncover important contextualization of the gender-based humors online as with the reactions of the participants in the study as they belong to diverse profiles. --- Population and sampling technique Purposive sampling was used to collect data for this investigation. The participants were initially determined as social media users who spend at least 1 h on social media. Essentially, purposive sampling ensures that the participants share specific characteristics , i.e., social media users who spend at least 1 h online. The social media sites that participants used are presented in Table 1. They were coming from various orientations and profiles to deeply portray the origins of online comedy manifestations. Purposive sampling, also known as purposive and selected sampling, is a method used by qualitative researchers to discover individuals who may give particular information on the issue under investigation . The qualitative researcher determines the qualifying criteria that each participant must satisfy in order to be picked for the research study, which is very subjective . There were 14 participants for this study who spend at least 1 h using social media daily and actively engaging for communication or for commentaries and conversations. The profiles of the number of the participants were showing in Table 2. --- Sampling technique Purposive sampling was used to acquire participants since this demands certain profiles, roles, and qualities. Purposive sampling may be advantageous when just a small number of people are available to serve as primary data sources due to the nature of the study design and aims and objectives. Purposive sampling investigates how to generate a sample population even if it is not statistically representative of the larger community under study, which is why discourse analysis was [i] anyone who is holding a public office, who leads a certain department, or who has a certain role in a publicly established office , he/she has to be in current position for more than 5 years; [ii] anyone who is in private institution and in public institution who has a high authority to certain department or office . 2) Educational attainment-[i] one who had been and graduated in higher education from either private or public university; [ii] one who had finished either high school, or elementary, and has access to social media platforms. 3) Gender roles-[i] anyone from any age who consider themselves as a member of any gender-based groups . 4) Generational classification-[i] two from adult bracket , two from millennial bracket . utilized. Purposive sampling allows information to be extracted from data. --- Research instrument To cover the extraction and analysis of the research, the study used an online gender-based-mapper, one-on-one interviews, and FGD-oriented tools. All participants were invited to the interviews and FGD, were informed of the goal of the study, were informed of their rights and privileges as participants, and were ensured of the confidentiality of the information retrieved. Consent was requested for the researcher's dissemination efforts. 1) Online gender-based humor expressions mapper: To identify prevalent discriminatory gender-based comedy expressions in social media postings, comments, and discussions. The researcher requested permission from the participants to monitor and submit inventory posts on humors. These were accomplished by taking screen photos or copying and pasting them into a Word document. This is accomplished by an examination of the degree of gender-based comedy, how people react to it based on gender, how marginalized individuals react to it, the existence of tolerant behaviors, and the repetition of actions over time. Identification tags on these documents were deleted during the study process and the researcher's dissemination activities. 2) Online humor-mapper and message pattern validators: To determine the message patterns and themes of online gender-based humors based on language promoting discriminatory practices, language reproduction of gender bias, gender-stereotyping in language use, and disempowering tool to personal images. This was a one-on-one in-depth semi-structured interview with all participants to validate the message patterns and themes found in their social media activities. There were two extractions of data in this part as well. The first was to categorize the message patterns and themes from the inventory of humor posts, commentaries, and conversations from the social media account of the participants. The second extraction of data took place through one-on-one interview with the respondents to validate and analyze their humor posts, commentaries, and conversations on social media. --- Data collection procedure The study used inventory monitoring, interview, and focus group discussion as the methods to gather in depth information from the participants as well as their humor expressions on social media in terms of statements, commentaries, and conversations. The monitoring-inventory of humor expressions covered their activities in the social media from November 2021 and twelve weeks earlier. The monitoring-inventory served as the collection of gender-based humors in social media. The inventory in Table 3 is collections of gender-based humors collected in different social media platforms that the participants are using. Most of these were among their friends and contacts. This inventory was presented to the participants during the interview and FGD. In conducting an FGD, the researcher used the guidelines from Hollis et al. . As noted, " [One] ingredient in a successful focus group is the questions or exercises that are used to stimulate ideas" . The researchers used the guide questions from Table 4 in collecting the data through FGD. This, along with the inventory in Table 3, stimulates the ideas of participants towards gender-based humors and its message patterns. What are the common traits or characterizations made for genders ? Are these portrayals of genders good and appropriate for them? Is it possible that these discriminated genders be treated as such by the rest of the population ? Can humors or jokes be converted into beliefs by others who read or listen to humors or jokes on social media? If I say that women are weak, or gays are cowards in humors, others will believe that they are? How do you think humors and jokes are constructed as a language in social media? Language reproduction of gender bias Gender-stereotyping in language use Disempowering tool to personal images --- Data analysis Based on Kiger and Varpio , semi-structured interviews and focus groups were used to validate these discourse indicators in order to gain a thorough understanding of the origins and circumstances of these humors. Generally, "unstructured interviews [serve] as a natural extension of participant observation, because they so often occur as part of ongoing participant observation…" . Coding and a thematic approach were used to determine the paths of the objectives as well as to reveal probable emergent themes from the data extraction and analysis process. Thematic analysis enables the researchers to translate and communicate the data . Thematic analysis is a powerful analytical tool for qualitative research because it is a versatile qualitative data analysis method that may extract information in an inductive or deductive manner. This study utilized the process outlined by Lochmiller where the researchers categorized the responses into smaller and compressible themes that share similar and relevant codes. In coding, the researcher should "fracture" or separate the narratives into tiny bits of information . --- Results --- Research objective 1: Common discriminatory gender-based humor expressions in social media in terms Table 5 showed that gender-based humors from fourteen participants for the last 12 months starting from November 2021 with the content on women as being hard to understand, overacting, and unreasonable comprise the highest frequency posted and commented in the online. It is followed by women being objectified based on their bodies, being portrayed as weak, dependent, as well as being the subject of entertainment on the online jokes. Furthermore, the LGBT community is also portrayed with different characterization in gender-based humors online including language content pertaining to their lesser rights compared to other genders, being weak and being easy topics for funny conversations. Table 5. Sample inventory of gender-based humors online based on themes. --- Posts Common discriminatory gender-based humor expressions 1 Do you think you can win an argument with a person who stands like this for 7 h just to look pretty? Women use high heels just to look pretty and arguing with them is not a good choice. 2 Ariel, why do you wear seashells? Because B-shells are too small and D-shells are too big. Bossy shaming of women based on their bust circumference. --- 3 Why can't women ski? Because there's no snow between the bedroom and the kitchen. Women should only be limited to doing household chores. --- 4 What's six inches long, two inches wide, and drives women wild? Money. It is described that women are concerned about wildly attaining money. 5 Philippine presidency is not a job for women. Women don't have the capacity to be president. 6 Q: Is Google male or female? A: Female, because it doesn't let you finish a sentence before making suggestions. Women don't allow someone to talk before reacting. 7 Did you hear they finally made device that makes cars run 95% quieter? Yeah, it fits right over her mouth. It describes sexually cutting off women to remain silent. 8 I'm not sexist. Sexism is wrong. And being wrong is for women. Women are described to be always wrong. 9 Sometimes, these feminists are really a bit OA. I mean, that's funny. Come on. Just laugh. Sexist jokes are just normal. The following gender-based humor materials were gathered from the online comments of the research participants from November 2022 and 12 months earlier. --- Research objective 2: Message patterns and themes The emerging trend of humor in social media was represented by message patterns and themes based on language promotion, biases and disempowering. LGBT members should be treated as jokes. 5 None of my straight friends are this straight. They love my bad gay jokes. LGBT members are topics for gender-based jokes. 6 What I find funny is that all my past friends have been bisexual. LGBT shall be unfriended. 7 Apparently, I don't have any straight friends, even the ones who appear to be. Gender expression is being neglected. --- 8 Bold of you to assume I have any straight friends. All his friends are gays. 9 Imagines a couple of girls. Married for thirty years...when one of them snaps and asks the question that's been bugging them the entire time. "Are you gay?". Gays cannot marry a woman. --- Language promotion as discriminatory practices There are different targets for the language promotion based on the intent of the gender-based humors online. The audience perceived reactions and language are the prevalent aspects of language promotions. Audience reaction is dependent on how particular gender-based humors "go viral" in the social media and this is where the message proceeds to the many actors in the social media. The algorithm of the online world is easier for the gender-based humors to propagate and reach its target audience or even the unintended audience. This is an open space for the attached discriminatory and sexist content and meanings to spread effectively and continue its influence on diverse audience. That is why Asemah et al. admits that the online space has shattered communication challenges by accommodating people to openly express their thoughts, opinions, and the many aspects of their lives. Because there is massive distribution and high impact on individual persons and cyber-communities, the prevalence of jokes with sexism and hate speech on the internet has become an expected concern. "They give negative characterization. Sometimes it is spreading. We are offending the specific gender. It can affect their emotions. It can affect their confidence. Sometimes, I can say that for women, the jokes are offensive. Not the natural way they react. Not all of them will react similarly to these kinds of matters. It can affect their emotions. Some are not true. What is trending… because a lot of people share it, the reader will believe."- "For me, I notice that they are putting pictures, they use words so others would understand. They use colors for backgrounds because sir image ad different backgrounds tend to attract. They will be curious. They will read it… after that, they will absorb it. They think what they read is real or true which is actually not."- --- Language reproduction of gender bias The messages of the gender-based humors are accepted and reproduced according to the type of audience or recipients on the web. Message patterns of the gender-based humor online are important dissection for the development and reproduction of gender biases which may have many routes with respect to gender and development and the actors advocating it. As more extensive representations of the messages from the humors online, gender biases and other forms of gender disenfranchisements, were patronized by the type of audience in terms of gender roles and preferences. These biases are particularly patterned on the "imbalance role" treatments among genders, particularly in "characterizing" certain genders, like women and LGBT as "weaker" and "softer" genders against the males or even against themselves. That is why Jiang et al. had to advocate for minimizing language biases on gender by transitioning into gender-neutral language: to reduce gender bias, different peoples' language patterns should be modified by making these audiences aware of the advantages of gender-neutral words, phrases, and language usage. "The characteristics is that women are different from other genders like straight. They cannot do the same things. Women are weak. LGBT is not right or acceptable."- "Comparison of the capability of the genders, women, or men. Normalization on what is acceptable to a woman or man as they relate to how women react or respond or think..."- "I think jokes are made for fun. In fact, it involves certain sectors. So, it makes the jokes not good anymore..."- --- Gender-stereotyping in language use The language on gender-stereotyping is one of the most common message patterns extracted from the sources of gender-based humor online. Recognizable from these humors are themes related to the traits of the actors in the humor itself. These message patterns are particularly elaborations of stereotyping in terms of "inequality", "characteristics", "qualities", and "capacities" of a particular gender group. Perception and assumptions determine the stereotyping tendencies of certain genders. These progresses as some gender and social profiles generalized the traits of certain genders in the language of underestimating their abilities, weakening their capacities, and subsequently defending their assumptions through their experience and group validations. Because there are message gaps in the humors, Budziszewska et al. reiterate that the prevalence of communication barriers and gaps influences word delivery, which is causing the humor language to be misconstrued in general. Stereotyping intentionally or unknowingly produces language that presents "bad images to someone" while representing it as a humor. Stereotyping was "illustrations" and "representations" of genders being portrayed negatively as these are formed by the author of the humors and the distributor of the humor material. Generally, the "illustration" of stereotyping for certain genders is automatically considered as negative portrayal of women and the LGBT members. --- "Stereotyping is underestimating their capabilities in attitudes and movements of the females."- "I think especially on the illustration, it tries to give bad image to women. Bring them to lower level in terms of the gender, where men are better…"- --- Disempowering tool to personal images Disempowering message pattern appears to be one of the most critical aspects in characterizing gender-based humor in social media because it directly and indirectly affects the campaign dynamics of mainstreaming gender and development. Much more dangerous because it empowers potentially powerful actors in the social media to continue the discriminative and sexist of the humors prevailing in the vast arena of the web. Message patterns are very powerful tools that target the very effect of the humor's content to the very identity of the gender group or groups because of its feature to characterize them in bad light such as their "abilities", "capacities", and "opportunities" to do it. This is the very initial impression of Peters where he contends that males possess innate qualities in them such as aggression that proceed to more power as well as social status. "Based on the materials, depicted in the messages, women are less dominant. Women are vulnerable to sexism. They shouldn't have equal rights…"- "With women here, women are just a homemaker, not capable of being handling positions, even presidency positions. Discriminatory, like not being able to become a doctor, lesbians are characterized as nobody…"- "I think they will believe. For gays, they have to be coward. I think somehow, when I encountered jokes on genders parang ma-change ang belief ko . It becomes a part of my belief. When I get to react to it, on those jokes it becomes a belief…"- --- Levels of beliefs and conversion A very important theme emerged in the study, specifically on nuancing and analyzing the message patterns on "beliefs" and "conversion". The narratives suggest that social media users, especially those with low educational background pose more susceptibility to passing through language and humor as a "belief " and a "social norm". From the other lens, those who advocate and who are aware of the progressive nature of GAD mainstreaming have less susceptibility to adapting to the culture surrounding the gender-based humor online. Conversion of these humors distributed online solidified beliefs and posed serious aspects for "discriminatory practices" as well as carrying the "influential" trait to other web actors. Segments of the online players who could recognize legitimate classification of neutral and sexist language from these humors were "less manipulated", considering the potential "misinforming quotations" from someone in power. Hodson et al. was correct in this assumption that gender-based humors or jokes are not thought to self-reflect or encourage biases due to the belief that it has lower perceived offense as created by sarcastic elements and very indirect communication. This is the possible reason why beliefs and conversion happen because these groups do not see anything bias among the humors online. "For some sir they do, for some they don't. It depends on the person how they understand the joke. For those who believe in these jokes, these are people that are less knowledgeable about live. Like if, you are not easily believing in these jokes. Those who less believe, are having knowledge of the meaning of these words…"- "I think, especially if consumers or users who are less educated are easily manipulated that these jokes are true. Example on women who are not fit for presidency, that followers will listen and believe…"- "At some points… But not all are converted because people search for it if they have spare time. To know what is true and not, we should choose what is right…"- --- Discussion Language promotion as discriminatory practices. The other aspect of promotion is the language used itself. The language used becomes a tool in promoting the content and message of the content including aggression . The type of language used in the gender-based humors online actually becomes a tool for comprehension of humor, including the use of "informal" and "unstructured" tones of the language which adds to the element of justified appeal to the humors towards a large portion of the population immersed in the social media . The language of the humors is simplified, giving it more relatability to the audience or potential recipients. As it is relatable and simplified, the gender-based humors are easily distributed to as many parts of the population even if it has cyber bullying content on genders . For its implication, this means that whatever gender-based undertones of discrimination and sexism, are easily promoted to the internet users who are exposed to the gender-based humors. Bergmann painted a clearer picture where discrimination, and sexist attributes are reinforced through the language of sexist jokes in the comfort of the features of humors which is spread easily because of its language features. Language reproduction of gender bias. Cameron believed, the implication was clear that from the conversations surrounding gender-based humors, "comparison among groups" was a prevalent pattern among the online actors. This type of comparison aids bigger pictures of biases against certain genders because the comparison releases direct differentiation of traits from the origin of the author of the humors who may not really represent the group or population's opinion towards the perceived traits of the "compared" genders . Manifested languages of biases shape into "one is better than the other" or "they cannot do the same things". Subsequently, these reproduced language of gender biases perpetrates on bigger audiences because there is a neutral ground for its control and corrections. This is supported by Tajfel , where gender biases come in different forms of humors expressions and language and groups. These comments or conversations are remnants of classifiable discourses which may promote gender biases especially in repeated interaction and frequency of use in arguing or long discussions, taking the language reproduction to longer survival on the Internet and even redistributing through sharing online. Gender stereotyping in language use. The implication supports Eagly expands the issue on stereotypes in social dynamics, where gender stereotyping is not really content-neutral since reflect power and status imbalances that favor the males and are associated with the social performances and roles that go with their perceived traits, including those who are in power and those in the mass segment of the society , stereotypes in the gender-based humors represented the message for language that demean or underestimate character and qualities of the compared genders . In the long run, it is an important move to clarify through language campaigns the levels of stereotyping where it already endangers the images of certain genders being forwarded by the gender-based jokes . Disempowering tools to personal images. The implication is on point. There are clear demarcation lines that tell humors based on gender characterized women as weak and fragile as they show them in the language of "nobodiness" or "nothingness" . These message patterns validate its disempowering element as they can discredit what certain groups can do. In many instances, "cowardice" is somewhat associated with LGBT , and this becomes a personal image and identity to that particular group as opposed to men who are touted as the stronger roles . This very image proceeds and extend its effects in "discriminating personal images". Gays are tagged to be easily dominated and subsequently disempowered by their ability to be equal in the standards of the society. This is very aligned with the idea of Eagly that there are power imbalances that favor males and are associated with the social roles that go with them which affects how women are portrayed in the societal strata including in the workplace. In many perception games on gender, women lose in the arena of equal playing field. Menegatti and Rubini succumb to a firm assertion that female characters are cast as the powerless or victim since they are more docile and sensitive. The woman's "ability of women to hold executive positions" is one of the most remarkable messages extracted from the narratives among the online actors. The very humor about gender neutralizes the fair characterizations of women in the workplace or any portrayals in the society. Women are magnified in the humors as the gender with "less" leadership traits, as echoed by Agassi ) and Blumberg ), that primary or supportive employment outside the home is not for women. Conversely, these portrayals may not necessarily fly in the belief system of some online actors because leadership as competence can be associated with women being a "homemaker", but these beliefs are minority in the population. Some of these progressive thinking about role portrayals is still dominated by personal images indicating women, as well as LGBT, as the weaker actors, as compared to men which further fall prey to more discrimination against women . --- Level of beliefs and conversion. There is no definite mechanism to control and regulate highly discriminatory beliefs which means that accountability will not be enforced to anyone who believe and spread the dangers of language in the gender-based humors. This emergence of these promoters and converters of the sexist and discriminatory humors are more alarming if they are coming from the group with authorities and position in the society. But unlike Bill and Naus , they affirm that public domains are presumably based on the belief held by the general public and humor theorists alike that there is gender-based humors online with negative consequences. Specifically, it is thought to create and reinforce beliefs of stereotypes of social groups and, thus, perpetuate prejudice. Also, there is even lesser accountability as the authors and distributors of the discriminatory and sexist humors are coming from anonymous actors, thereby continuing the efficacy of believing and converting recipients into assuming that these humors are normal and tolerable. --- Conclusion Gender-based humor online has multiplied language use in creating messages that express biases towards women which proceeds to send expressions that they are lesser genders in many jokes online. As part of its strong traits, these humors online cannot be controlled or regulated because there are no institutionalized policies to police discriminating or sexist jokes. This study was able to explore complex social contexts within gender-based humor and its linguistic implications. Language promotion as discriminatory practices indicated that gender-based humors were unstructured such that its message transcends across social media. Discriminatory practices in essence indicate that women were inferior to men because of their demographical characteristics. Moreover, it is notable that gender bias upholds more message patterns that pertain to comparison between men and women, i.e., differentiation, disproportionate representation, and between-group segmentation. In terms of the general implication of gender-based humor online, it served as a channel for disempowering social images that it focused on instilling social norms as standards to be attained by subordinate groups. Nevertheless, this was able to identify theoretical components relevant to social dominance theory. Particularly, there is a need to extend the definitions of social dominance in social media especially because social media heavily relies on linguistic and visual messages. Future study has to consider complex human relations beyond the scope of messages considering humors act as both positive and negative reinforcers. --- ---

This study explored the message patterns of gender-based humor in social media in different layers of discriminatory practices against certain genders, language biases against women and LGBT including elements of stereotyping and disempowering tools against the personal images of subordinate genders. This research used discourse analysis based on the mapped-out online posts and comments of the fourteen ( 14) profiles of individuals and extracted their important testimonies based on the collected online gender-based humor to elicit the message patterns. Gender-based humor online enhanced the language use in creating messages that express biases towards women and the LGBT. Humor has both implicit and explicit messages that stereotype women and LGBT as weak and slow. These senses of humor also disempower the women and LGBT's personal images as groups who are easily dominated or are cowards. As asserted, gender-based humor posed a threat to community as it highlights hierarchy-enhancing social roles. Gender-based humor in social media appeared as a mainstreamed form of social differentiation.

PREVIOUS RESEARCH REGARDING MILITARY TRANSITIONS These insights regarding educational transitions can be applied to military transitions, such as enlisting in the armed forces, being assigned to combat occupations, deploying to war zones, and experiencing combat. A large body of research has examined the first military transition, enlisting in the armed forces, and has shown that the people who enlist in the military differ from those who do not in terms of their family background. Men were more likely to enlist in the military if they grew up in families with lower socioeconomic status . They were also more likely to do so if their parents had lower educational attainment . More recent research has suggested that men from both the bottom and the top of the income distribution are under-represented in the armed forces . At the very least, these findings suggest that servicemembers differ from civilians in terms of their class background. Previous research also suggests that people are more likely to enlist in the armed forces if they come from the middle of the academic achievement distribution. On the one hand, people are less likely to serve in the armed forces if they have lower cognitive test scores . On the other hand, they are more likely to enter the military if they had poorer high school grades . And they are less likely to enter the military if they have college aspirations and greater abilities than they are to go to college . Only two articles have examined the second military transition, concluding that servicemembers are assigned to combat positions not on the basis of race but on the basis of aptitude. During both the Vietnam and Iraq wars, black service-members were not disproportionately likely to serve in combat positions . However, one article finds that the service-members were assigned to these positions on the basis of test scores. Service-men were more likely to serve in combat positions if they had lower entry test scores than if they had higher scores . These findings suggest that service-men are more likely to serve in combat roles if they have lower academic aptitude. Relatively more research has examined the last two military transitions and indicates that service-members may have been more likely to be exposed to or killed in combat if they came from disadvantaged backgrounds. During the Vietnam War, men from families with lower socioeconomic status were more likely than those from families with higher socioeconomic status both to serve and fight . Similarly, men who came from families in which the fathers had less education and lower status and from neighborhoods with lower median income were more likely than more privileged men to die in combat . In 2003In -2006, blacks were overrepresented in the armed forces but were not over-represented among the casualties . These findings suggest that the men who see combat differ from those who do not in terms of class, though not of race. --- COMBAT AND CLASS BIAS Previous researchers examining these military transitions have been guided, in part, by efforts to test the class bias hypothesis. This hypothesis states that people are exposed to combat on the basis of their family background . According to the hypothesis, men who see combat are more likely than those who do not to come from poor and workingclass families. Men from wealthier and higher status families are less likely to serve and fight during wartime . Such differences in combat exposure could stem from one or a combination of two processes. First, men from families with more resources may be less likely to enter the military, which could account for the fact that they are then less likely to see combat. If this process governs selection into combat occupations, servicemembers should differ systematically from civilians. However, the service-members who see combat should have similar socioeconomic status to the service-members who do not. Second, men may be equally likely to enter the military at all levels of socioeconomic status. Once in the armed forces, however, service-members may be more likely to be assigned to combat occupations if they have fewer than if they have more family resources. If this process governs selection into combat occupations, service-members in such positions should differ both from civilians and from other service-members. Most scholars describing the class bias account leave the mechanisms underlying the association between class and combat unspecified. However, class bias in combat may stem from one or both of two possible mechanisms. First, men from families with more resources have, net of skills and abilities, greater access to education and therefore human capital . They may use this human capital to get civilian rather than military jobs and thus avoid the chance of combat. Alternatively, once in the military, they may use it to enter non-combat occupations . Second, men from families with more resources have, net of skills, abilities, and human capital, greater cultural capital . They learn skills and values that enable them to successfully navigate bureaucracies . They may use this cultural capital to obtain civilian jobs or higher education, rather than to enter the military, and thus avoid the risk of combat. Alternatively, once in the military, they may use it to obtain non-combat positions. The preceding reasoning leads to the following predictions: Hypothesis 1: Among all men, those from disadvantaged backgrounds should be more likely than those from advantaged backgrounds to serve in combat occupations. Hypothesis 2: Among service-men, those from disadvantaged backgrounds should be more likely than those from advantaged backgrounds to serve in combat occupations. --- SUPPLY, DEMAND, AND MILITARY OCCUPATIONS While many researchers have evaluated class bias in military outcomes, few researchers have assessed how such outcomes may be affected by supply and demand .1 Sociologists and economists, however, have previously assessed the relative importance of supply and demand factors for determining how people come to work in civilian occupations . Supply factors refer to the characteristics and choices of workers. Demand factors refer to forces that are outside of workers' control, such as institutional choices, employer preferences, economic trends, and discrimination. Apparently only one previous study has applied the concept of supply and demand to the types of people who serve in the military. It was conducted by economists and evaluated how supply and demand affected recruits in terms of their average quality, as measured by test scores . The researchers assessed how supply-side factors affected troops by looking at external trends and events that affect the choices of people. They calculated, for example, the ratio of the average pay that people receive when they are in the military to the average pay that people receive in the civilian labor market. According to this view, people should choose to enlist when they would earn less as civilians than they would earn in the military. The researchers also measured demand-side factors by looking at external trends and events, but at those that affect the choices of the institution, in this case 1 In the following discussion, I loosely adapt the terminology used by economists regarding supply and demand and juxtapose the impact of measures of these concepts with the impact of measures more commonly used by sociologists to capture family background and class. Economists use the terms supply and demand to describe how prices and volume are determined in markets, including the labor market. With reference to the market for products, they refer to supply factors as those that affect producers and demand factors as those that affect consumers. With reference to the market for labor, they refer to supply factors as those that describe the suppliers of labor, the potential workers. Typical supply factors include the preferences and abilities of workers, including the wages that workers would like to earn. Similarly, economists describe demand factors in the labor market as those that apply to the consumers of labor, potential employers. They describe demand factors as referring to the number of workers that employers would like to hire and the wage that employers would like to pay. The intersection of these supply and demand preferences, or curves, determines the number of people who will work and the wages that the workers will be paid . In this article, I adapt this framework loosely to describe how the military labor market works, labeling as supply factors those that refer to the preferences of the suppliers of military labor, the potential recruits, and demand factors those that refer to preferences of the consumers of military labor, the armed forces. Instead of examining how these factors determine wages, however, I evaluate how these factors may determine the observed characteristics of the people who serve in the military. I also juxtapose how supply and demand factors shape these observed characteristics with how these characteristics relate to family background, particularly focusing on race and class. the military. They measured whether the nation was at war or conducting a draft. When the nation is at war, the military will need more soldiers, and therefore the demand will be greater. When it is conducting a draft, the military will need fewer volunteer soldiers, and therefore the demand will be less. In this economic framework, all potential servicemembers are affected by supply and demand factors in the same way . The following analyses, however, adopt an alternative concept of supply and demand factors that has been used to examine how these factors lead the members of one group to have labor market outcomes that differ from those of the members of another group. Using this framework, scholars evaluate, for example, why women and men tend to work in different occupations . In this case, they conceive of supply factors as reflecting whether women and men have different characteristics that make them suited for different types of jobs. Women may prefer to work in office jobs, while men may prefer to work in mechanical jobs, either innately or because they are socialized to do so. These researchers describe demand-side factors as reflecting whether employers prefer to put different types of workers in different jobs. Employers may expect women workers to be better at working with people and men to be better at working with things. These researchers acknowledge that supply factors could produce the same outcomes as demand factors . On the supply-side, for example, women could prefer to enroll in the humanities disciplines in college, while men could prefer to enroll in math and sciences. Women would then earn less than men on the basis of their different educational credentials. On the demand-side, however, employers could value more technical work because it is performed by men . Men would then earn more than women on the basis of employer choices. As a consequence, research in this vein often cannot disentangle the effects of supply and demand as definitively as described above for the economic research on supply and demand in the military. --- Supply Factors and Combat Occupations Some previous scholars assess how supply factors influence military outcomes. They focus on how the characteristics of service-members are affected by people's motives. Military sociologists, for example, evaluate motives for joining the armed forces according to the Institution/Occupation model . They juxtapose an "institutional" framework, which sees military service as a calling, with an "occupational" framework, which focuses on the material benefits of service, such as transferable skills and high pay . Recent theory argues that institutional factors, or symbolic rewards, and occupational factors, or material rewards, represent overlapping continuums . Similarly, recent empirical research suggests that people join the military for multiple reasons, some of which fall into institutional or symbolic categories, while others fall into occupational or material categories . Some previous research also assesses people's motives for choosing particular military occupations. For example, previous research suggests that Israeli soldiers are socialized to prefer combat occupations . Economists also implicitly emphasize how supply-side factors affect military service, drawing on human capital theory . They focus on the material rewards of service, such as pay and transferable skills. Accordingly, they test the impact of the relative wages paid by the military on the odds that people will enlist. During the draft period and the early years of the AVF, the wages paid by the armed forces were not as high as the wages paid in the civilian labor market . Therefore, people who had greater skills and academic achievements were deemed more likely to want to continue their education or to enter the labor market than to enlist in the military. This reasoning leads to the following hypothesis: Hypothesis 3: Men should be less likely to enter the military and thereby to serve in combat occupations if they are in the top of the academic aptitude and educational attainment distributions. --- Demand Factors and Combat Occupations Similar to civilian labor market outcomes, military outcomes may also be determined by demand-side factors. The military screens potential incumbents for entry to the institution and to positions. The armed forces exclude applicants on the basis of both ascribed and achieved status. They have partially or wholly excluded people on the basis of race, sex, sexuality, attainment, and ability. Homosexuals are still not officially allowed to serve openly in the military . Up until 1948, blacks played a limited role in the armed forces. During World War II, they served in segregated units and were excluded from combat . Today, women still are not technically allowed to serve in combat positions . Because little previous theory has examined the impact of demand-side factors on the military, the following two sections describe the expected effects of military screening at the levels of the institution and of the occupation. --- Institutional Screening For most of the twentieth century, the military theoretically rejected people at the bottom of the aptitude distribution and restricted the numbers they would accept from the bottom of the educational distribution. Since World War I, the armed forces have administered tests to determine which recruits to accept. After World War II, the tests were combined, standardized across the services, and normalized on the men serving at the end of that war. The resulting entry test is referred to as the Armed Forces Qualifying Test . In 1980, the armed forces re-normed the AFQT on the National Longitudinal Study of Youth 1979 , a nationally representative sample of the military age population . Thus, the AFQT percentile scores represent percentiles in the population, not just among those taking the test. The test has been included in the National Longitudinal Surveys as a measure of ability for all respondents . The armed forces use the scores from this test, along with educational attainment, to determine which applicants to accept. Since 1951, they have theoretically excluded people who score in the bottom 9 percent on the test and have restricted the number of recruits with relatively low test scores, those between the 10th and 34th percentiles, who can enter the services . The armed forces use the scores on this test along with educational attainment to determine which recruits to accept. They accepted, for example, men with high scores even if the men had dropped out of high school . These military policies suggest the following prediction: Hypothesis 4: Men should be less likely to be admitted to the military and thereby to serve in combat occupations if they are in the bottom of the academic aptitude and educational attainment distributions. --- Occupational Screening While institutional screening should lead the people who serve in combat occupations to be less likely than civilians in their cohort to have cognitive aptitude at the bottom of the distribution, occupational screening may produce the opposite result. Due to occupational screening, the service-members serving in combat occupations should be more likely to have cognitive aptitude near the bottom of the distribution than other service-members and, possibly, than civilians. The armed forces assign positions on the basis of military needs and on the basis of the recruits' characteristics, experiences, skills, abilities, and aspirations. At any given time, the armed forces have jobs that they have a greater or lesser need to fill. In filling these positions, they consider the types of jobs that the recruits say they want to perform in the short and long-term. They exclude women from positions that involve combat or that are in units that primarily engage in combat . New service-members are excluded from jobs that are only available to those with military experience . They are allocated to positions that are open to them because they obtained skills or qualifications before they entered the military, through education or training. They are also assigned to positions based on the scores they receive on the Armed Services Vocational Assessment Battery , which consists of ten specialized tests that are designed to measure skills and achievements. One of these tests, for example, assesses recruits' familiarity and experience with mechanics . The armed forces also classify occupations according to the cognitive skills needed, including medium-skill jobs, such as those involving mechanics, support, and administration; and high-skill jobs, such as those involving communications and intelligence. The bulk of combat occupations, those in the infantry, fall into a category labeled as requiring lower cognitive skill, which also includes occupations that involve service and supply handling . Thus, service-members with higher ASVAB test scores should have greater access to technical, non-combat occupations. By extension, those with lower scores should be more likely to be assigned to combat occupations, which leads to the following prediction: Hypothesis 5: Among those in the military, men should be less likely to serve in combat occupations the higher they are in the academic aptitude and educational attainment distributions. --- DATA AND METHODS --- Data To evaluate the roles of class bias and supply and demand factors in determining who serves in combat occupations, we use data from the NLSY, a nationally representative survey of 12,686 men and women who were between the ages of 14 and 22 in 1979. These data are ideal for the current purpose because they include an over-sample of people serving in the armed forces as of September 30, 1978. The NLSY dropped many of the respondents from the military over-sample after the 1984 wave of the survey. Therefore, we focus on information reported by the respondents in the 1979-1984 waves. As mentioned above, we analyze data provided by the male respondents only. Of the 6,403 men in the NLSY sample, data on all the variables used in our analyses exist for 5,569 . We use weights for survey waves 1979-1984 and correct for clustering within families. Part of the following analysis presents findings regarding how socioeconomic, racial, cognitive, and educational characteristics relate to the odds that men would enlist in the volunteer military. The NLSY has been used for this purpose before . The remainder of the analysis extends these findings to look at the associations between these characteristics and the odds that service-members would serve in combat positions. The data have not previously been used for this purpose. Indeed, only one article has examined selection into combat positions , and none have examined such selection during the volunteer military. One article does look at the racial makeup of combat positions during the Iraq war, but it does not consider the association between serving in such positions and any characteristics other than race . --- Outcomes The following analyses simultaneously evaluate two related outcomes. The first outcome is whether or not the respondent had served in the military by 1984. The men are categorized as having served in the military if they were part of the military over-sample or if they answered "yes" to a series of questions about military service asked every year. A total of 1,647 male respondents had served in the military by 1984, of which 1,244 provided answers to all of the analysis questions. The second outcome is whether or not the respondent served in a combat occupation. The men are classified as having served in a combat occupation if they reported serving in an MOS that fell within the "Infantry, Gun Crews, and Seamanship Specialists" category established by the 1977 Department of Defense 3-Digit Enlisted Occupational Classification System . This category includes people who work in infantry, combat engineering, artillery, and other occupational specialties. A total of 365 men had served in a combat MOS by 1984, of whom 316 had complete data. Unfortunately, the NLSY does not include a measure of whether or not these men who served in a combat occupation also experienced combat. Ideally, research examining the effects of supply and demand should assess all of the decision points leading to a particular outcome. The following analyses, for example, would include all of the people who applied to serve in the military but were rejected. In practice, researchers often look at outcomes among a population and infer the preceding choices that led to those outcomes . The current analyses follow this procedure. --- Predictors Birth Year-The NLSY respondents were born between 1957 and 1964. They first became eligible to serve in the armed forces between 1975 and 1982, the early years of the AVF. Between 1975 and 1980, the number of men entering the military declined. In 1981 and 1982, that number increased again, but only slightly . Accordingly, the analyses include a linear measure of birth year. Race/Ethnicity-As mentioned above, research regarding class bias has focused not just on class but also on race . Therefore, we evaluate the possibility that people had different probabilities of serving in the armed forces and in combat occupations if they came from different racial and ethnic backgrounds. We construct two dummy variables to measure race/ethnicity: one if the respondent reported being black and the other if the respondent reported being non-black and Hispanic. Family Background-We use two family background measures. The first measure reflects whether or not the respondent grew up in an intact family. We consider those respondents who indicated that they lived with their biological father and mother when they were 14 years old to have been raised in an intact family. Respondents who provided a different answer are coded with a 0. Intact families tend to have greater financial resources than do other types of families . The second family background measure is the average years of schooling completed by the respondents' parents. People with higher educational attainment are more likely than those with lower attainment to earn more and work in jobs with higher status . Scholars examining selection into military service and wartime mortality have commonly used parental education to assess class background . If the respondent only provided information on the educational level of one parent, we use this as the measure of parental education. In some households, the NLSY surveyed more than one respondent. If information for one respondent was missing, we substitute the parental education information from a sibling who reported information on parental education and also reported the same family structure. Neither of these two measures of family background captures class position directly. In order to test whether the results are robust if education and family structure are used rather than a more direct measure of class, models were estimated on two different sub-samples: one for which a measure of father's occupation was present and the other for which, if father's occupation was not present, mother's occupation was present. In each of these subsamples, models were estimated with two different sets of variables, all the measures of family background and the measure of parents' education, or the measure of father's or parents' occupation . In all cases, the pattern of results was the same, which suggests that the measure of parents' education is a relatively robust indicator of family background. AFQT Score-In 1980, 94 percent of the NLSY sample took the ASVAB test. Results from this test were used to construct an unofficial AFQT score. Researchers have previously used the scores in the NLSY to norm the AFQT scores and as a measure of intelligence for all the survey respondents . In the current analyses, the scores are used to measure academic aptitude. They are calculated as percentiles ranging from 0 to 100. For the current analyses, the percentiles are converted to the categories used by the military: 1 = 91-100; 2 = 65-90; 3 = 31-64; 4 = 10-30; 5 = 0-9 . Because there are so few respondents in Categories 1 and 5, we combine the respondents in each of these categories with those in Categories 2 and 4, respectively. An alternative would be to treat the scores on the test as linear and to test for non-linearities in the effects of AFQT score by including a quadratic measure. The analyses tested the robustness of the effects in this way, and the results of this robustness check are mentioned when the results are described. Educational Attainment-The measure of education is based on the respondents' reports of the highest grade of schooling that they completed. Among service-members, the measure is based on the respondents' answers to a question asking them to report the highest grade they had completed before they entered the military. If the answer to this question is missing, the measure is based on the answers to three questions: the highest grade completed as reported each year, the date on which that education was completed, and the date the respondents entered the military. If the answer is still missing, it is based on the highest grade the respondent reported. The linear measure of educational attainment is then transformed into a categorical measure with the following categories: 0-11 = less than high school; 12 = high school; 13-15 = some college; > 15 = college graduate. Among the national population of students enrolled in post-secondary education, the proportion of students who enroll in post-secondary education at neither 2-year nor 4-year institutions is less than 2 percent . This article, therefore, refers to the broader category of post-secondary education as college. Higher education is potentially endogenous to military service. Therefore, we report models with and without educational attainment. We are accordingly cautious in our causal claims about the relationship between educational attainment and military service. --- Models The model used is a sequential logistic model, based on one that is popular among scholars who study educational attainment . It has also been called a continuation ratio or "Mare" model . It expands on the traditional logistic regression model to take account of the sub-samples that are "at risk" of making particular transitions. For example, with regard to education, only the people who have finished high school are at risk of entering college, and only those who have entered college are at risk of attaining a college degree . While the model estimates conditional logistic regressions simultaneously, these regressions could also be fitted individually . Thus, the effects of the background variables can differ across the outcomes. This model has not previously been used to study military outcomes. However, the same logic of nested transitions applies to the process by which people are assigned to combat occupations. Only people who have entered the military are at risk of being assigned to combat occupations. In the language of the model, people who did not serve in the military were at risk of entering the military, but not of being assigned to combat occupations. People who served in the military in non-combat occupations were at risk both of entering the military and of serving in combat occupations. They accomplished the first transition but not the second. Similarly, people who served in the military in combat occupations were at risk both of entering the military and of serving in combat occupations. They completed both transitions. Figure 2 shows a schematic of the model, which is based on the processes outlined in Figure 1. The virtue of the model in the current case is that people can only serve in combat positions if they have enlisted in the military. However, net of this assumption, the factors that affect the decisions regarding entering the military and serving in combat positions are assumed to be independent. The general model is shown in equation 1: where X represents a matrix of explanatory variables, λ' represents a vector of estimated parameters linking the explanatory variables to the outcome, and α represents the constant for the outcome. Pass ki-1 indicates whether or not a person has passed the previous transition. The following analyses include two transitions , where 1 represents the transition into military and 2 represents the transition into a combat position. This model is estimated in Stata using the sequential logit option developed by . --- FINDINGS Table 1 presents the characteristics of different groups within the sample. The left panel compares the characteristics of service-men to civilians. The right panel compares the characteristics of service-men who served in non-combat occupations to those of servicemen who served in combat occupations. As shown in the left panel, the men who enlisted in the military were slightly older than men who did not; they were born, on average, a third of a year earlier. They were also more likely to be black and less likely to have grown up with both biological parents. They came from families in which the parents had, on average, a third year less of schooling. On average, they themselves had one year less of schooling than did civilians. Despite the fact that the military theoretically excluded men with the lowest entry scores, 5 percent of service-men had scores in the bottom category of the AFQT. Similarly, despite the fact that the military theoretically restricted the number of high school dropouts they would accept, service-men were as likely as civilians to have dropped out of high school. These findings suggest that service may be related both to family background and to supply-side factors. As shown in the right panel, service-men were more likely to serve in combat occupations than in non-combat occupations if they grew up in intact than in non-intact families. Because previous researchers have shown that intact families provide children with more privileges than do non-intact families , this finding provides evidence that is consistent with the view that disadvantaged men were more likely than advantaged men to serve in combat occupations. Combat service-men also were more likely to have low and less likely to have high cognitive test scores than non-combat servicemen. They had fewer years of education and were more likely to have dropped out of high school. These findings suggest that men served in combat occupations due to both supply and demand factors. Table 2 reports coefficients from sequential logistic regression models predicting military service and combat occupation. Family characteristics are more strongly associated with the earlier transition, that into military service, than they are with the later transition, that into a combat occupation. The differences between these coefficients are statistically significant according to joint and separate chi-square tests. The first model shows that the odds of entering the military were associated with the following measures of family background: race, family structure, and parental education. Black men were more likely to enter the military than were white men. Men were less likely to enter the military if they grew up in intact families than in non-intact families. They were also less likely to enlist if their parents had college degrees than if their parents had less education. By contrast, the odds of completing the later transition were associated only with family structure. Among those in the military, service-men were more likely to serve in combat occupations if they grew up in non-intact families than if they grew up in intact families. These findings suggest that serving in a combat occupation was associated with family disadvantage. Net of family background, the odds of entering the military and of serving in a combat position are associated with AFQT scores. However, the test scores are associated with the two outcomes in different ways. As above, these differences are statistically significant. According to the model, men were less likely to enlist in the military if they had high AFQT scores than if they had median test scores. Among service-men, they were also less likely to serve in combat occupations if they had high test scores. In addition, men were less likely to enlist in the military if they had low scores than if they had median scores. However, once in the military, they were equally likely to be assigned to combat occupations. These findings suggest that, net of family background, combat service may have stemmed from the intersection of supply and demand factors. Figure 3 presents predicted probabilities derived from the preceding model, Model 1. The predicted probabilities are derived by multiplying the probability of making the transition by the percent of the population at risk of making the transition. All men are at risk of entering the military, for example, while only those in the military are at risk of being assigned to combat occupations. For the purposes of the figure, disadvantaged men are defined as those who did not grow up with both biological parents and whose parents had, on average, just a high school diploma. Advantaged men are defined as those who did grow up with both biological parents and whose parents had, on average, a college degree. The dashed line at the top of the figure represents the predicted probabilities that disadvantaged men who were black would enter the military. The solid line below represents the predicted probabilities that advantaged men who were white would enter the military. The remaining two lines represent the probabilities, respectively, that disadvantaged black and that advantaged white men would serve in combat occupations when compared not just to those in the military but to the population. The figure demonstrates the importance of family background. Black men who came from disadvantaged families were nearly four times as likely as white men from advantaged families to enter the military and therefore to serve in combat occupations. Among the men with test scores in the middle of the distribution, the predicted probability that a black man from a disadvantaged family would serve in a combat occupation was equal to the probability that a white man from an advantaged family would serve in the military in any capacity. Among those with the lowest test scores, white men from advantaged families were slightly less likely to serve in the military than black men from disadvantaged families were to serve in combat occupations. These findings are consistent with the family disadvantage account. Model 2 in Table 2 presents evidence regarding the importance of human capital for selection into military service and combat occupations. This model incorporates the categorical measure of educational attainment, which reverses the apparent association between cognitive test scores and military service at the top of the distribution. Holding education constant, men with higher scores were no less likely to enter the military than those with median scores. This finding suggests that the association between high test scores and military service was mediated by human capital. Men with high test scores were probably less likely to enter the military than men with lower scores because they were more likely to get a college degree. According to Model 1, men were less likely to enter the military if they had higher than if they had lower AFQT scores. According to Model 2, however, they were more likely to do so at each level of schooling. These findings provide further evidence consistent with demand-side screening. All else equal, the armed forces would rather enlist recruits with higher than those with lower cognitive test scores. The marginal association between educational attainment and military service in this model is quite large, suggesting that college may have kept men from enlisting in the military and therefore from obtaining combat positions. Indeed, the model shows that, net of AFQT and family background, college graduates were much less likely to enter the military and serve in combat occupations than were high school graduates. In contrast, a college degree had little relationship with occupational assignment among men in the military. This result suggests that service-men were assigned to occupations primarily on the basis of AFQT scores rather than on the basis of educational credentials. The model including educational attainment provides some insight into the mechanisms underlying the association between military service and both race and family structure. The effect of coming from an intact family becomes smaller, though still negative, when education is included in the model. This finding suggests that the association between military service and family structure may be partially mediated by educational attainment. Men from intact families were more likely than those from non-intact families to attain higher education and correspondingly less likely to enter the military. The race association, by contrast, becomes larger. This finding suggests that, net of AFQT, at all levels of education, blacks were more likely than whites to enter the military, which is consistent with previous research suggesting that whites negatively select but that blacks positively select into the military . It suggests that higher ability working-class black men may have different outcomes from both higher ability whites and from other workingclass blacks. The analyses next evaluate whether and how the associations between the military outcomes and the independent variables differ by race in three steps. In the first step, four sets of models were separately estimated by race. In the second step, models were estimated using data from the full sample and including interactions between race and the independent variables, except for birth year, which was not significantly different when models were estimated separately by race, in order to test for racial differences formally. These models suggest that most of the independent variables, except for family structure, interact with race based on comparing the models' Wald statistics. In the third step, the coefficients from the models with significant differences in the Wald statistics were examined more closely. In the models with the interactions between race and parents' education or AFQT scores, the estimates of the main effects are similar to the model without interactions. The estimates of the interaction terms are, for the most part, insignificant. Based on these results, the interaction that is most informative suggests that blacks and whites experience different associations between military outcomes and educational attainment. The findings suggest that the military was more likely to exclude high school dropouts if they were black than if they were white. They also suggest that having more education provided black service-men, though not white service-men, with the option to avoid serving in combat positions. Figure 4 is derived from the model with the interaction between race and education. The figure presents predicted probabilities that are derived for the same groups and represented as described above for Figure 3. In this figure, panel A contains predicted probabilities by AFQT score and family background for high school graduates, while panel B contains those predicted probabilities for men who attained more education than a high school degree. The contrast between panel A and panel B demonstrates the importance of human capital. More educated men had extremely low predicted probabilities both of serving in the armed forces and of being assigned to combat occupations, suggesting that attending more than high school gives men options that make military service and therefore combat service unlikely. Among high school graduates, family background continues to be strongly associated with military service and combat occupations. White men from advantaged families were much less likely to enter the military than were black men from disadvantaged families. The association between family background and combat MOS is somewhat smaller. Disadvantaged black men were more likely than advantaged white men to be assigned to combat occupations regardless of their AFQT scores. This association predominately stems from family structure, as neither race nor parental education significantly affects combat assignment among service-men. Net of measured cognitive achievement and human capital, more privileged men may have unmeasured cultural capital that enables them to navigate the military bureaucracy. --- CONCLUSION The preceding analyses show that men come to serve in combat positions as a result of supply and demand factors but also at least partly as a result of family background. They confirm the importance of demand factors that have led the volunteer era armed forces to select people with higher ability. They also confirm the importance of supply factors, which suggest that, among service-members, people with less education and cognitive ability are more likely to be assigned to combat positions. Net of these supply and demand factors, however, men were still less likely to enlist and were less likely to serve in combat occupations as a consequence of both race and class. Only one previous article has looked at how both personal characteristics and race affect selection into combat positions. The current analysis differs from that previous article in two respects. First, the previous article used data provided only by veterans , whereas the current article uses data from a sample of the entire population. Thus, the current article provides insight into the cumulative nature of the processes by which the members of particular groups end up first serving in the armed forces and then serving in combat occupations. To draw on the analogy of educational attainment, the strategy of the first article is akin to examining the factors that determine who graduates from college among the people who enter college. The strategy of the current article is akin to evaluating the preceding question but also evaluating the factors that determine who enters college among the broader population. Second, the previous research focused on the Vietnam war era , while the current analysis has focused on the years shortly after that war, the early years of the all-volunteer period. In the previous research, the veterans were more likely than the veterans examined in the current analyses to experience combat because they served during wartime. They also served during the draft era, which means, in theory, that they were less likely to differ systematically from the general population. Indeed, during the peacetime draft era, men who served in the armed forces grew up in families that had similar characteristics to those of the men who did not . Despite the different samples and eras, the current analysis reaches similar conclusions to the previous article regarding the effects on combat assignment of race and cognitive test scores. In wartime and peacetime, and volunteer and draft militaries, service-men were assigned to combat occupations on the basis of their AFQT scores . Most likely service-men with higher cognitive test scores were offered higher skill, non-combat occupations, leaving those with lower scores to serve in lower skill, combat occupations. To be sure, some combat occupations require skill. However, in general, the armed forces consider higher skill occupations to be ones that do not involve combat . The previous article and the current research both also show that, once in the military, race did not affect combat assignment . Among service-men, blacks were as likely to be assigned to combat occupations as were whites. The findings from the current article, however, differ from the findings of the previous article in two respects. First, they suggest that serving in a combat occupation was associated with family background. During the peacetime volunteer period, the men who entered the military and thus were at risk of serving in combat occupations differed from those who did not in terms of their family background, as measured by race, family structure, and parental education. The previous article did not evaluate the relationship between combat service and family characteristics other than race . Second, the analyses show that part of the impact of test scores on serving in a combat occupation was mediated by educational attainment. Men with higher test scores were more likely to go on to college, less likely to enter the military, and therefore less likely to serve in combat occupations. Educational selection into the military may also have been important during the draft era, when men could avoid the draft by attending college . Indeed, one article shows that Vietnam veterans were less likely to have a college degree than the general population . The previous article could not explore this relationship because the data did not include respondents who did not enlist in the armed forces . The current research reaches similar conclusions regarding the importance of family background in combat occupations as does other research examining combat exposure. Previous research has suggested that the men who fought and died during the Korean and Vietnam wars came from families and neighborhoods with fewer advantages . In contrast to this previous research, the current analyses also include measures of personal characteristics that could mediate the association between military outcomes and family background. These characteristics, particularly cognitive test scores, strongly affected who enlisted in the military and who served in combat occupations. They did not, however, explain the impact of family background characteristics completely. More generally, the findings contribute to the literature regarding social stratification, particularly with respect to transitions made early in the adult life course. They show that military transitions partially resemble educational transitions. Similar to the educational transitions research , the current analyses shed light on the relationship between equality of opportunity and of result. They simultaneously assess the impact of social background on the odds of accomplishing multiple outcomes. The educational transitions research focuses on the distribution of social rewards. It is concerned with the characteristics of the people who attain particular levels of education . The current research on military transitions focuses on the distribution of social risks. It is concerned with the characteristics of the people who may fight in wartime. In both cases, the analyses show that socioeconomic background affects earlier transitions more strongly than it does later transitions. The unequal results, whether educational or military, stem from unequal opportunities to make or avoid earlier transitions. Scholars have explained the pattern of a diminishing association between social background and educational attainment by developing two competing theoretical accounts. First, some scholars have argued that this pattern reflects maximally maintained inequality. This theory suggests, in part, that social background effects are strongest at the earliest transitions that are not universal . Based on the similar pattern with respect to military outcomes, the theory may apply not just to rewards but also to risks. The data used in the current article are derived from a period in which entering the military was far from universal. Thus, the relative strength of the association between family background and military service is consistent with the theory. Second, other scholars have suggested that the diminishing relationship between social background and educational transitions is consistent with an explanation derived from the life course perspective . This theoretical perspective, in part, emphasizes the importance of timing and of linked lives . Consistent with this view, people make earlier transitions, such as enrolling in college, when they are economically and socially closer to their families. They make later transitions, such as graduating from college, when they are farther from their families. The findings regarding military transitions are also consistent with this view. The current analyses cannot adjudicate between these two theories because they do not include enough historical variation. This theoretical contrast represents a promising avenue for future research. In addition, the preceding analyses shed light on the relative importance of supply and demand for determining occupational segregation. Some previous research on the segregation of civilian occupations has suggested that such segregation stems primarily from actions taken at the organizational level . The preceding analyses lead to a similar conclusion. They show that the segregation of military occupations stems primarily from the fact that particular types of people, defined by race, family structure, and parental education, are more likely than others to serve in the military. They also show that both supply and demand factors affect who goes into the military, but only demand factors affect who is assigned to particular occupations. Future research should examine the extent to which service-men in combat positions are more likely than service-men and women in non-combat positions to see combat. This research could evaluate the risks that not just service-men but also service-women face, which would have increased relevance today because women are more likely to serve in the contemporary military and to be exposed to combat than they were in the past. In the late 1970s and early 1980s, when the NLSY was first fielded, women constituted 10 percent of the armed forces. Today, they constitute 15 percent . In 1994, the Department of Defense opened some previously closed military occupations and units to women, while still technically excluding women from combat positions and from combat units . As a consequence, service-women appear to face more of the dangers of war today than they have in the past . Future research would ideally examine data that include measures of both military occupations and combat exposure among both men and women. The preceding analyses also potentially shed light on the question of unequal risk in the wars in Iraq and Afghanistan because they include not just service-members on active duty but also those in the reserves. In the current wars, the troops are composed of a mixture of service-members who volunteered for active duty and service-members who volunteered to serve in the reserves and National Guard. Over the last 8 years, the armed forces have deployed to war zones more than 1.7 million service-members . They have also placed on active duty more than 700,000 service-members who were in the reserves and National Guard . Approximately 24 percent of the service-members in the NLSY had not served on active duty by 1984. The current analyses therefore evaluate military service and combat occupations among a mixture of service-men who were and were not on active duty and may shed light on the selection processes in the current wars in which many service-members in the reserves have been deployed. The analyses suggest that people differ systematically from each other in the odds that they will fight and die during wartime. They show that people serve in combat positions as a consequence of the ways that supply and demand preferences intersect. On the demand side, the armed forces want recruits with higher test scores and are more likely to put those people into non-combat positions. On the supply side, people with higher test scores have potentially more desirable options than enlisting in the military. Yet net of these competing supply and demand preferences, men with fewer pre-service advantages, as captured by race, parental education, and family structure, are still more likely than those with more advantages to serve in the military and, therefore, in combat occupations.

This study evaluates the characteristics of the men who served in the volunteer military in combat occupations. It examines whether these characteristics stem from supply-side or demand-side decisions, or reflect class bias. The findings suggest that, on the supply side, men who had greater academic abilities were more likely to go to college, thereby avoiding military service and the possibility of serving in a combat occupation. On the demand side, the armed forces were more likely to exclude men with lower academic abilities but were more likely to assign such men in the military to combat occupations. Net of the impacts of these supply-side and demand-side decisions, men who served in combat occupations still differed from those who did not in terms of their family background. The impact of family background was stronger on entering the military than on being assigned to combat occupations once in the military.Much research has examined how the people who served in the U.S. military differed from those who did not (e.g., Bachman, Segal, Freedman-Doan, and O'Malley 2000) and how the men who fought and died in U.S. wars differed from those who did not (e.g., Allen, Herrmann, and Giles 1994;Wilson 1995). However, little is known about the process by which people come to be at the greatest risk of fighting and dying, the process by which they come to serve in combat occupations. Only one previous article has examined this process. It focused on the Vietnam war era and showed that servicemen with low entry test scores were more likely to serve in combat occupations than those with high test scores (Gimbel and Booth 1996). This article evaluates the characteristics of the men who were assigned to combat occupations in the late 1970s and the early 1980s, the early years of the All-Volunteer Force (AVF). Even during peacetime, service-members who perform combat occupations suffer the greatest risks relative to other service-members and receive the fewest rewards. If a war occurs, they are the ones most likely to see combat (Gimbel and Booth 1996). In addition, previous research suggests that they learn skills that are less transferable than those learned by service-members in non-combat occupations and have lower earnings in the civilian labor market (Mangum and Ball 1987).

INTRODUCTION A major challenge in Computational Social Science [6,12,15] consists in modelling and explaining the temporal dynamics of human communications. Which interactions lead to more successful communication or productive meetings? How can we infer temporal models of interactions? How can we explain what these temporal interaction really mean? Current statistical analysis techniques do not explore the full temporal aspect of time-series data generated by interactive systems, and certainly they do not address complex queries involving temporal dependencies. We investigate Markov rewards models for human-human interactions in social group meetings and how to interpret them. We identify various queries predicating over the temporal interactions between different roles, the impact of different sentiments in interactions or in decision making, causality between particular states, etc. We use probabilistic computational tree logic with rewards [4,11], which is a type of probabilistic temporal logic variant, to formalize these queries. We then use the PRISM tool [11], a symbolic probabilistic model checker, to analyse the formal queries and thus interpret the temporal interaction models. Probabilistic model checking [4] is a well-established verification technique that explores all possible states of a Markov model in a systematic and exhaustive manner, and computes the probability that a temporal property of the system under analysis holds. We can ask queries such as 'What is the average count of the project manager's interventions until a decision is taken?', 'What is the probability of a decision to be taken without anybody commenting about their understanding?', or 'What is the average interaction count from one decision to another decision without a negative sentiment being expressed in the interim?'. Figure 1 illustrates the method we propose for probabilistic modeling analysis of social group behaviour. The main contribution of this paper consists in empirically demonstrating the expressiveness of probabilistic temporal logic properties and probabilistic model checking for the analysis of temporal dynamics of social group interactions in meetings. --- RELATED WORK Our work is most closely related to the Markov Rewards Model by Murray [13,14] for analyzing and querying social sequences. In that work, social interactions are represented as a sequence of states, and particular states are associated with rewards or costs that are dependant on the query being asked. A Value Iteration algorithm is then used to estimate the expected value of every state, with a state's value indicating how it is related to the outcome of interest being queried. In our work, we will use the same state representation as Murray, but show that our probabilistic model checking framework allows us to ask queries that would be difficult or impossible to ask in the Markov Rewards Model framework. More generally, our approach is an example of social sequence analysis [7], where the goal is to analyze patterns in social sequences or to compare social sequences to one another. These social sequences might unfold at the macro scale or at the micro scale , and the present work is concerned with social sequences at the micro scale. The past decade has seen an increasing amount of work on developing technologies for supporting meetings, including the use of machine learning for making predictions on meeting data. This includes detection of decision items [10] and classification of dialogue act types [8], in addition to predictions for many other meeting phenomena [16]. The field of Social Signal Processing consists of work that examines social interaction through primarily nonverbal cues [18], such as gesture, gaze, and prosody. There is also a growing inter-disciplinary field of meeting science that aims to understand the processes that take place before, during, and after meetings [1]. --- CORPUS The dataset used in this paper is the Augmented Multimodal Interaction meeting corpus [5]. Each meeting group in the corpus consists of four people, and the group completes a sequence of Figure 1: Overall process of modeling and analysis of group interactions four meetings where they are role-playing as members of a company that is designing and marketing a product. Each person in the group is assigned a role; the roles are Project Manager , Marketing Expert , User Interface Designer , and Industrial Designer . Despite the artificial scenario and the assigned roles, the speech is spontaneous and unscripted, and each group is free to make decisions as they see fit. We discuss further aspects of the corpus in Section 4.1, where we describe the state representation used in this work. --- PROBABILISTIC TEMPORAL MODELLING AND ANALYSIS OF INTERACTION In this section we describe the state representation used in our Markov models, the probabilistic temporal logic properties and reward structures used for formalising queries about group interactions captured by the Markov model, and the probabilistic model checker PRISM used for formally analysing these queries. --- Markov models of social group interactions In our representation of social sequences in meeting, each state is labelled by a 5-tuple consisting of the following information: the participant's role in the group: PM , ME , UI , and ID ; the dialogue act type taking one of the 15 values listed and briefly described in Table 1; the sentiment being expressed: nosentiment, positive, negative, posneg ; whether the utterance involves a decision: nodecision, decision; whether the utterance involves an action item: noaction, yesaction. In addition to the complex states described in the preceding section, there are START and STOP labeled states representing the beginning and the end of a meeting. Example states include the following: • <PM-bck-positive-nodecision-noaction> describes the situation where the project manager makes a positive backchannel comment, unrelated to a decision or action; • <PM-el.ass-nosentiment-nodecision-yesaction> represents the project manager eliciting feedback about an action item; The Markov aspect of the Markov models is that the probability of a given state depends only on the preceding state in the sequence. The state transition probabilities are estimated directly from the transition counts in the data. This way we obtain a discrete-time Markov model of the behaviour seen in the meeting data, where the states labels and the transition probability function are defined as above, the initial state is labelled by START. A path in a Markov model is a non-empty sequence of states such that the transition probability from one state to the next one in the sequence is strictly greater than zero. --- Probabilistic temporal logic and model checking Probabilistic model checking is a technique for modelling and analysing stochastic systems, usually focused on investigating correctness properties of the real-life system. It requires an abstract, high-level description of the system and specifications of the properties expressed in a suitable temporal logic. In the first step a probabilistic model checker tool builds a model of the system from its description, typically a Markov model . In the second step, the tool uses model checking algorithms to verify automatically if a temporal logic property is satisfied or not, or to compute the probability of a temporal logic formula to hold. These model checking algorithms explore the model in an systematic and exhaustive way. Probabilistic Computation Tree Logic [4,11] is a probabilistic branching-time temporal logic that allows one to express a probability measure of the satisfaction of a temporal property by a state of a discrete-time Markov model. The syntax is the following: State formulae: Φ ::= true | a | ¬ Φ | Φ ∧ Φ | P ▷◁ p [Ψ] | S ▷◁ p [Φ] Path formulae: Ψ ::= X Φ | Φ U ≤N Φ where a represents an atomic proposition, ▷◁ ∈ {≤, <, ≥, >}, p ∈ [0, 1], and N ∈ N ∪ {∞}. For a path π starting from a state s, we define the satisfaction relation π |= Ψ as follows: • π |= X Φ is true if and only if Φ is satisfied in the next state following s in the path π ; • π |= Φ 1 U ≤N Φ 2 is true if and only if Φ 2 is satisfied within N time steps and Φ 1 is true up until that point where Φ 2 is satisfied for the first time. The syntax above includes only a minimal set of operators; the propositional operators false, disjunction ∨ and implication =⇒ can be derived. Two common derived path operators are: the eventually operator F where F ≤n Φ ≡ true U ≤n Φ and the always operator G where G Ψ ≡ ¬. If N = ∞, i.e., the until operator U is not bounded, then the superscript is omitted. For example, how do we check whether the probability of reaching a yesaction within 50 utterances while the sentiment being expressed is not a positive one is greater than 0.75? The corresponding PCTL property represented as P ≥0.75 [¬ "positive" U ≤50 "yesaction"]. The model checking algorithm computes the reachability probability for all states satisfying the atomic proposition "yesaction" provided that all previous states visited do not satisfy the atomic proposition "positive"; if the resulting probability is greater than 0.75 then the model checking problem returns true; otherwise it returns false. PRISM is a probabilistic model checker [11] used for formal modelling and analysis of systems that exhibit random or probabilistic behaviour. Its high-level state-based modelling language supports a variety of probabilistic models, including discrete-time Markov chains. In PRISM we can replace the bounds ▷◁ p in the properties with =? and thus obtain the numerical value that makes the property true. PRISM also allows models to be augmented with reward structures, which assign positive real values to states and/or transitions for the purpose of reasoning over expected or average values of these rewards. In PRISM we can specify the following reward-based temporal properties: • R r wd=? C ≤N in a state s computes the expected value of the reward named rwd accumulated along all paths starting from s within N time-steps. • R r wd=? [ F Φ ] in a state s computes the expected value of the reward named rwd accumulated along all paths starting from s until the state formula Φ is satisfied. In PRISM, filters check for properties that hold when starting from sets of states satisfying given propositions. In this paper we use the filter operators state and avg in the following two types of properties: • filter evaluates the satisfaction of the state formula Φ in the state uniquely identified by the Boolean proposition cond1; • filter computes the average over all states where cond2 is true. In the following, for convenience, we refer to PCTL properties with or without rewards simply as properties or queries, though strictly they also include PRISM operators. --- EXPERIMENTS AND RESULTS In this section, we first define the behavioural model used, followed by a set of queries, their encoding as probabilistic temporal logic properties, and their results, which demonstrate the flexibility and expressiveness of the method presented in this paper. --- Defining a behavioural model of social group interactions The behavioural model is a Markov rewards model initially inferred as described in Section 4.1 to which we add labels and reward structure definitions as required by the queries. In our case the atomic propositions associated with each state are the state labels and the individual particles composing the state label. The PRISM model encoding the Markov model for the input data set considered for this paper as well as the PRISM properties analysed later in this paper are available at http://www.dcs.gla. ac.uk/~oandrei/resources/imsgi_gift18. The PRISM model has a relatively small state space of 196 reachable states and 4002 transitions, therefore the model checking process for one temporal property is not time-consuming . We defined the following reward structures: • r_Steps assigns a value of 1 to each transition or time-step. We use this reward structure when computing the average number of time-steps from one state to another state. --- Querying the Markov model We use the command line of the PRISM tool to execute each of the queries presented in this section through the probabilistic model checking engine and export the results. For some of the PRISM properties below we make the following notation for the sake of brevity. We use the placeholder roleLabel to be instantiated with any of the roles PM, ME, UI, or ID. The atomic proposition y=j refers to the state variable y in the PRISM model with the identifier j; in this case j takes values from 0 to 268. --- 5.2.1 Queries for validating the model. We first start with examples of queries and results that confirm our expectations about meetings generally and the AMI scenario specifically. For example, some of the results reflect the fact that project managers tend to begin meetings, and -in the AMI scenario, at least -are the most active participants. Some of the results of this first set of queries are merely artifacts of the AMI scenario, and in particular of the fact that participants are assigned clearly-defined roles and have to progress through distinct phases of a role-playing exercise. We then move on to queries and results that generate more insight into meeting interactions. Q1: How long does it typically take in a meeting before each type of role has participated? These queries are encoded in PRISM as: R{"r_Steps"}=?[F "PM"] R{"r_Steps"}=?[F "ME"] R{"r_Steps"}=?[F "ID"] R{"r_Steps"}=?[F "UI"] Each of the PRISM queries above computes the average accumulated number of time steps it takes to reach a state corresponding to a particular role. The actual average number of steps is computed using the transition reward r_Steps. The analysis results are 2.13 time steps for PM, 5.26 for ME, 5.99 for ID, and 6.03 for UI. This is an intuitive result, showing that the project manager tends to begin the meeting discussions, but also that all members participate early on in the discussion. Q2: How long does it typically take in a meeting before each type of non-PM role has participated after a Project Manager? The PRISM properties encoding of Q2 are: filter filter filter Then the PRISM property encoding Q3 is: χ 3 / + χ 3 + χ 3 + χ 3 ) Checking this property instantiated with each of the four roles, we obtain that PM participates 32%, ME 24%, while UI and ID are participating in equal measure at 22%. This results reflect the fact that project managers tend to be more dominant in the meeting discussions, and in particularly in regards to decision-making. Q4: How many times in average a PM is involved in decision-making within 100 time steps? Let χ 4 denote the PRISM property that computes the average visit counts to states where roleLabel made a decision within 100 time steps: --- R{"r_roleLabel_decision"}=?[C<=100] Then the PRISM property encoding Q4 is: χ 4 / + χ 4 + χ 4 + χ 4 ) After checking the four instances of this property, we obtain the following results: 86% for PM, 9% for UI, 3% for ID, and 1% for ME. As expected, project managers are making the majority of decisions, and the differences between the other three roles are likely an artifact of the AMI scenario. Q5: Which type of non-PM roles is more participatory following a PM within 100 time steps? The PRISM property encoding this query averages over all PM states the visit counts to roleLabel within 100 time steps: filter and the results of model checking it are: 36% for ME, 33% for ID, and 32% for UI. This shows that the non-PM roles are approximately equally likely to participate after the PM, with the ME being slightly more frequent. Again, this may be an artifact of the AMI scenario. Q6: Which roles with positive sentiment have the highest probability in the long-run? The PRISM property encoding this query looks at the probability in the long-run to be in a particular type of role with a positive sentiment: and the results are as follows: 34% for PM, 32% for ME, 18% for ID, 16% for UI. These results largely reflect the fact that the PM tends to be most active person in the AMI meeting discussions. --- 5.2.2 Queries for further exploration of interactions. Many of the preceding sets of queries and results conform to our expectations about meeting behaviour and the AMI scenario. We now turn to a set of queries and results that generate more valuable insight into meeting interactions. Q7: Which non-decision states are most valuable in contributing to decisions being made within 100 time-steps? The PRISM property encoding this query computes the probability of reaching a decision state within 100 time-steps when starting from a specific non-decision state: filter&"nodecision") The top ten most valuable non-decision states are the following: The most noticeable trend is that states containing sentimentboth positive and negative -are highly associated with decisionmaking. A second trend is that non-decision states belonging to the PM are highly associated with decisions being made. Both of these findings are intuitive; for example, participants tend to express a variety of opinions before mutually deciding on a solution or course of action. --- State Q8: Which PM states tend to lead to more participation by non-PM participants within 50 time-steps? The corresponding PRISM property for the ME role sets a reward of 1 for each visit of a ME state and hence computes the average visit counts to ME states within 50 time-steps when starting from a specific PM state. These results tell us that the PM is particularly likely to get participation from other members when he or she explicitly seeks input and when expressing sentiment. Q9: Which non-sentiment states are highly associated with positive sentiment? The PRISM property encoding this query looks at each state tuple with no sentiment being expressed and then computes the probability of the next state to include a positive sentiment: filter & "nosentiment") The top ten non-sentiment states most likely to be associated with positive sentiment in the next state are the following: These results show that states containing dialogue acts that are explicitly eliciting information are likely to be followed by expressions of positive sentiment. In particular, the top state represents the PM explicitly seeking an assessment from one or more of the other group members, and this is very likely to be followed by a positive sentiment state. Q10: Which non-sentiment states are highly associated with negative sentiment? Similar to Q9, the PRISM property encoding Q10 is: filter & "nosentiment") The top ten non-sentiment states most highly associated with negative sentiment in the next state are the following: Interestingly, states that explicitly elicit information and belong to somebody other than the PM are associated with negative sentiment. This result coupled with the previous result suggest that participants may be eager to please the PM through expressions of positive sentiment and agreement, and more willing to express negative sentiment to non-PM participants. --- State Q11: Which non-decision states that occur early in meetings tend to cause decisions to be made quickly? The PRISM property encoding this query is: P=?[F<=50 & "nodecision" & P>=1[X "decision"])] where we considered early meetings to be within 50 time steps. This property computes the probability of eventually to reach a nodecision state identified by j within 50 time steps and in the neXt state a decision is taken . Interestingly, none of these states involve sentiment, and they belong to a variety of the roles. However, the top two results both belong to the PM. This reveals that sentiment and decision-making are less associated with each other early on in the meetings. Q12: If one person expresses positive sentiment, does it lead to other people expressing positive sentiment? We compare the average probability of expressing one type of sentiment after another or the same type of sentiment using the following PRISM properties and their results: --- 0.04 For example, the last property above computes for each positive state s the probability of reaching a negative state when starting from s, and then returns the average over all positive states s. These results show that an expression of positive sentiment is very likely to be followed by another expression of positive sentiment, and similarly with negative sentiment following negative sentiment. It is less common for negative to follow positive and viceverse, which is partly reflecting the fact that negative sentiment is much less common in this corpus. --- Q13: If a PM person expresses positive sentiments, what is the probability that it leads to positive sentiment expressed by a non-PM person? This query is a form of causality relation between positive sentiments expressed by a PM person and a non-PM person. We formalise query Q13 as a probabilistic constrained response [9] where we instantiate roleLabel by ME, UI, or ID: P>=1 [G => P>=p [ & !) U<=N ])] This PRISM property states the following: whenever PM expresses positive sentiment then, with probability greater than p, roleLabel and PM do not express negative sentiment until roleLabel expresses a positive sentiment within N time steps. This property helps us identify the maximum probability p for which the answer to the query is true when instantiating the roleLabel for non-PM roles. For N = 100, then the maximum probabilities p for which the answers to Q13 are true are 0.1 for ME, 0.06 for ID, and 0.05 for UI respectively. For N = 500, then the maximum probabilities p for which the answers to Q13 are true are 0.4 for ME, 0.25 for ID, and 0.25 for UI. We conclude that ME is approximately twice as likely than ID and UI to respond positively to a PM positive sentiment. This result is likely to reflect the structure of the AMI scenario. It tells us that the ME has a great deal of responsibility and can perhaps be seen as a secondary leader of the meeting. --- CONCLUSION In this paper we demonstrated the expressiveness of probabilistic temporal logic properties for formalising various probabilistic and reward-based queries about group interactions in meetings and then analysed them with the probabilistic model checker PRISM and interpreted them for the AMI corpus. Some of the queries analysed above do not need probabilistic temporal logic properties to be asked on the initial data set. However, all queries involving bounded time steps and in particular the steady-state properties, e.g. Q11 and Q13, cannot be expressed in any other way than as temporal property formulae. The queries Q1 -Q6 validate our behavioral model as their results confirm expected interactions, while the queries Q7 -Q13 highlight novel insight into the AMI dataset we analysed. In this paper we analysed the Markov model inferred from state transitions counts in the data. For future work we will consider admixture models inferred from the data using classical Expectation-Maximisation algorithms where each component in the admixture model models a particular pattern of behaviour, similar to the work of [2,3]. The challenge will be in identifying suitable classes of probabilistic temporal properties for characterising and discriminating between the patterns for the particular type of interaction data contained AIM corpus. In future work, we will experiment with alternative state representations, particularly representations that are less specific to the AMI corpus scenario and its roles. For example, we will include demographic characteristics such as gender and the native language of the speaker. We will also apply this representation and methodology to other group interaction datasets such as the ELEA corpus [17].

There may be differences between this version and the published version. You are advised to consult the publisher's version if you wish to cite from it.

INTRODUCTION The importance of user groups and patient associations in research and policy-making related to the provision and quality of healthcare is growing. 1 Nonetheless, in perinatal audits, the inclusion of patients and families is rare and the limited experiences reported in the literature are recent. [2][3][4][5] Including bereaved women in perinatal death audits could make it possible to obtain more complete information, especially about their antenatal care and socioeconomic circumstances. However, recruiting women who have just lost a child into a research protocol requires careful thought because of their vulnerability after this experience. In the disadvantaged French district of Seine-Saint-Denis, an audit was commissioned by the regional health agency in 2013 because of high rates of infant and perinatal mortality. At that time, infant mortality was 4.4 per 1000 live births compared with 3.2 in France. 6 A Delphi consultation with district --- Strengths and limitations of this study ► A multifaceted audit to investigate high rates of perinatal mortality in a socially disadvantaged district included data abstraction from medical records, interviews with bereaved women and peer-reviews by an expert panel. ► A referral protocol was put into place for bereaved women identified as having unmet needs identified by the midwife-investigators. ► A support group for the three midwife-investigators was established and they contributed written narratives of their experiences 6 months after the study's end. ► Women who refused an interview did not differ from women who accepted, but there was wide variation in acceptance rates by maternity unit. ► This study sheds light on the benefits and the difficulties of interviewing bereaved women for perinatal audits and stresses the need for designing studies with support for women and investigators. Open access clinicians and parents' organisation to sound out their views on causes of the high mortality rates before the audit's onset highlighted difficulties linked to access to care and social disadvantage. 7 8 Because this information is not always available in medical records, interviews with bereaved women were integrated into the study design. The interviews were conducted by three experienced midwife-investigators. A protocol was developed with the district perinatal health network to refer women with unmet needs identified during the interview. In addition, a support group run by a clinical psychologist was set up for the midwife-investigators. The aim was to support them for their frequent contact with bereaved women and their families as well as with hospital staff. The purpose of this article is to describe and discuss the contribution of interviews with bereaved women to the audit's research aims, the benefits for the women themselves, especially related to use of the referral protocol and the midwife-investigators' experiences with the support group. --- MATERIAL AND METHODS --- Study design The audit included all stillbirths and neonatal deaths at or after 22 weeks of gestation in the 11 maternity units of the Seine-Saint-Denis district of births occurring in in from 1 January to 31 December 2014. All cases were included unless the family explicitly refused. The research team did not directly approach potential participants, rather the medical team acted as a liaison between the women and the research team. During the maternity hospitalisation or after the death in the neonatal unit, the medical team provided information about the study and asked women if they agreed to be contacted for an interview. Women's care did not vary as a result of their decision to participate. Women who agreed were contacted by one of the three midwife-investigators 4-6 weeks later to arrange an interview in the location of their choice. Professional interpreters with previous experience in qualitative research were available, if necessary. The midwife-investigators were recruited by advertising the positions on an employment website for epidemiological research and through the French midwives' professional society. In interviews with potential candidates, we sought to confirm sufficient clinical experience and maturity to manage complex medical or psychosocial situations that could arise during the maternal interviews or in interactions with medical staff. Personal considerations were also discussed, including the candidates' own experiences with perinatal death. The three midwife investigators who were recruited for the study had at least 15 years working in maternity care and none had personal experience with perinatal bereavement. Their role was to collect socioeconomic and medical data from medical records and the results of follow-up tests. They used a standard protocol which did not include any identifying information . They also conducted interviews with the women. After further anonymisation , each case was assessed by at least one expert from a panel comprising 14 senior clinicians . The purpose was to identify suboptimal factors and to determine whether the death might have been prevented by their modification. More details on the audit's methodology audit have been published elsewhere. 9 Questionnaire for bereaved women A multidisciplinary group with experience in bereavement care , sociologist, psychologist) developed the questionnaire in order to minimise any negative impact on the women of being interviewed about the death of their baby. The goals of this questionnaire , designed to take about 40 min, was to provide information not systematically available in the medical records about demographic and social characteristics, living conditions and care during the pregnancy. The questionnaire included questions about women's understanding of the medical follow-up and their evaluation of staff support during delivery. The questionnaire was study-specific and did not include preexisting psychometric scales. Extra time at the end of the interview allowed women to express other concerns. The midwives also noted information about the conditions of each interview: the location, presence of an interpreter, the woman's apparent understanding of the questions, her attitude towards the interview and use of the referral protocol. The midwife-investigators received a half-day training in qualitative methods covering the different steps for conducting and analysing interviews and the importance of verbatim transcription of answers to the open-ended questions. During the interviews, the midwife-investigators made notes about the women's responses and then transcribed them at the end of the interview. We chose not to record the interviews for fear that it would increase the refusal rate, and the women did not receive a copy of the transcripts. When the midwives encountered situations in which they considered that the woman required medical or other services, they referred in accordance with the referral protocol. This protocol was developed together with hospital teams and parents' associations in the district and identified the professionals to whom women should be referred for different types of care or services according to their place of residence. --- Midwife-investigators' descriptions of their experience of the support group A half-day support group led by a clinical psychologist who was also a midwife was set up every 2 months to enable the investigators to consolidate their knowledge of perinatal bereavement care, have a space in which to share their experiences and be supported as appropriate. --- Open access They could also schedule telephone consultations with the psychologist if needed. Six months after the end of the study, each midwife-investigator was asked to provide a written document describing her experiences during the study and in the support group. The aim of this document was to provide feedback to the coordination team about this component of the study protocol; there was no predefined format for this document. --- Analysis Data from the medical records, responses to closed questions from the maternal interviews and information about suboptimal factors from the external assessments were entered into a database and analysed quantitatively. For comparisons of categorical variables, we used χ 2 tests or Fisher's exact tests when cell sizes were below 5; means were compared with t tests. Free-text responses from the maternal interview and the midwives' written descriptions were extracted by one investigator using N'Vivo 10 for the analyses. We identified comments that were related to women's experiences of care which were used to illustrate how the interviews contribute to understanding the care process and its failures. The writings of the midwives were analysed thematically, according to the method of grounded theory. 10 First, we coded the texts line by line with N'vivo 10, using an inductive approach , followed by thematic grouping . 11 These analyses were then shared with JZ, in order to obtain an agreement on the themes. Dissentions did not occur in the coding of this corpus. Finally, we consulted with the midwife-investigators as well as the psychologist who had led the group, who reviewed the results; they had no further information to add. --- Patient and public involvement Two parents' organisations were involved in the Delphi process prior to the audit and in the elaboration of the referral protocol. --- RESULTS --- Maternal interviews Description of the women who agreed to be interviewed Over the study period, there were 172 stillbirths and 77 neonatal deaths among 25 037 births that took place in the district maternity units. Because 22 families refused to participate in the audit, the study included 218 women with 227 infants . Of them, 101 agreed to be contacted for an interview. However, there were 19 secondary refusals, and seven more could not be reached. Interviews were therefore conducted with 75 women . Inclusion rates by hospital ranged from 6% to 60%. Two centres had an inclusion rate below 15%, four between 20% and <40%, four between 40% and 50% and one included 60% of women. The 75 women were similar in age, parity, country of birth and nationality to those who refused, but were less likely to have no health insurance coverage , as shown in table 1. Obstetric history, preexisting medical risk factors and pregnancy outcomes were also similar between women who were and were not interviewed. --- Conducting the interviews and use of the referral protocol The interviews lasted an average of 90 min and were conducted between March 2014 and March 2015. Most women chose the option of meeting at home, although 11 interviews took place in a healthcare setting and two women chose a public park . Professional interpreters were used for five interviews. Most women responded to the questionnaire without difficulty, but questions had to be reformulated or explained for some women who did not need translators but who did not speak French fluently. At the end of the interview, 85% of the women reported that they were satisfied and pleased to have shared their experience or to have participated in this research: 'It's gratifying to have someone come to us; it shows that there is someone who is asking questions' . Eleven women were more neutral, and one was unhappy because of a misunderstanding; she had expected to get further information about her child's autopsy. The referral protocol was used for 35% of women. As shown in table 2, most referrals were to the district's parental bereavement services, followed by maternal and child health services, parental organisations and the maternity unit. --- Information provided by the maternal interviews The interviews made it possible to complete information in the medical files about care during pregnancy and in particular difficulties accessing care. Of the women interviewed, 22% described difficulties making appointments and 24% getting to them, because of lack of child care, work responsibilities and financial constraints. Fifteen percent of women reported going without care for financial reasons. These difficulties impacted on women's care pathways: 'I had been told in the previous pregnancy that because of the problem of the detached placenta, as I had a cesarean and my baby had been hospitalized for 3 weeks, that for a future pregnancy I had to be followed at the hospital. I was also told about the need to take [aspirin. But when I called the maternity unit to make an appointment, I was told it could be done only from the 6 th month' . Later in the interview, she regretted that her general practitioner had not insisted that the hospital agree to provide her prenatal care earlier. For women requiring specialised care, the time and financial burdens could be high: 'For the specialized ultrasound examinations, it took me about 3 hours to get there and back. And appointments were frequent. They are very expensive in addition, I spent almost €600. Fortunately, my Open access insurance company reimbursed a part of the costs' . Once they entered the healthcare system, however, most of the women reported very good or good treatment by professionals, emphasising the kindness of the medical and paramedical staff. Nonetheless, 10.7% of women were not at all satisfied. Their comments illustrated a gap between the care expected and received. 'Then I had my appointment with the doctor at the prenatal diagnosis center. She did not examine me because, she said, 'You've been to the emergency room too many times, I don't want to be intrusive'. The relationship went wrong. I was expecting someone to tell me what to do on a daily basis: to move, not to move, to be on strict rest… because I had heard so many different opinions. The tone and manner in which she spoke was neither right nor reassuring' . --- Usefulness of the interviews for expert review of cases A panel of medical experts reviewed suboptimal factors and potentially preventable deaths. The experts were given access to the transcripts of the maternal interviews, when available. As part of the assessment, they were asked whether the interviews had allowed them to identify additional suboptimal factors and contributed to their assessments . In 32 of the 78 deaths with maternal interviews, the experts responded Open access yes to this question. Reasons provided included a better understanding of the woman's living conditions and social circumstances: 'No family support may have contributed to the absence of pregnancy follow-up' or to the woman's isolation: 'The interview with the woman highlighted her psychological distress and loneliness , which did not appear in medical observations because it was not spotted' . In some cases, the questionnaires provided medical information that was not included in the file: 'The interview makes it possible to know that her blood pressure was elevated at the last visit' . The experts in their reports also pointed to problems in care coordination between different practitioners or different facilities, for example: 'The information in the interview demonstrates step by step the major lack of coordination of care for this patient: Due to her complicated diabetes, pregnancy might have been contraindicated in her case. A preconceptional consultation should have been very strongly considered, with optimization of her glycemic control if she chose to become pregnant. The patient should probably have been managed from the start by a multidisciplinary team of specialists at a high level maternity unit with an adult ICU, given her risks. The follow-up here started gradually with the usual gynecologist and no clear consultation with diabetologists or nephrologists. The patient's providers were her general practitioner and a private practice nurse, which shows the low level of medical supervision she received' . The perceived value of having the woman's perspectives is underscored by an expert's comment in one case without an interview: 'It seems to me that the interview could have provided significant additional information, concerning in particular the monitoring of this pregnancy'. --- The midwife-investigators' support group, led by a psychologist The support group was an important part of the three midwife-investigators' decision to participate in the study. 'This proposal tempted me from the beginning because it converges with a question that is essential to me, both personally and professionally: how to support people dealing with death in a society that hides it?' . The analysis of their experiences, described 6 months after the study, revealed three themes about the benefits of the group related to the emotional weight of their encounters with bereaved women, interactions with the healthcare teams and personal well-being. The first theme was the need to discuss their experiences managing the grief expressed by the women during the interviews. Although perinatal mourning is inherent to the practice of midwifery, death is uncommon in a conventional practice: 'For us, even more than for providers who treat sick people, death is a failure, even if we are regularly confronted with life-threatening events such as potential preterm birth or miscarriages… Some interviews were very hard. Even if you feel armed against the pain of the patients, their story, the duration of these stories, the emotional burden that goes with them, are hard to handle' . 'Another very positive aspect of these sessions was being able to hear the others [midwife-investigators] talk about their difficulties. At these moments, we became aware of how hard what we were doing really is and the fact that we must not minimize the weight it might have' . A second theme centred on experiences interacting with medical teams. The study's protocol required the midwives to obtain medical records and other documents Open access from these teams, whose members did not necessarily want to remember or think about the perinatal death which they experienced as a failure. 'It is enough to meet a maternity team facing such an event to confront our own feelings of failure and our personal suffering in the face of death' . Completing the files led the midwives to analyse the cases from a retrospective point of view and they could note the dysfunction and failures of the health system: 'One of the major difficulties for me was containing my emotions in the face of what I could identify as practice mistakes. What do you do when you find information that, for you, is synonymous with a lifethreatening situation and was not interpreted as such, leading ultimately to the death of a fetus?' . Finally, a third theme was their personal appreciation of the value of the exchanges within the group. By giving them support in difficult situations, they reported feeling that their importance and worth was being recognised: 'This time for reflection seemed to me to be very respectful of us as professionals… The fact that we could think of ourselves and the consequences of what we were going to experience was very gratifying and empowering' . Being able to express and share the difficulties of the project during the group support sessions enabled the midwives to accept and make sense of them: 'The sessions allowed me to express anger at what I could identify as a failure, violating either the principle of neutrality or the necessary confidentiality […] It allowed me to distance myself and to hold on over time, but also gave me time for introspection' . Only one of the three midwives used an individual phone session with the psychologist. But all three stated that it had reassured them to know that they could use this type of support between the group sessions. --- DISCUSSION As part of an audit of perinatal deaths that took place in 2014 in a disadvantaged district in the Parisian region, the research team sought to include the bereaved mothers at the data collection stage. Seventy-five questionnaires in an interview format were conducted with women with characteristics similar to all women in the district who lost a child to stillbirth or neonatal death. These interviews provided important information about the women's prenatal care and were valuable to the experts assessing the cases. Our analysis of this experience revealed the importance of support protocols for both the bereaved mothers and the investigators. More than a third of the women were found to be in need of health or social services, according to the referral protocol. The experience of the midwifeinvestigators participating in the support group that met every 2 months illustrates the challenges for research teams supporting perinatal bereavement, medical teams in distress and their own emotional experiences. The strengths of this study are its populationbased design and its socioeconomic context, focused on disadvantaged women who are at higher risk of perinatal death, but are less likely to participate in research studies. 12 Although there have been a few initiatives to include bereaved women in studies of perinatal death, 4 13 this approach is still very uncommon. 2 14 15 It is thus important to illustrate its benefits and challenges in different contexts in order to develop this research further. A further strength of this study is its parallel focus on the midwife-investigators; although support groups for hospital staff have been reported, 16 17 this is an original approach for a research team. 18 Our analyses were based on the midwives' written assessments 6 months after the study was over, giving them time to gain perspective on their experiences; however, other designs, such as keeping a journal during the study, might have captured other, more immediate, reactions. Among the study limitations, we did not have information on why women refused an interview. Differences in the inclusion rates by hospital, alongside similar maternal characteristics among responders and non-responders, suggests that the medical teams' approach to presenting the study played a role. 19 However, only one-third of women were interviewed and we cannot exclude the existence of other biases. Although one concern was that the interviews would cause women to relive the experience of their child's recent death, study design features were developed to minimise these risks including holding the interview after several weeks, having experienced midwifeinvestigators and designing the questionnaire with multidisciplinary specialists. One of the reasons for conducting research on study designs that include bereaved women is precisely to provide evidence of their importance and feasibility for medical personnel who are hesitant about upsetting or placing an additional burden on women soon after they have experienced a stillbirth or neonatal death. Our results showing that interviews improved the quality of the audit, were useful for helping women still struggling to deal with their loss, and were well accepted by women are reassuring in this regard. Further studies on women's reasons for refusal as well as their experiences participating in research could help improve protocols; these could include more participatory designs, such as getting input from women on the transcripts of their interviews. Finally, having research from different countries is important for assessing the external validity of these findings. Audits are a key tool for surveillance and evaluation of in utero and infant deaths as shown by those conducted routinely in the UK, 20 Australia and New Zealand 21 and the Netherlands, 22 but bereaved mothers do not participate in providing information for these audit procedures. Audits have not been widely used in France, although there are a few regional examples [23][24][25][26] which have not included bereaved parents. Our study showed that speaking to women provided important information on their social circumstances, obstetric history and difficulties in access to care that were not available in medical records and that affected assessments of the appropriateness of care. This constitutes a strong rationale for Open access including interviews with mothers in perinatal audits to ensure that preventable health system factors are identified. An English study is testing an innovative method of including bereaved parents in clinical mortality reviews in two hospitals 4 and this could also be a way to ensure that information on the woman's circumstances and care experiences are readily available in the files for external audit. Our study shows that it is possible to interview mothers after a perinatal death, including women in disadvantaged areas and who are not fluent in national languages. The provisions to adapt to women's needs in the protocol, such as allowing for interviews in places other than the woman's home and having professional interpreters who could accompany the study midwives, were necessary for including some women and used in one out of five interviews. These interviews 4-6 weeks after the death also showed insufficient support after discharge, as found elsewhere. 13 Despite the presence of a specialised service for perinatal bereavement care in the district, 27 not all of the maternity units referred women for these services. High use of the study's referral protocol underscored the importance of having experienced midwives conduct interviews in order to identify and appropriately refer women with health or other needs. 28 The difficult experience of professionals dealing simultaneously with their response to perinatal death and with mourning parents is recognised as particularly acute. 29 30 As a recent literature review shows better training and support is essential 31 for teams facing perinatal loss. 32 Small groups and the opportunity to contact a psychologist individually appear to enhance support for research teams substantially. This issue has been raised in the literature for teams conducting research with qualitative methods 33 34 but not for those conducting clinical audits. The write-ups by the three midwife-investigators showed the extent to which daily contact with bereaved women and witnessing shortcomings in care could bring up strong feelings related to their professional ethics and personal beliefs. The psychotherapist and the group provided a contained space for sharing these experiences. This approach would be valuable for investigators or medical teams working on audits in other areas, for example, maternal mortality or childhood cancer, who face similar challenges. --- CONCLUSION We found that interviewing bereaved women as part of a perinatal audit is feasible, provides valuable information and can have benefits for the women themselves. Women's experiences navigating the healthcare system or with difficult social circumstances are important for accurately identifying suboptimal care and designing quality improvement and prevention campaigns. Interviewing bereaved women also provides information about postdischarge care. However, our study shows that research including bereaved parents requires specific design features. It was imperative to have adequate arrangements for referral to healthcare providers and parental associations which requires having investigators who are experienced healthcare professionals. Another key feature was providing support to the study investigators for their daily contact with maternal grieving and the distress of healthcare professionals. Setting up a psychologist-led support group made the investigators feel recognised and respected, and improved their ability to provide support to the women. In conclusion, our results provide a strong justification for including interviews with bereaved women in perinatal audits, but only if sufficient attention is given to the needs of bereaved women and investigators. --- Data availability statement Data are available upon reasonable request. Ethical restrictions regarding participant privacy prohibit the authors from making the entire data set publicly available. The corresponding author Priscille Sauvegrain may be contacted in order to request the data. --- Ethics approval This study obtained the authorisation of the Advisory Committee on health research information on 29 October 2013. The collection of the medical data and the constitution of databases were authorised by the CNIL on 8 January 2014, which was followed by an amendment on the modalities for obtaining consent. The Ethics Committee of Inserm/IRB00003888 issued a favourable opinion n 13-39 for the study as a whole and more particularly for the maternal interviews on 11 December 2013. This study was approved by the French Advisory Committee on Health Data in Medical Research and the French National Commission for Data Protection and Liberties. Provenance and peer review Not commissioned; externally peer reviewed.

Objectives To describe experiences including interviews with bereaved women in a clinical audit. Design The data come from an audit of all stillbirths and neonatal deaths at ≥22 weeks of gestation in Seine-Saint-Denis, a disadvantaged French district in 2014. We included bereaved women using a questionnaire that also contained open-ended questions administered in an interview format by a midwife-investigator several weeks after the death. The study included a referral protocol for bereaved women with unmet needs revealed during the interviews. A psychological support for the three midwifeinvestigators was set-up, in the form of a support group. Setting The 11 maternity hospitals in the district. Participants 218 women (227 deaths). Analyses Data come from medical records, maternal interviews, the reviews of the audit's expert panel and written narratives of their experiences provided by the midwife-investigators. Quantitative data were analysed statistically, and qualitative data thematically. Results One-third (75) of the women agreed to an interview, but acceptance ranged from 6% to 60% by maternity unit. Characteristics of respondents and nonrespondents were similar. Members of the audit's expert panel reported that 41% of the interviews contained new information relevant to their assessment. Of the women interviewed, 35% were referred to a medical professional, psychologist or social worker. Midwife-investigators' experiences illustrated the benefits of a support group with three main themes identified: improving their interactions with bereaved women as well as medical teams and protecting their psychological well-being. Conclusion These results showed that including interviews with bereaved women in audit designs was feasible and provided valuable information on women's care and social circumstances that were not available in medical records. They also highlight the importance of implementing referral protocols for the bereaved women, used in over one-third of cases, as well as providing support for study investigators.

Introduction --- I n this article, we discuss the extent to which sociological theories produced within the Brazilian academic field can be correlated with a broader intellectual movement that has been gaining attention in recent decades, which critiques coloniality and the Eurocentric foundations of the social sciences. However, it is important to note that the effort to consider these possible homologies and dialogues presents us with some initial difficulties. The first challenge is related to dealing with two theoretical approaches that are complex and heterogeneous, as is the case with Brazilian sociology and postcolonial perspectives, in their diverse forms and developments. On the one hand, no single tradition or specific theoretical approach within Brazilian sociology has focused on questioning the Eurocentric foundations of sociology. On the other, postcolonialism itself is an "umbrella" term, combining multiple academic and extra-academic approaches. Therefore, in this analysis, we consider both approaches from a panoramic perspective, which includes critical perspectives within Brazilian sociological theory, particularly concerning Eurocentrism, modernity, and colonialism, as well as the various facets of postcolonial approaches in contemporary scholarship. The second challenge is to consider both perspectives in their specificities -contextual, theoretical, methodological, and epistemological -to avoid homogenizing readings that label any critical interpretation of a social problem as postcolonial. Our aim is to avoid anachronisms or vague adjectives that offer little analytical insight into the substantial contributions of the mentioned authors. In the article, we face these challenges by discussing the relevance of Brazilian sociologist Alberto Guerreiro Ramos ), understood as an author with postcolonial inspiration. We look at his contributions in criticizing academic Eurocentrism and "intellectual dependence" in order to investigate his proposition of an epistemological approach that sought to deal with the challenge of the appropriation by Brazilian sociology of concepts, theories, methods, and techniques developed in other contexts. In the first part of the text, we address the underlying challenges in attempting to define these two approaches. In the second, we examine the convergences and divergences observed between Brazilian sociological theory and Postcolonial thought, which provide a basis for reflecting on possible research agendas that combine these approaches. In the final section, we explore the epistemological potential of one of these agendas, which involves rereading Brazilian sociological theory in light of the tensions provoked by Postcolonial thought in contemporary social theory at a global level. For that, we analyze the heuristic potential of the Guerreiro Ramos' reduction sociology, which is conceived as a critical-assimilative method of foreign social theories whose relevance may suggest a reciprocal inquiry between Brazilian sociology and postcolonial thought -a decentered approach to our sociological tradition that also highlights significant contributions to the future of postcolonial epistemologies. --- Preliminary considerations Sociological theory distinguishes itself, on the one hand, due to a lack of consensus in terms of its ontological, epistemological, and methodological dimensions 1 ; on the other hand, due to the existence of a plurality of schools, perspectives, approaches, and paradigms . This tendency was no different in the institutionalization of the social sciences in Brazil, which, since the 1940s, has been characterized by intense debates and by the internal constitution and competition between specific fields of study. Even when considering the internal constitution of such fields, it is worth noting that they are not based on univocal interpretations or any monolithic theoretical or political entity. This is what Lívio Sansone argues regarding racial studies in Brazil, a field that, since the 1930s, has been "full of tensions, agendas, points of view, perspectives, and desires". This context becomes more complex when one observes the disputes, which arise from divergent explanatory principles of class and race for the structure and dynamics of inequality in Brazil . The debate acquires new dimensions with the perspectives that accentuate the structuring and overlapping characteristics of these categories, in addition to gender . These discussions encapsulate the challenge of approaching Brazilian sociology more comprehensively, particularly in terms of rereading the criticism of coloniality and Eurocentrism. As suggested, such critiques are not confined to a single theoretical strand. On the contrary, they can be observed in authors who advocated epistemically divergent proposals for Brazilian sociology, such as Guerreiro Ramos and Florestan Fernandes, who engaged in an emblematic debate on how to understand sociological practice in the 1950s . Therefore, our effort to systematize the theoretical commonalities 1. The distinction between social theory and sociological theory is sometimes considered subtle. By social theory, we understand the abstract, metatheoretical manner of research in the social sciences , since, when searching for more general models for analysis, social theory can schematize, conceptualizes, and explains the forms and dynamics in the interactions, classifications, hierarchies, reproductions, and social changes. In its turn, sociological theory is "less general and more concrete, which does not mean that it is less abstract, but offers not so much a reflection regarding society as such, but rather about any given society or […] any determined set of societies" . Thus, we agree with Vandenberghe that sociological theory tends to be more disciplined and historically informed compared to social theory. Additionally, the analytic and metatheoretical propositions of social theory become important presuppositions for sociological theory, which in turn informs empirical research in sociology. and differences between Brazilian sociology and postcolonial theories generally considers the most critical strands of the country's sociological tradition regarding the persistence of colonialism and Eurocentrism in the construction of national sociological knowledge. The postcolonial tradition is equally thorny. There are at least two conceptions of the term. The first refers to a historical-political condition that points to the "later" processes of decolonization in "Third World" countries between 1950 and 1970 . The second refers to the range of theoretical, methodological, and epistemological perspectives that emerged in critical historiography and cultural studies starting in the 1970s, with a strong impact in India, the United States, and England, aiming to produce knowledge that is unaffected by Eurocentrism, even if at times invoking European currents such as poststructuralism, deconstruction, and Marxism . Sérgio Costa argues that, despite the theoretical dissimilarities of authors as distinct as Ftantz Fanon, Edward Said, Homi Bhabha and Gayatri Spivak,, who are seminal for social theory in the second half of the 20 th century, they share a common intellectual and political program. All of them seek to break with the tyranny of a single history sustained by Western metanarratives; overcome ideologies of singular modernization and progress; develop a critique of colonial domination and the ethnocentric "civilizational" processes; and deconstruct the essentialisms based on class, race, ethnicity, nation, and gender. In Latin America, the postcolonial perspective can be observed in discussions regarding decoloniality, especially within the Latin-American Subaltern Studies Group and, subsequently, in the Coloniality/Modernity Group. The creation of the latter was guided by a concern to break with the Eurocentric canon and achieve truly decolonized forms of knowledge and social organization . However, Luciana Ballestrin and Joaze Bernardino- Costa point out that one of the issues with decolonial approaches in Latin America has been the limited engagement with Brazil. Significantly, no Brazilians have been associated with these groups, indicating that the intellectual output of Brazilian social sciences remains excluded from what Santiago Castro-Gómez and Ramón Grosfoguel called the "decolonial turn". Examining the convergences and divergences between Brazilian sociology and postcolonial approaches constitutes a first step in our effort to analyze connections among these perspectives. In this endeavor, we will consider the term postcolo-nialism as broadly as possible, encompassing the various tendencies and variations cited, regardless of their specificities and dissimilarities. As Pedro Borba and Guilherme Benzaquen assert, "a constructive dialogue does not depend on rigid demarcations, but rather on uncovering pertinent connections". --- Connections between --- Brazilian sociology and postcolonial theory Analyzing the conditions that led to the emergence of a critical discourse on coloniality and the Eurocentric foundations of the social sciences, which gained notoriety in many peripheral countries of capitalism since at least the 1950s, provides a productive path for discussing convergences between Brazilian sociology and postcolonial approaches. We do not intend to reconstruct the entire history of this process of theoretical decentering but rather to identify some of the social, economic, cultural, and institutional conditions that allowed the emergence of this debate in peripheral contexts between the 1950s and 1970s. Decolonization struggles in Africa and Asia were responsible for the political and intellectual reorientation observed in countries on the periphery of capitalism. Furthermore, as noted by Ballestrin and Meneses , the Bandung Conference in 1955, the Non-Aligned Movement in 1961, and the Tricontinental Conference in Cuba in 1966 helped to shape a geopolitical identity for the so-called Third World countries, which sought to achieve a better position in the global context beyond the polarization imposed by the Cold War. In Brazil, there was a change in foreign policy in the 1960s with the election of Jânio Quadros and, shortly after, João Goulart. They began to adopt a more independent position towards North countries and affirmed an anticolonial and antiracist stance in solidarity with other Third World countries. During his brief time in power, Quadros opened embassies in Africa and granted scholarships to African students with funding from Itamaraty . Mário Augusto Medeiros da Silva highlights the growing connection between Brazilian activists, intellectuals, and anticolonial activists from Lusophone Africa during that period. He cites the foundation of the Afro-Brazilian Movement for the Liberation of Angola in 1962, which was linked to the Pro-Liberation of Angola Movement and established partnerships with black cultural activism in São Paulo and Rio de Janeiro, as well as with white and Black Brazilian intellectuals. During this period, Brazil underwent an intense process of industrialization, creating an atmosphere of pursuing change that was evident in both the intellec tual and political spheres. There was also a growing desire for a more independent geopolitical position on the global stage. These factors led to a critical approach to the Eurocentric foundations of the social sciences and the creation of intellectual projects aimed at developing an autonomous sociology engaged with national concerns and the country's future direction. These conditions allowed for theoretical and methodological convergences between Brazilian intellectuals and those from other peripheral contexts. --- The status of sociology in peripheral contexts We could say that the first point of convergence between Brazilian sociology and postcolonial theories concerns a theoretical preoccupation with the status of sociology in peripheral contexts. This discussion is crucial to the postcolonial debate and is woven into the national sociological field in various ways. In the history of Brazilian sociology, we see an intense debate about applying social theories and normative ideals produced in the Global North to the Brazilian reality while simultaneously critiquing the Eurocentric foundations of the social sciences. The influence of Euro-North American sociology on the formation of Brazilian sociology is undeniable. As in other peripheral contexts, the Brazilian sociological canon was fundamentally constituted by European and American models. Despite this, there has been a constant concern with the peripheral status of Brazilian sociology at different moments in the discipline's history. It is worth mentioning the debate between Guerreiro Ramos and Florestan Fernandes in the 1950s regarding the possible directions of Brazilian sociology and society during a time of institutionalization of the social sciences in Brazil . The debate began during the 2nd Latin-American Congress of Sociology, held in Rio de Janeiro andSão Paulo, in 1953, and, according to Bariani , revealed the divergences in understanding sociological research, its methods, and the conditions and development of research in Brazil, demonstrating different projects for the discipline and the nation. In the proposals presented at the Congress -and later in the works Cartilha brasileira do aprendiz de sociólogo, published in 1954, and A redução sociológica, published in 1958 -, Guerreiro Ramos criticized the transplantation of measures adopted in "developed" countries to solve domestic problems. He believed that solutions should be sought in the actual conditions of Brazil's national and regional structures. Concerning ideal research conditions, the sociologist from Bahia be-lieved they should be adapted to the "underdeveloped" nature of Brazilian society, prioritizing the formulation of general interpretations of the social structure rather than studies of the "minutiae of social life". These general interpretations should be able to guide the implementation of central planning policies aimed at national development through industrialization . By endowing sociological Thought with a salvational orientation, Guerreiro Ramos believed that teaching sociology, even as a school discipline, could best promote "the diffusion of a critical consciousness of national problems and the promotion of emancipation concerning cultural colonialism" . Florestan Fernandes, in turn, disagrees with Guerreiro Ramos in his analysis of the scientific working pattern that should be adopted by Brazilian sociology. In a text originally published in 1958, The scientific working pattern of Brazilian sociologists, Fernandes emphasizes that sociology should not be guided by the nation's system of interests and values but rather by the system of norms and values of scientific knowledge; otherwise, it would not be possible to put science at the service of the community. The author sees the defense of the autonomy of the scientific field as a way to counter external influences that weigh on the sociologist at various levels, especially in a context marked by the persistence of archaic and authoritarian relationships. Therefore, he advocates for methodological rigor in the conduct of sociological research -which should not be subordinate to the material conditions of underdevelopment in the nation -and the universal character of scientific knowledge, which can be absorbed from major academic centers. The incorporation of community studies -rejected by Guerreiro Ramos -would allow Fernandes to apprehend the variations in internal, economic, and sociocultural development that characterize the country's various regions. In other words, science itself should be explored as a factor in development . Although it is beyond the scope of this work to reconstruct the multiple meanings and subsequent interpretations taken on by this debate, it is worth noting the centrality of the concerns of both authors regarding the future of sociology in Brazil at the moment of its institutionalization. Such concerns were focused on reflecting on the nature of the relationships to be established with the centers of scientific production and the limits and possibilities of Sociological Thought in promoting the country's development. These critical approaches converge with a decentralized transnational intellectual movement that, since the 1950s, has challenged the Eurocentric status of social sciences practiced in North Atlantic countries. According to Marcelo Maia , this reflection contributed to two analytical procedures observed in our academic field and other peripheral countries. The first refers to "conceptual criticism", developed from other places of enunciation. The second corresponds to the refutation of mid-range theories due to the biases of their empirical bases and the proposition of new analytical approaches for specific phenomena. These procedures allow for another convergence between Brazilian sociology and postcolonial epistemologies: the critique of dominant theories of modernity. --- Critique of the dominant theories of "modernity" The critique of modernization theories has been one of the main features of postcolonial debates, which seek to highlight coloniality as its obscure but constitutive side . Equally, it is possible to observe the development of critical readings in Brazilian sociological theory that seek to break with the normative and evolutionist understanding of the hegemonic theories of modernity. These theories adopt Europe as a universal model, basing themselves on a methodological dualism between European modernity and the premodern, traditional, backward nature of peripheral countries. If such criticisms can be found in our intellectual context throughout the 20 th century 2 , it is especially in the last century's final decades that new interpretations emerge. These focus on i. the construction of a critical theory of Brazilian modernity, which affirms the fully modern character of Brazil's national society, and ii. the development of a truly global theory of modernity. Jessé Souza ) drew up a critical theory of Brazilian modernity through a rereading of the classical interpreters of Brazil, rooted in the theoretical models developed by Pierre Bourdieu and Charles Taylor . Souza seeks to reinterpret the process of modernization and constitution of a class society in Brazil, returning to a central problem in the Brazilian intellectual context regarding the "specificity" of our society, in contrast to countries central to the modern West. According to Souza , the essayistic tradition in Brazil produces an interpretation of Brazilian modernity grounded in a European distortion, characterizing it as an inauthentic, "epidermal", and superficial process from its inception. His argument presents two fronts. On the one hand, against a "culturalist" legacy of Brazilian political-social Thought, imported by Gilberto Freyre from the Boasian circles in the USA, which Souza labeled the "personalist" tradition, whose representatives included Gilberto Freyre, Sérgio Buarque de Holanda, and Raymundo Faoro. On the other, he opposed the sociological tradition from São Paulo organized around the figure of Florestan Fernandes and dominant in the second half of the 20 th century. According to Souza , it is possible to observe, in both traditions, an emphasis on the persistence of premodern elements in the modernization of Brazilian society, which marked its incomplete character and its "specificity" about fully modern societies 3 . Contrasting these readings, Jessé Souza affirms the modern character of Brazilian society, whose modernization began at the start of the 19 th century with the arrival of the Portuguese crown to Brazil, thereby favoring the introduction of modern institutions . It is possible to observe the gradual abandonment of personalist and patrimonialist criteria, typical of traditional domination, and the subsequent adoption of properly capitalist criteria of social classification -such as merit and personal performance. These would be supposedly valid even for those who remain on the margins of the formal labor market in the post-abolition context. In the reinterpretation of the modernization process and social class structures in Brazil, Souza adopts Bourdieu's Theory of class based on a sociocultural understanding of these structures. They manifest themselves as appropriating impersonal economic and cultural capital, evolving over time, and understood through the notion of "trajectory". This theoretical model affirms the modernity of Brazilian society -whose social structure can be conceived through the same concepts applied to modern central societies -and the reanalysis of historical transformations in Brazilian class structure. Souza points out that the modernization process of Brazilian society is characterized by the marginalization of a significant part of its population. These individuals are denied not only the capacities that would allow them entry into the labor market as productive workers but also the social, cultural, affective, and emotional preconditions that could facilitate this entry. The distinctive features of Brazilian society would be the formation of a class characterized by a precarious habitus -marked by an absence of the economic and political dispositions that make up the "disciplined, productive, and useful subject", typical of a primary habitus . The author labels this class as the ralé estrutural . Its historical origins go back to the period of slavery and how the constitution of modern society on the periphery of capitalism, whose modern institutions emerged as "ready-made artifacts" , was disconnected from an extended moral and cultural maturation capable of universalizing an egalitarian logic throughout the internal conditions of societies that act on the character assumed by their respective processes of modernization population. In this sense, Souza stresses the selective character of Brazilian modernization. With this interpretation, Souza aims to go beyond Pierre Bourdieu's theorization, who proposed the generalization of those minimum dispositions to the entire population, using the example of the French case. According to Souza , such generalization does not occur in the periphery of capitalism. Therefore it is necessary to advance in the construction of new analytical instruments and a critical theory of modernization that apprehends the material and symbolic forms of domination prevailing in the current phase of capitalism in central and peripheral countries. However, Souza has been criticized for reaffirming the "peculiar" character of Brazilian modernity, employing the idea of selective modernization, which would deprive the national experience of a fully modern condition. This criticism has been made by Sérgio Tavolaro , who is amongst the current theorists interested in situating the experience of modernity in Brazil in the global context. Tavolaro seeks to distance himself from what he labels the "hegemonic sociological discourse of modernity", underlining its limits for understanding the contingent and historical processes of modernization worldwide. In analyzing Brazilian sociological production, both classic and contemporary, Tavolaro asserts that despite their theoretical differences, they converge on an interpretation of modernity in Brazil that does not place it on equal footing with the experiences lived in central countries. The so-called "sociology of inauthenticity" would encompass not only the tradition of atavistic culturalism of authors such as Freyre and Buarque de Holanda but also the Sociology of Dependency and the Marxist approach from the Escola Paulista, together with their interpretations of peripheral modernity -such as Caio Prado Jr., Florestan Fernandes, Fernando Henrique Cardoso, Octávio Ianni, among others . Thus, the center-periphery binomial has been consolidated, where the prevailing view is that Brazilian modernity remains incomplete. This is because, according to Tavolaro , our experience of modernity is evaluated according to an external model standard. Furthermore, in both traditions, there is a tendency to relegate historical variations in sociability to the background and to assume a normative pattern that has remained practically intact in Brazil for the past two centuries. In his critique, Tavolaro proposes the "de-essentialization" of modernity, which is seen as a process in which historically contingent institutions emerge. His analysis relies on two key notions: i. the modern pattern of sociability, and ii. modernity as a tendentially global and decentralized process from its original birthplace. The first notion is anchored in a critique of the modern episteme -represented by the sociological classics and, more recently, in figures such as Luhmann, Parsons, and Habermas. However, they fail to consider variations in three pillars of modern experience: a) social differentiation/complexification; b) secularization of normativity; and c) separation between public and private domains. This episteme, reproduced both by classical sociology and Brazilian sociology, projected an unchanging image of central societies based on these pillars, relegating "societies of delayed modernity" to the condition of "peculiar modern arrangements". Seeking to overcome the theoretical and empirical limitations of the modern episteme, Tavolaro proposes his version based on the ideas of varying patterns of social differentiation, varying patterns of secularization, and varying patterns of separation between public and private spheres. Such a derivation would allow an understanding of the configurations assumed by the "later modern societies" and those experienced by "central societies". Additionally, Tavolaro argues for the need to understand such patterns of modern sociability beyond simply national borders, highlighting the idea of modernity as a potentially global process. In this manner, even if central societies can be understood as the historical origin of modernity, they can no longer be considered the exclusive disseminators of this type of modern sociability. Therefore, the very assumption of the existence of coherent, linear, and national scripts in the direction of modernity becomes problematic . Based on this synthesis of the debates, we can sustain that Brazilian sociological theory is close to postcolonial concerns in its critique of the modern episteme, even though there are significant differences in the formulation of the critiques and the very conception of modernity defended by these perspectives. It is worth investigating the differences and specificities observed between Brazilian sociology and postcolonial epistemologies based on systematization and analysis of what we will call "strategic distances". --- Strategic distances In the review of research regarding modernity in Latin America, Sérgio Costa stresses that contemporary Brazilian sociological theory constitutes an intermediary response between the dominant interpretations of modernity and the decen-tering provoked by postcolonial approaches, such as in the case of the Theory of coloniality of power proposed by the Peruvian sociologist Aníbal Quijano. According to Costa, in the case of post and decolonial theories, the decentering is even more radical, to the extent that the very ontological and historical origins of the idea of modernity become geographically and temporally decentered. It no longer refers exclusively to the European/Latin American axis but rather is shaped by an understanding of the colonial power relations between Europe and the "rest" of the world more generally . The emphasis of these perspectives is much more on the constitutive link between modernity and coloniality, the relations of power and domination anchored in racial classifications, and the traffic of people and goods that formed the modern project. Such elements would not be considered failures or exceptions but rather elements integral to this project. Authors such as Walter Mignolo and Madina Tlostanova seek to take the decolonization of social relations and knowledge production to the limit, proposing a radical rupture with the univocal idea of modernity and the Euro-North-American canon. Although providing a counterpoint to the hegemonic theories of modernity, even the most critical voices within Brazilian social theory do not fully break with the "modern project" paradigm. Instead, many Brazilian theorists have emphasized the dangers of abandoning the entire theoretical-methodological tradition that could help understand colonial sociability patterns and relations in their diverse manifestations. José Maurício Domingues , for instance, examines the idea of modernity in Latin America, considering the "anti-modern risks" that its wholesale rejection could generate. On the one hand, locating the "evils of origin" of modernity on the European continent turned out to be unproductive. On the other, negatively essentializing or characterizing modernity as a whole, based on its "dark side", as Mignolo proposes, results in the promotion of the "native tradition" and nostalgia for the "autochthonous purity". Even if the discussion regarding "entangled modernities" involves a constant and, at times, conflicted negotiation between different, overlapping historical contexts, this does not presume the undoing of Eurocentric theories or the relations of epistemic dependence between North and South. Domingues proposes a "third phase of modernity" -more decentered, heterogeneous, and complex than the previous phases -and advocates for the validity of Critical Theory in producing a diagnosis of the present. The proposal seeks to respond to the systems of domination imposed on the global periphery that undermines democracy. Domingues points out emancipatory elements and the horizon of expectation that social movements in Latin America have demonstrated since the 1980s, mainly those centered around racial, gender, or environmental concerns. For him, theorizing on the periphery of capitalism should lead to changes in concepts and a more radical perspective on modernity , considering the sociopolitical dynamics these societies manifest. Therefore, Brazilian sociology, beyond producing a description of the local problems and case studies, should pursue general theorizations with universalist aspirations, playing its part in future processes of emancipation on the subcontinent and at the global level. Marcelo Maia also problematizes some ossified paradigms in postcolonial theory, warning about the perils of adopting such perspectives in the same manner European metanarratives of the 19 th century were embraced throughout the history of global sociology. According to him, in reviewing Eurocentrism and seeking alternative paradigms, we cannot fall into the trap of uncritically formulating a "Brazilian postcolonialism" since this would obscure other equally critical perspectives rooted in national particularities. Brazilian sociological tradition sought to understand the country starting from its specificities while constituting a conflicted, at times harmonious dialogue with the social imagination of other countries in Latin America and Europe . A possible way of replacing the unequal logic of the geopolitics of Thought, which mirrors global inequalities in the social and economic spheres, is to understand how questions once conceived as endogenous developments of our intellectual tradition can assume a sense and intelligibility when situated within transnational peripheral networks of cooperation. Brazilian political theorist Luciana Ballestrin has also pointed out the theoretical limitations and challenges that post and decolonial perspectives confront: […] is it possible to break with the logic of the coloniality of modernity without abandoning the contributions of Western/European/ Enlightenment thought […] for decolonization itself? Would the success of its project depend on its subaltern and peripheral condition? What is the limit of undoing the epistemological foundations of the social sciences? […] How should we deal with the European paternity of our political institutions and social Thought? How should we empirically verify the "colonized subject" today? Are experiments considered decolonial such as the new Andean Latin-American constitutionalism, free of contradictions? […] How does one methodologically operationalize the analysis of scales, levels, and spheres that coloniality includes? . These questions lead us to discuss the main theoretical divergences between the critical lineage in the Brazilian academic field and a more radical strand of the de-bates on post and decoloniality, especially regarding their relationship with the Euro-North-American sociological canon. We hypothesize that, in Brazilian sociology, there seems to be a greater willingness, in general, to engage in critical dialogue with the canon while still acknowledging its limits, i. either because they recognize in it a grammar that allows for a cross-cutting dialogue, ii. or because they claim the use of methodological repertoires considered useful for analyzing social phenomena, iii. or even because they defend the existence of elements of resistance and emancipation in the modern imaginary, which should not be entirely discarded. Our reflection reveals how the dialogue between a critical strand of Brazilian sociology and postcolonial theories can operate as a two-way street, marked by reciprocal contributions that enable both advances in theoretical approaches that are more decentered in relation to the sociological canon and the construction of appropriate repertoires for the analysis of the most diverse local realities. It is based on this hypothesis that we propose an exercise of "rereading" one of Guerreiro Ramos" main works, The sociological reduction, from 1958, which allows us to highlight the recurring concern of part of our academic field with theoretical-methodological issues and creative responses to the challenge of dealing with the hegemonic sociological tradition. --- A rereading of Ramos' sociological reduction Since the early decades of the 20 th century, Brazilian intellectual history has been characterized by a modernist ideal of a critical inflection in comparison with other theoretical traditions, which expresses a certain "anthropophagic" tendency of selective appropriation of different traditions in the process of intellectual creativity. Guerreiro Ramos" project of "sociological reduction" , in our view, expres ses this anthropophagy in the sociological field insofar as it systematizes the "critical-assimilative" procedure of foreign explanatory schemes without completely breaking with "alien theories" of the Global North while adapting frameworks and techniques to local circumstances. Therefore, we propose a critical exercise of "rereading" this author, which seems to us a productive path for bringing together a more critical tradition of Brazilian sociology and postcolonial epistemologies. For some time now in Brazil, research programs have sought to emphasize the relevance of certain intellectuals as "precursors" of postcolonial perspectives. According to specific interests, researchers have held up the intellectual production and trajectory of Guerreiro Ramos as a pioneering effort to articulate lines of investigation for a decentered theory and a critique of Eurocentrism, looking to understand sociology starting from the idiosyncrasies of the national context . We aim to present Ramos and his program of sociological reduction as a methodological alternative for expanding postcolonial praxis and then propose the hypothesis of possible "rereadings" of Brazilian sociology. Our argument starts from an observation and a wager. We believe that certain epistemological turns in Contemporary social theory, such as postcolonial theory, have reconfigured the way of analyzing our own intellectual history. The wager emerges when we observe that research programs such as Ramos' sociological reduction are methodologically original procedures that can be returned to, cultivated, and practiced not only in current Brazilian sociology but also by contemporary sociology at a global level. A rereading of Guerreiro Ramos in light of postcolonial critique helps us to confront what the author referred to at the time as "centripetism", which is the tendency of some peripheral intellectual traditions to always turn outward -especially toward hegemonic centers -in search of ready-made and model solutions to our problems. This tendency, according to his critique, is not limited to the academic field; it affects all levels of our lives, creating tension between the anachronisms of the country, the potential of its "structures in generation", and the difficulties in generating effective solutions: In terms of superstructural elements, this tension reflects a conflict between two perspectives: that of the old country and that of the new country, between the colonial or reflexive mentality and the authentically national mentality. In the realm of social sciences, this tension is also present. Until now, many scholars have conducted their work without considering the historical and ideological presuppositions of their scientific work. Their approach was reflexive, and they passively and mechanically adhered to criteria from developed countries. [...] To the literal and passive assimilation of imported scientific products, one must oppose the critical assimilation of these products. Therefore, the term "sociological reduction" is proposed here to designate the methodical procedure that seeks to make critical assimilation systematic . The "reduction" expresses, among other things, a concern to systematically reflect on the historical, theoretical, and ideological presuppositions that shape the social sciences. In his essay O negro na sociologia brasileira, originally published in 1954, the author highlights the ideological bias -for him, imperialist, colonialist, and ethnocentric -of fundamental concepts in European and North American sociology and anthropology, such as "acculturation" and "social change", which would constitute a "rationalization or disguise of colonial exploitation" . As these and other concepts -such as social structure -are uncritically appropriated by scholars from peripheral countries, through a mimetic process, they begin to act as a "powerful factor of alienation", contributing to the consolidation of a "quietist conception of society", which favors the "concealment of the decisive therapy for human problems in underdeveloped countries" . It is interesting to note that although Ramos proposes, in this text, an assessment of studies on the Black population in Brazil, his criticism directed at various Brazilian authors is oriented not so much by the conception of race they adopted -racist and, in any case, outdated from a scientific point of view -, but by the attitude assumed towards the foreign theoretical repertoire. Therefore, it is possible to observe in this criticism a sketch of what would be his proposal of sociological reduction as a critical-assimilative attitude and a methodological concern, considered fundamental to ensure sociological work in peripheral contexts. In this sense, Ramos converges with postcolonial approaches in his criticism of the colonial bias inherent in hegemonic sociological theories. However, in our view, the Bahian sociologist advances in the attempt to propose theoretical-methodological alternatives to deal with this legacy, responding to a yearning that marks the national academic field of the time and which can still be felt in the criticisms of contemporary Brazilian theorists addressed to postcolonial approaches, as discussed in the previous section. According to Ramos , sociological reduction, whether practiced in the theoretical-comprehensive domain or in the area of empirical operations, consists of eliminating everything that disturbs the effort due to its ideological, accessory, or secondary character of understanding and obtaining the essential of a social fact. In epistemological terms, to recover the "reduction" is to invest in a framework that, on the one hand, reflects a decentered view of Brazilian sociological Theory and, on the other, can entail contributions from critical Brazilian sociology to the future of postcolonial criticism. However, before advancing this discussion, it is worth clarifying what the notion of "rereading" consists of and its possibilities. --- Three potential avenues of rereading The idea of "rereading" should be understood as an analytical category. Rereading presupposes retrospection and implies reading in a different, displaced, and hete rotopic manner. It does not mean "redefining" a certain historical tradition based on its foundational authors, ideas, projects, contexts, and disputes, framing them in light of contemporary movements as if this group of factors prophetically announced the future or expressed avant-garde premonitions of epistemological discoveries. Instead, by sociological "rereading", we mean an epistemic displacement and decentering, a change of perspective, a new prism through which the sociological tradition in Brazil can be rediscovered, reevaluated, resignified, and refigured critically. This was the main challenge that postcolonial paradigms posed to Brazilian sociology. This problematization involves discovering new inflections within the same tradition as well as outside of it. The analytical perspective of "rereading" that we propose returns to and reconsiders the past in all its nuances while critically reappropriating it. This perspective is different and integrative, yet it is also capable of applying the "necessary corrections" to the tradition, which tends to be neglectful due to its historical determinations, as argued by Alatas and Sinha and Connell . Following this line of Thought, we can identify three significant sets of possible rereadings in social theory. The first type of rereading involves an analytical recovery procedure that provides necessary temporal corrections, capable of uncovering the tradition's elements that were neglected in their time. This procedure is analogous to what Alatas and Sinha practiced concerning European sociological classics . These authors demonstrate the timeless methodological qualities of the classics that remain useful regardless of different local contexts. At the same time, they also expose their conceptual, methodological, political, and ideological limitations in light of new displacements enabled by the epistemological turns in social theory. In the Brazilian case, an example would be the criticism of culturalist racism inherent in the interpretations of the generation of essayists from the 1930s, as highlighted by Jessé Souza regarding the concepts of "cordiality" and "patrimonialism" developed by Sérgio Buarque de Holanda. Even Caio Prado Jr., who starts from a materialist paradigm, is criticized for the cultural racism inherent in his proposal for modernizing the country, which is anchored in raising the cultural standards of our colonial roots -Amerindian and African . The second type of rereading enables the rediscovery of intellectuals marginalized by the hegemonic production of knowledge in the country or who belonged to silenced, unrecognized, or forgotten traditions. There are many examples, such as Alex Ratts with Beatriz Nascimento, Érika Mesquita with the sociology of praxis by Clóvis Moura, Sandra Siqueira , who analyzes the ostracism of theorists from the "radical current" of Dependency Theory, such as Vânia Bambirra, and Flávia Rios recovering the political and intellectual trajectory of Brazilian sociologist Lélia Gonzalez. This article also proposes a third possible form of rereading. This one seeks to uncover elements and biases from the margins of the history of sociological ideas in Brazil. It is more precisely a reorientation of the focus onto questions that were excluded or interpreted as contingent compared to what was epistemologically valued at the time. In other words, it is a recovery of authors, concepts, and theoretical-methodological projects from the intellectual limbo they were thrown to give them renewed relevance and recognition 4 . Mário Augusto Medeiros da Silva , for example, undertakes a critical reexamination of black sociologists such as Virginia Bicudo and Eduardo de Oliveira e Oliveira, amongst others, in São Paulo between 1950and 1970. Muryatan Barbosa reconsiders the idea of "black personalism" of Guerreiro Ramos, based on his activism in the Teatro Experimental do Negro of Abdias do Nascimento, and in a debate with theorists of negritude. Lastly, Adélia Miglievich-Ribeiro investigates the relationships of Darcy Ribeiro and Paulo Freire with intellectuals from other Latin-American and African countries. These three ideas of rereading seek to correct traditions, recover or rediscover authors, and uncover debates and reevaluate marginalized intellectuals. With these in hand, we can systematize both the impact of postcolonial perspectives on the sociological context in Brazil and the influence that Brazilian sociology, viewed from a transnational and "desprovincialized" perspective, can exercise on postcolonial approaches. These rereadings were made possible thanks to the problematizations provided by contemporary turns in social theory, especially by postcolonial epistemologies. To reread sociological traditions like the Brazilian one -in its most critical incarnation -can offer relevant contributions to postcolonial approaches. To better demonstrate this argument, we propose rereading the "diagnosis of the intellectual context" present in A redução sociológica, a pioneering work of Guerreiro Ramos. 4. However, different to the "presentist" perspective of Olli Pyyhtinen , the rereading that we are proposing here does not return to "the classics" of sociology to update them in light of contemporary problems and topics, but rather it divests those that were undervalued by their tradition of their secondary status, reevaluating their contributions in light of the new methodological and theoretical possibilities opened up by the epistemic turns in contemporary social theory. --- The sociological reduction against the colonial mentality As previously discussed, Guerreiro Ramos identified in the sociological research of the 1940s and 1950s Brazil a reflection of "academic and intellectual dependency", characterized by "scientific alienation" and "conceptual servitude" -a theme further developed in his work Myth and truth of the Brazilian Revolution . Ramos accused a portion of the national sociology community of his time of imprecisely applying foreign, "alien" production to Brazil in a mechanical and subservient manner without regard for the "historical and ideological presuppositions of scientific work" in a peripheral capitalist country . On the one hand, there is a sharp criticism directed at the arrogance of some authors who composed the sociological tradition in Brazil, which according to him, is an ambiguous expression of a "viralatismo" that reflects the "condition of copier and repeater" of the national intellectual elite in relation to European habits . On the other hand, there is an apprehension about "academic dependence" and the challenges to producing an autonomous sociology based on the abandonment of umbilical ties that relegate the sociological field of the country to a secondary and, therefore, unimportant position on the proposition of "instruments of self-knowledge and development of national and regional structures" 5 . Ramos suggests specific measures to contest Brazil's academic dependency: a) the removal of the "colonial mentality" 6 and its effects at the level of culture, ideas, and politics; and b) the presentation of the historical and ideological foundations for the "new critical awareness of the Brazilian reality". This awareness would be the foundation for an autonomous, authentic, dynamic, and less alienated sociology, free of "imperial shackles" and productive of theories adjusted to the demands of the national reality , which would set the stage for the emergence of an "epistemic subject" or, as he would have it, of the "parenthetical man". In this regard, Ramos argued for a type of "systematic training", which could prepare citizens" to transcend, as far as possible, circumstantial factors that conspire against its free and autonomous expression" . Sociological culture is then a qualitative component of resistance against "the robotization of conduct due to organized social pressures" . However, Ramos accuses the Brazilian tradition of presenting a "canned sociology" and a "consular sociology" . In both A cartilha brasileira de aprendiz de sociólogo and A redução sociológica, he asserts 5. This concern with epistemological colonialism and scientific autonomy appeared in other sociological fields of the Global South during the same period. We could mention Syed Hussein Alatas , who undertook a similar discussion in Malaysia, Pablo Gonzalez Casanova in Mexico, and Anouar Abdel-Malek in Egypt -who, incidentally, was the one who provided the concept "Orientalism" for Edward Said. 6. The critique of the "colonial mentality" is close to the phenomenon of the "captive mind", which Syed Hussein Alatas argued was persistent in peripheral scientific traditions. Regarding Alatas and Ramos, see: Maia . that Brazil was not producing sociologists capable of making a "sociological use of sociology". With this diagnosis in the 1950s, Ramos claimed that conventional Brazilian sociologists had become accustomed to mechanically incorporating foreign production, sacrificing their critical sense for the prestige they could gain from the lay public by using concepts and techniques imported from abroad, where the "best sociology" was being produced, but which were not effective in dealing with historical obstacles in the nation . It is worth noting that A redução sociológica is a work in which Ramos responds to criticisms he received at the 2 nd Latin-American Congress of sociology in 1953, reformulating and expanding some of the proposals presented on that occasion. Amongst these was his position concerning methodological questions. In the "Preface to the Second Edition", Ramos asserts that many "leading figures of conventional sociology in the country" were unwilling to "adjust" foreign research techniques to the material conditions of Brazilian society, fearing that such adjustments could alter the methodological character of their contribution and hinder their theoretical work. Ramos also criticizes the "provincial purism" that constitutes part of this national sociology and advocates that sociological work must always involve a modulation of scientific techniques and methods. Similarly, concepts and analytical models needed to be constantly adapted to situated realities and concretely underlie all authentic sociological inquiry -whose authenticity is measured by the degree to which it is linked to real problems of social life. The "authentic sociology" discussion had already been raised in "Sociologia enlatada versus sociologia dinâmica", presented at that Congress. Authentic sociology's essence, directly and indirectly, reflects an intention of salvation and social reconstruction. It is grounded in an experience of the community lived by the sociologist, as a function of which it gains sense […]. Whoever speaks of life speaks of questions. The essence of life is its incessant problematization. Therefore, to the extent that sociologist vitally practices their discipline, they are forcefully led to connect their thinking with their national and regional circumstances . Ramos also criticized the so-called "consular sociology" in Brazil for being "Bovarist", that is, distorting empirical reality, generating a significant error by perceiving an unbridgeable gap between the "world of the sociologists" and the "world of the layperson" . He grounded this analysis in a parallel between "conventional sociologists" and the "puritans" of grammar, who sought a linguistic vernacular for Portuguese in the 16 th century, which was not only unachievable but also nonexistent. Ramos argues that the "hypercorrection in sociology" is a contra-diction in terms "because there is very little sociology and a great deal of alienated and mystified consciousness". Therefore, what puts us in a balanced relationship with "foreign colleagues is not the memorized knowledge of their output but rather a mastery of the reasoning that it implies, which allows sociologists to do different things in different circumstances without harming the scientific objective" . As a counterpoint, Ramos advocated replacing "centripetism", the purist attitude of importing Euro-North American theories, which he called the "literal and passive assimilation of imported scientific products" , with a "critical-assimilative" methodology. This would lead to a "sociological use of sociology", the foundation of his sociological reduction. He developed the reduction method based on Husserl's phenomenology and Gurvitch's sociology of knowledge 7 . However, Ramos warned that the idea of reduction is foreign to European intellectuals who did not experience the challenge of decolonizing their sociological work 8 . They did not need to deal with what Cheik Anta Diop once denounced as a falsification of history, namely, the fact that the history of peripheral societies has been written solely "from the European point of view" . In this sense, The sociological reduction emerges as an anthropophagic method of "critical assimilation", selective of the intellectual patrimony from overseas 9 . This method is a way of overcoming what he described as "mimetic imperialism" and giving visibility to "autonomous and authentic sociology" . Reduction is not a mere transposition of knowledge from one social context to another but rather the quintessence of sociological work: a critical rereading of reality in its diverse situated expressions . Guerreiro Ramos underlines three senses of his "sociological reduction": i. as a method of "critical assimilation" of the situationally adjusted foreign production; ii. as a "parenthetic" attitude, guided by a readjustment of our analytic perspectives, namely a capacity to put social facts "between brackets", grounded in a systematic training of our perception, to perceive the world from another epistemic point of view; and iii. as a form of sociological overcoming in its merely institutional and academic dimensions, which takes into account sociology's potential as a science of making and being made, an unfinished project of "elab-orating a new knowledge" , whose elements are posed within a concrete society. Ramos" proposal converges with the analytical procedures highlighted by Maia in that sociological reduction invites both conceptual critique from other sites of enunciation and the proposal of new analytic approaches for specific phenomena. Moreover, revisiting Ramos" work, especially his proposal of sociological reduction, can offer inventive contributions to the critical fortune of postcolonial epistemologies as it enters the realm of discussions on methodological challenges faced by sociological practices in peripheral contexts. Adrián Scribano highlights that constructing a theoretical corpus involves integrating five ways of understanding society: epistemological, ontological, critical, theoretical, and methodological. While postcolonialism has emphasized the first four domains, little attention has been given to methodological questions. This suggests that sociology has not had much impact on "anti, post and decolonial" epistemologies and supports criticisms made by scholars such as Gurminder Bhambra and Julian Go , who argue that Postcolonial Thought risks becoming another "missing revolution" for sociology. Guerreiro Ramos does not relinquish the intellectual challenge of reflecting on methodological issues. In this sense, we highlight some elements of his "sociological reduction". Even if it is inductive, it is a systematically methodical approach to the extent that it apprehends social reality as it shows itself immediately to our inattentive eyes without evaluating its foundations and presuppositions, its conditions of possibility, and its connections of sense. Additionally, it is perspectivist and situated. As such, it postulates a world in which individuals and objects encounter one another in an infinite and complicated "web of references", within which they mutually constitute one another. Therefore, if we displace the analytic focus onto another perspective, both problems initially formulated and the objects analyzed cease to be exactly what they were. Due to this variability, a sociological problem cannot be understood as "disconnected from a determined context" . While presupposing a collective support , which suggests that sociology, in a generic sense, is not an "act of individual insight", The sociological reduction is also grounded in a type of "material logic immanent to society", It is highly rigorous and developed in methodological terms, and makes use of historical knowledge, the systematic study of social facts, and scientifically grounded reasoning. Lastly, the reduction constitutes a critical-assimilative procedure of foreign influences. This does not imply scientific "isolationism", nor the romantic exaltation of . local, regional, or national traditions, but rather be guided by the "aspiration to the universal" mediated by the force of local circumstances. It modulates the analytical method based on investigating real social problems from a specific society according to how they present themselves to us, without "distorting them" to better integrate them into foreign theories . Thus, Ramos seems to advocate for a certain flexibility of concepts and theory in light of empirical reality and the pragmatic interests of national sociology. Critiquing the philosophical fascisms that emerge in the chauvinist manner of staking a claim to methodological nationalisms, he distinguishes science in act from science as a habit. This latter one presupposes a fictitious imagination about the relationship between theory and practice in the field of intellectual work and therefore tends to "hypostasize" the sociological discipline, making it a highly privileged knowledge restricted to a few. In this sense, sociology would be ideological and informational, just another European legacy. "Science in the act", on the other hand, reflects a methodic and scientific attitude in the face of concrete reality. It bases its method on what it encounters in concrete reality, in the dynamics of lived experience. Its future and main goal is to stop being the knowledge of a few specialists and become practical knowledge for general citizens. In the "Preface to the Second Edition" of A redução sociológica, written around 1963, just prior to the civil-military coup in Brazil, Ramos advocates for an ideal type of public and activist sociology as a horizon for research when criticizing scholastic, conventional sociology, "separated from the life world" and represented by other Brazilian sociologists. According to Ramos, "the vocation of sociology [in Brazil] is to become public knowledge. Sociology will volatize itself in the very global social process". These elements show us the strength of The sociological reduction, a sufficiently rigorous and creative research program that allows for the anthropophagic appropriation of social theories, concepts, intellectual experiences, and normative ideas produced in countries of the Global North. However, this always occurs within the construction of autonomous sociological knowledge, concerned with confronting concrete social problems to guarantee peripheral societies" political and intellectual self-determination. Therefore, we understand that the "reduction" coincides with and complements more radical projects of decolonization of knowledge found in postcolonial epistemologies. --- Final considerations Rereading Brazilian Sociology implies revisiting diverse traditions -some well-established, others hidden -and taking a renewed look at them. Nevertheless, as we argued at the start, this is not to "fit" authors from the past into contemporary epistemological dynamics, nor would it be to attribute to them concerns that were not necessarily theirs. Both operations would be anachronistic equivocations. Even though Guerreiro Ramos mentions intellectuals such as Frantz Fanon, Cheikh Anta Diop, Aimé Césaire, and Abdoulaye Ly, amongst others, it is not our intention here to situate him as a postcolonial movement member, even more so because he makes use of categories that are criticized by these movements 10 . This does not prevent us from understanding him as an author with "postcolonial aspirations" and especially as a genuinely critical, anticolonial, anti-Eurocentric, and antiracist theorist who is a relevant theoretical reference for contemporary debates and who can contribute to the future development of postcolonial epistemologies in the global level. We proposed in this article a possible rereading of Guerreiro Ramos, suggesting that his contributions should be understood not only as an object of study for the history of ideas in Brazil but as a source of sophisticated, relevant, and up-to-date methods and theories. We argued that his idea of "reduction" could be understood as contributing to postcolonial epistemologies from Brazilian sociological theory. Ramos set out the fundamental preoccupations for the sociology of his time, proposing an innovative method of investigation to examine peripheral societies and offering a critique of forms of domination, servitude, colonialism, and dependency. For Ramos, sociology is not mere philosophical speculation. Rather, it is an occupation that requires to "roll up your sleeves", a praxis with practical and empirical consequences. If postcolonial approaches have already established unavoidable questions for social theory and have provoked a necessary decentralization of its main intellectual traditions, we believe in the interlocution between such approaches and Brazilian sociology, considering that the latter can contribute to the advancement and deepening of this critical movement. The proposal of possible rereadings of Brazilian sociological theory allows for the recuperation of the main contributions of this academic field, especially in terms of: i. the importance assumed by the methodological dimensions of the social analysis, including as a way of overcoming the relations of epistemic dependency between North and South; 10. There are various examples of it: "social evolution", "modern history", "development", "universal man", "human nature", and "progress", concepts of their time and place, and of the circumstances of the intellectual and political debate between the 1950s and 1970s. ii. the meeting of challenges that present themselves to an analysis of the "colonized subject", which recuperates part of the European and American philosophical and sociological legacy; and lastly iii. the emphasis on the variable, contradictory, and decentered character of the idea of "modernity", understood as an invitation for new theorizations of a global reach, seeking the critical formation of new emancipatory horizons and political connections at the South-South level. All these contributions suggest that, beyond simply opening the discipline's canon, Brazilian sociology appears engaged, with all its specificities and limitations, in the decolonization movement of our disciplinary field. BARIANI, Edison. Guerreiro Ramos e a redenção sociológica. São Paulo: Editora USP, 2011. ______. Padrão e salvação: o debate Florestan Fernandes × Guerreiro Ramos. Cronos, v. 7, n. 1, p. 151-160, 2006. BARRETI, Paula; LIMA, Márcia; LOPES, Andrea; SOTERO, Edilza. Entre o isolamento e a dispersão. A temática racial nos estudos sociológicos no Brasil. Revista Brasileira de Sociologia, v. 5, p. 113-141, 2017. BARRETO, Paula; RIOS, Flavia; NEVES, Paulo; SANTOS, Dyane. A produção das ciências sociais sobre as relações raciais no Brasil entre 2012 e 2019. Revista Brasileira de Informação Bibliográfica em Ciências Sociais , v. 94, p. 1-35, 2020. BERNARDINO-COSTA, Joaze. Decolonialidade, Atlântico Negro e intelectuais negros brasileiros: em busca de um diálogo horizontal. Sociedade e Estado, v. 33, n. 1, p. 117-135, 2018. BHABHA, Homi K. The location of culture. London: Routledge, 1994. BHAMBRA, Gurminder K. Sociology, and postcolonialism: another "missing" revolution? Sociology, v. 41, n. 5, p. 871-884, 2007. BORBA, Pedro; BENZAQUEN, Guilherme. Teoria crítica nas margens: um diálogo entre marxismo e pós-colonialismo. RBCS, v. 35, p. 1-17, 2020. --- 7. According to Ramos , "the sociological reduction, although permeated by the influence of Husserl, is divergent from an eidetic social science. What we take from Husserl is not so much the philosophical content of his method as a fragment of his terminology. Additionally, the idea of the sociological reduction as it is conceived of and presented here never occurred to Gurvitch. This idea is foreign to Gurvitch who does not experience the problem of the decolonization of sociological labor". 8. "Sociological reduction" is something different from an eidetic science of the social. What we took from Husserl was less the philosophical content of his method than a fragment of his terminology. Moreover, the idea of sociological reduction as conceived and expounded in this book never crossed Gurvitch's mind. This idea is foreign to Gurvitch, who did not experience the problem of decolonizing sociological work" . 9. The "criticalassimilative" methodological procedure is not against international theoretical influences, and therefore, it is not THERBORN, Göran. Entangled modernities. European Journal of Social Theory, v. 6, n. 3, p. 293-305, 2003. VANDENBERGHE, Frédéric. Metateoria, teoria social, teoria sociológica. Prefácio à tradução brasileira. In: ______. Uma história filosófica da sociologia alemã -v. 1, p. 1-25. São Paulo: Annablume, 2011. YOUNG, Robert. Postcolonialism. Malden, MA: Blackwell, 2001.

The article discusses to what extent sociological theories produced in the Brazilian academic field dialogue with a global intellectual movement criticizing coloniality and the Eurocentric foundations of the social sciences. Initially, we analyze the challenges regarding the attempts to define two theoretical approaches, Brazilian sociology, and Postcolonial Thought, without overlooking their internal heterogeneities. Then, we address the tensions between these approaches as conditions for research agendas that bring both contributions into proximity. Finally, we explore the epistemological potential of one of these agendas, which corresponds to a rereading of Brazilian sociological theory in light of postcolonial criticism. This exercise in rereading the canon is based on the methodological program of sociological reduction of Guerreiro Ramos, which indicates a reciprocal interrogation between Brazilian sociology and postcolonial thought, i.e., a decentered look at our sociological tradition that also reveals contributions from this tradition for the future of postcolonial epistemologies.

Introduction When we go back there, our children don't love us and don't obey us and don't treat us like we were their parents. They have other ways of thinking. So, those are the things that are truly harming families. In Guatemala, there are people, who are a little better off, who are teaching our children to misbehave and do things that are not good. That's how our lives are, but we put up with it to stay here. Since we're so far away, if a family member dies, we can't go there. So, we have to put up with it. Here in the United States, we feel good because we have food and jobs. We can find money and a job. For that reason, we like to be here in the United States. One central problem in studying labor migration-and in understanding capitalist market economies in general-is how to evaluate decisions made in the face of both wages and well-being, economic necessity and human dignity. In the excerpt above, one migrant farmworker offered to us this apparent contradiction between valued remuneration and social woes, illustrating that his ability to favorably construct his life circumstances was limited, and in specific ways. In order to understand migrant farmworkers' housing, employment, and migration decisions better, the purpose of this research is to determine how far and in what ways Latino migrant farmworkers living in eastern North Carolina labor camps feel their agency extends in responding to substandard housing and work conditions, while also considering who among them has the power to exercise this agency. Such an understanding is critical for developing high-impact policies and interventions that address the lived realities of migrant farmworkers and families in the United States. Accordingly, we propose two specific aims: to describe why, in what ways, and by whom housing conditions are accepted, contested, and resisted by migrant farmworkers and their families; and to delineate the selfperceived limits of farmworker agency at the intersections of their employment, housing, and immigration status. This analysis builds on that of Keim-Malpass et al. , which provides specific descriptions of housing conditions. Prior research on farmworker perceptions of their housing environment is scarce , even more so when analyzed in terms of agency. It is important to remedy this dearth of research because housing serves as a valuable lens through which one can survey many aspects of farmworkers' and their families' lives, given housing's role as a major social determinant of health from stress levels to respiratory disease-and its linkages to employment and legal status for so many migrant workers. Therefore, if housing is so fundamentally entangled with critical domains like health, employment, and legal status, it is crucial to know how individual farmworkers and communities perceive their housing conditions, how they do and do not act to improve those conditions, and the material and subjective contexts of those agentive practices. --- Dark Anthropology and Agency Moving away from the 1970s' and 1980s' celebratory, resistance-based conceptions of the agency of marginalized groups , much of anthropological theory now "emphasizes the harsh and brutal dimensions of human experience and the structural and historical conditions that produce them," a new iteration of the field that Ortner terms "dark anthropology." Oftentimes combining-whether coherently or haphazardly-a Marxian focus on expanding material inequality and classbased, capitalist exploitation with a Foucauldian conception of ubiquitous and subjectifying power, scholarship within this paradigm presents a world system and the lives that populate it as determined by the structures of neoliberal capitalism and its totalizing rationalities . With so much of contemporary theorization of agency emanating from this scholarly response to contemporary dislocation and domination, it is no wonder that labor figures prominently among leading empirical contexts for the study of agency. In this vein, we define agency as "both the intention and the practice of taking action for one's own self-interest or the interest of others" ). Conceptually, we rely heavily on the work of Rogaly andCarswell andDe Neve to construct a theoretical framework of agency that spans multiple analytical, spatial, and temporal domains, allowing us to consider migrant farmworkers' agency as a force and set of practices not limited to their workplace or housing, to their sending or receiving countries. In this conceptual framework, we aim to analyze agency by considering both the effects and intentions of farmworkers' and their families' agentive acts. In some political economic contexts, agency has been found to have a structurally transformative potential: researchers studying South Indian garment workers argue that agency and structure are mutually conditioning to such an extent that their relationship is a "two-way, iterative interaction in which capital responds to labour's agency and labour to capital's strategies" . On the other hand, Rogaly ) concludes that unorganized migrant garment workers in India are only able to achieve "incremental changes in conditions at micro-scales" through their agentive practices. These two, contrasting summations of the effects of agency lead us to believe that only through a study of the specific social, political, and economic circumstances of North Carolina migrant farmworkers can we reach any conclusions about their capacity or lack thereof to transform their living conditions. Furthermore, by describing the intentions that motivate their actions, we will consider whether the structures shaping migrant farmworker agency are merely materially restrictive or are more totalizing in their colonization of farmworker subjectivities, as darker anthropologies of hegemony may have us believe. Scholars working in California's agricultural industry, in particular, have explored the relationships between migrant farmworker agency, on the one hand, and the organization of contemporary agribusiness production, on the other. This area of research has produced two main lines of inquiry that guide our analysis: the transformative capacities of migrant farmworker communities and the influence of farm production on the organization of non-wage migrant labor. In reference to the former theme, researchers have marshalled diachronic ethnographic evidence and historical accounts of regional political economies to counter popular narratives surrounding farm communities of migrant and non-migrant laborers and the capacity they may have to transform those places . Du Bry's case study of rural Mecca, California, describes how communities of migrant farm laborers can change over time in the context of local labor market restructuring, leading to signs of socioeconomic mobility, increasing complexity of community life, and budding possibilities for community building. We aim to extend this line of research to North Carolina agricultural sectors structurally distinct from California's and to ask how a predominance of one form of housing tenure-employer-or contractor-provided-influences the development of migrant farmworker agency. In reference to the second theme, researchers have documented multiple ways in which informal labor practices subsidize capital and reproduce its position vis-à-vis labor in contemporary agribusiness . One important object of study within this body of research is the influence of production schemes on non-wage labor within migrant laborer households . Figueroa Sánchez looks to the relationships between California family farmers and Mexican immigrant sharecroppers to see how "capitalist agriculture requires the interlocking of strawberry arrangements with familial patriarchal relations." We extend this line of research beyond Figueroa Sanchez's California strawberry patches and into the domestic labor performed in maintaining and improving North Carolina labor camps, with similar concerns surrounding self-exploitation and the reproduction of exploitative conditions, albeit with the caveat that our limited sample size precludes a similarly "deep" analysis of gender relations in this dynamic. --- Migrant Farmworkers: Representation and Identity Migrant farmworkers' extreme legal, economic, and social vulnerability belies their basic rights, human dignity, and valuable contributions to society. Migrant farmworkers, commonly defined as persons who "travel at least seventy-five miles during a twelve-month period to obtain a farm job," form an integral part of United States agricultural production . Although migration patterns vary greatly, many migrant farmworkers relocate in spring months from "home base" communities in Florida, Texas, southern California, Mexico, the Caribbean, or Central America to growing season employment opportunities farther north, only to return to their "home base" in the fall or winter . The farmworker population is predominantly male, foreign-born, Spanish-speaking, having dependents, and lacking authorization to be in the United States . However, this characterization must be considered with caution because stereotypes of migrant farmworkers shape the interventions, policy, and discourse surrounding them. Specifically, this nationwide characterization does not include family members travelling with farmworkers, nor does it illustrate the high degree of regional variation in farmworker demographics. Our analysis specifically addresses migrant farmworkers and their families living in employer-provided labor camps in North Carolina, the sixth largest farmworker population in the United States with around 100,000 individuals . One dated enumeration profile of the state suggests that it hosts a higher proportion of migrants than the national figure , making analysis of the migrant-rather than seasonal-workers in the state especially important . Keeping in mind the regional variations and diversity of individuals who accompany farmworkers, we consider who-based on life experiences, attributes, and identities-lives in these labor camps and how they respond differently to substandard conditions, showing how structure is differentially contested and agency dynamically constructed. --- Agentive Acts in Institutional Isolation Although patterns of housing acquisition vary regionally, migrant farmworkers often have unique housing arrangements that divest them of choice and control over their circumstances. Along the Atlantic Coast, the primary means of housing acquisition for migrant farmworkers entails employers providing housing at no cost . Studies have found that North Carolina farmworker families living in grower-provided housing have the lowest quality housing, compared to housing acquired by other means . This observation is even more troubling when one considers that this form of housing tenure is directly contingent upon maintaining that specific employment. The institutionalized forms of agentive action for farmworkers are limited, complicated, and underused. Although states vary in their protections for migrant farmworkers,1 United States farmworkers are almost entirely excluded from key labor protection laws on the federal level, including most of the Fair Labor Standards Act's minimum wage and hour guarantees, the National Labor Relations Act's collective bargaining protections, and many of the basic safety standards issued by the Occupational Safety and Health Administration . The resulting legal exclusion and invisibility means that farmworkers are often left to individual acts of negotiation in securing adequate housing and employment. The current system of statutes regulating migrant farmworker housing is notoriously complicated: standards vary by state, and the overarching OSHA regulations do not apply to farms with less than ten employees. In North Carolina, the law that regulates migrant farmworker housing is the Migrant Housing Act . It requires owners and operators of temporary housing for migrant workers to register with state authorities who will then perform preoccupancy inspections to enforce standards for heat, hot water, cooking facilities, fire prevention, and kitchen sanitation . However, many aspects of farmworker housing are considered optional, such as having showers and toilets in separate rooms, and having privacy-dividers between individual toilets and showers . Standards that are mandatory, such as adequate storage for personal belongings, beds raised above the floor with clean mattresses, and laundry facilities, are not always well enforced, with one study documenting a mean of 11.4 total violations per camp among North Carolina labor camps . Post-occupancy inspections-those when residents are present and generated in response to "complaints, referrals, a randomly generated assignment, or an accident or fatality report" -are very rare, with the Agricultural Safety and Health Bureau conducting sixty-seven post-occupancy inspections and issuing only 116 violations in all of 2014 . This, alongside the widespread violations found within regional studies, leads us to believe either that farmworkers rarely utilize the framework provided by the North Carolina Department of Labor for addressing noncompliance or that those complaints do not always result in follow-up inspections, both of which prompt important questions concerning how labor camp residents act to improve substandard housing and the effects of those actions. --- Methods Data for this study came from twenty-three in-depth interviews conducted between August 2009 and October 2010 as part of a broader community-based participatory research project documenting North Carolina farmworker housing quality and health. For this research, collaborating partners included the North Carolina Farmworkers Project , Student Action with Farmworkers , Harvest Family Health Center , Piedmont Health Services , and Wake Forest School of Medicine . The research plan was reviewed and approved by the Wake Forest School of Medicine Institutional Review Board. Interviewers spoke with thirty farmworkers and partners for the broader research project. Of the thirty farmworkers interviewed, twenty-three were migrant farmworkers or partners living in employer provided housing, or what we refer to as "labor camps." This analysis is limited to these twenty-three migrant farmworkers and partners. Inclusion criteria were being an adult farmworker or an adult partner living with a farmworker and the ability to speak Spanish or English fluently. Community partner organizations identified potential participants, who were then introduced to project interviewers. The sample was recruited to reflect a diverse set of housing situations and household compositions, including unaccompanied men and members of migrant families in small and large labor camps. Staff members screened potential participants relative to their informally gauged interest, making it impossible to provide the number of potential participants who refused to participate. Once introduced to potential participants, interviewers described the project, told them that the interview would take about ninety minutes to complete, that it would be recorded, and that they would receive $30 as an incentive. The interviewer then answered any questions the participant had and obtained signed informed consent. All individuals who met with interviewers completed the interview process. Interviewers conducted semi-structured interviews with migrant farmworkers or their spouses living in labor camps based on an interview guide developed a priori by the research team and community partners. The topics addressed in the interview guide included descriptions of the housing itself; of household composition; of knowledge of housing regulations, rights, and avenues for addressing grievances; of experiences with landlords/growers; and of knowledge, experience, and skills in repairing their housing. The interview guide was then translated into Spanish by a professional translator with experience among North Carolina farmworkers. Fluent Spanish speakers who work with farmworkers then reviewed the translated guide. Investigators trained three interviewers to conduct the interviews. All three spoke Spanish fluently; two were also fluent English speakers. Interviews were completed in participants' homes and were conducted in Spanish. Interviews were transcribed verbatim and translated into English. 3 Using Hsieh and Shannon's techniques for directed content analysis, we analyzed textual data by reading and re-reading the transcripts line-by-line in order to develop codes and categories. ATLAS.ti v5.0 software was used for organizing and coding textual data into a preliminary coding dictionary. Codes and categories were then connected and themes formed by utilizing the concepts of structure and agency to chart the consistencies and contradictions among descriptions, evaluations, and actions across space and time in participants' lives, thereby plotting the basic contours and characteristics of migrant farmworkers' and their partners' agency. Data saturation was achieved when no new themes emerged. Bias control was enacted by allowing members of the research team and community partners to review aspects of the study and by the lead author maintaining a field notebook during analysis to facilitate reflection on prior assumptions. Once identified, salient codes and themes were evaluated, analyzed, and included in the results. Based on Arcury et al. 's analytical framework, a salient code or theme is a domain of belief, knowledge, or behavior that is widely shared within a group. Salience, therefore, relies upon a consideration of three factors: how widely among group members a domain is referenced, the affect with which it is described, and its explanatory capacity. --- Results Twelve participants lived in trailers, seven in barracks, and four in houses. About threequarters of participants were men, and six individuals were women. Nearly all participants had dependents, but only nine individuals had any dependents living with them. The ages of the participants ranged from twenty-three to sixty years, with a mean of forty years. Eight participants had an H-2A visa at the time of the interviews, and only one participant had permanent resident status. Slightly over half of the participants had spent less than a year at their labor camp at the time. --- Farmworker Perceptions of Grower-provided Housing Farmworkers reported ubiquitous and persistent deterioration in many labor camps. These statements highlighted housing unit structures, neighborhood features, and housing appliances that were broken or in disrepair. Instances of deterioration were common across almost every aspect of housing, from stove burners and roofs to door locks and shower floors. One woman said, "These trailers are in ruins, and sometimes we feel like we will fall through the floor, and sometimes we feel like the roof is falling down" . Another man living in a trailer said, "The pipes that are underneath leak. And that's where animals such as snakes or caterpillars could get in." Fluctuations in crowding, nearby pesticide use, and structural integrity were reported among all types of housing structures. One woman living in a barracks demonstrated the fluid, unpredictable nature of her social housing environment and its close relationship to employment: Right now, there are not many [other tenants] because they didn't like the job and left. They said it was too hard. They used to work in construction and they made a lot, so they didn't like the money they were making here. So, they left. Right now, I think, there are about 15 people. No, not even 15 because some went to a house that [NC contractor] has in Benson. --- Putting Up with It "Putting up with it" was a highly salient theme among farmworkers when explaining their housing, employment, and health. One man described the suffering he put up with in his leaking and overheated house: "Sometimes, when it rains and there is a lot of water, we feel bad. We feel the same when it's hot inside here, but since we have a job, we put up with those conditions. We just suffer. When it's cold, we feel very cold" . As this man pointed out, there were many aspects of housing with which participants put up, but housing itself often served as the reasoning by which the other domains of negative employment and health circumstances were tolerated. So, even when participants put up with non-housing related issues in the domains of employment and health, those instances of putting up with it were still grounded in the housing environment. One man simply stated, "We always have pains, but we put up with them" . Putting up with it was what farmworkers did when confronted with the limits of their agency within their own lives; it was not an option of first-choice. When asked what he did when the stench from the nearby chicken houses became too strong, one man responded, "What can we do? We have to sit here and bear the smell" . Framing their actions as "putting up with it" was a way to label poor conditions as such and entailed a begrudging endurance of those conditions' worst manifestations as they arose. Rhetorically, it highlighted the structural constraints that impeded farmworkers' ability to avoid and substantially improve poor housing and working conditions. But materially, putting up with it functioned as a forbearance of those very conditions. One man articulated a sense of inevitability and acceptance inherent in putting up with it, initially with reference to his health, but pointing out its broader applicability: "Well, I don't have any worries because we have to accept anything that comes. Diabetes is what is attacking me, but I look at it like everything else. I take care of myself, but I don't worry about it" . As this man demonstrated, participants often rationalized putting up with it as a form of selective self-care. By coming to a frank recognition of what was and was not in his control-and by putting up with the latter-he attempted to bypass the emotional toll of grappling with the aspects of his life he could not change and focused on the aspects of his life that he perceived he had the ability to control. The possible adverse effects of such a life management strategy were evident in the testimony of one fieldworker: "We're very strong. If something in our body hurts, we don't say anything. We put up with it until it goes away. We don't go to a clinic just because we get cut. We put up with it. It's the way we manage our lives" . --- Limits of Agency The above quote from participant ID-14, a sixty-year-old man living in labor camp barracks, also shows how putting up with it was grounded in a sense of farmworker identity. Multiple farmworkers made comments that rationalized hardship in housing, work, and health for Latinos, Mexicans, immigrants, the poor, or migrant farmworkers, in general, based on those identities or combinations of those identities. One man said that his boss and many others spray pesticides while workers are still in the fields, explaining, "We put up with it because we're Mexicans. They say that Mexicans put up with more than anyone else"4 . Here, common stereotypes about Latino, Mexican, and Central American workers were internalized by migrants and then used to partly rationalized their treatment. A forty-twoyear-old participant with an H-2A visa living in a labor camp trailer highlighted how his internalized socioeconoimc, nationality, and employment statuses informed his perceived imperative to put up with poor housing conditions: "We have to put up with [leaking pipes and pests] because we're poor, and we're not from here, and we want to work. For that reason, we put up with it and continue to be here. Even though we're afraid of those things, we keep living here" . Putting up with it illustrated how farmworkers were sorting out in real time the conflicts and constraints they felt between their economic necessities and the unpleasant or harmful phenomena associated with their work and housing, from nearby farming operations to poor health. These real time considerations were evident in the occasional contradictions in workers' evaluations. Upon being asked what health worries he had, one sixty-year-old migrant farmworker began his interview by describing how he put up with the diabetes that was "attacking" him , a moment that was followed a few minutes later by this contrasting exchange: I: We were talking about your rooms. How do you feel living here in the United States? R: Well, I feel great. I come here wanting to work. When you come willing to work, you feel good. This shift of evaluative tone-between negative and positive evaluations-during the interview was not uncommon among participants and was closely related to the subject being discussed. This man considered his health and well-being to be negative aspects of his migrant experience in the United States, but that did not stop him from positively evaluating his overall experience when he considered it from an employment perspective. One man with a wife and three children in Mexico explicitly described the economic imperative that drove him and other migrant farmworkers to put up with poor housing conditions: "If we don't want to do it like this [live in those housing conditions], we have to rent, and that's expensive. For example, in Pennsylvania, we paid $210 each for a month's rent. Here, we don't feel very comfortable, but at least we save that [$210]" . Most of the behaviors connected to putting up with poor living and working conditions consisted of small adaptations that maximized benefits for farmworkers in the face of aspects of their work and housing that they could not change. Farmworkers described their attempts to appropriate what value they could out of their environments, actions that included decreasing perceived crowding by using living rooms and hallways as bedrooms ; pooling resources for expenses within buildings and labor camps, especially to support new arrivals ; building shelves on porches to create sufficient storage for all housemates when storage was not provided by employers ; and utilizing outdoor space for social events . Labor camp residents oftentimes used materials and skills from their employment to repair their constantly deteriorating housing. But these frequent repairs were rarely effective in creating satisfactory living conditions: "If we fix the bathroom, the ceiling falls. If we fix the ceiling, the floor falls. Everything is falling apart" . For the women who shouldered many cleaning responsibilities, their unpaid labor in maintaining dwelling structures was demanding and constant: "I struggle the whole day. I clean, and they make a mess. I spend the whole day cleaning. I never finish. You might think that I don't do anything, but I'm always working" . In sum, these actions were completed with little to no assistance from labor camp owners, non-uniformly among residents, and outside institutional avenues for redress. In fact, only one participant described actions in which he formally resisted his employer's mistreatment. At one point, he had physically resisted and afterwards formally reported his boss' refusal to drive his housemates and him to the grocery: So, we walked to Walmart, and we got a shopping cart. Then, I asked the manager if I could borrow it because I lived so far away, and I didn't want to carry all those bags.… When we were almost home, someone who knew the patron saw us and beeped his horn and told us to get in the truck. I told him that I wasn't going to. Some of the guys, who I came with, did get in the truck because they didn't care about the point I was trying to make. My other coworkers did support me in what I was trying to do. He [the patron] asked me why we had [refused the ride from Walmart], and I told him that it was because we didn't have food. It is noteworthy that this man had a relatively high level of education , had a driver's license, had an H-2A visa, and spoke English. This man consequently occupied an elevated position in the farm's labor hierarchy as an informal supervisor, with responsibilities including occasional transportation of other workers and communicating with the employer about housing. After publicly refusing the ride from Walmart and reporting that violation of transportation responsibilities, this man did find improvements in working and housing conditions but exclusively for himself: "He started to treat me well, but only me because he did whatever he wanted to with the rest of them. He yelled all sorts of things at them because the other guys didn't know [they could report him]" . Nonetheless, other participants' descriptions of putting up with poor housing hinted at future possibilities of no longer putting up with it. One man, age sixty-two and with a wife and three daughters in Mexico, had been coming to work for the same boss for fifteen years, but his seemingly consistent migrant round was marked yearly by moments when he considered ending it: I tell my daughters to not ask me about [the living conditions in North Carolina] because I don't want to remember anything about it. For that reason, they don't ask me anymore, but I continue struggling. My daughters tell me not to come here anymore, but I tell them that I'll only do it one more year. Every time I come, we argue about that because they don't want me to come and work here since I'm sick and the heat makes me sicker. Another man emphatically demonstrated the limited extent to which he accepted the poor conditions of his trailer: "Sometimes, I tell my wife that I just want to take a match and burn it down" . --- Discussion We argue that putting up with it was a way for those living in labor camps to delineate their agency and manage their resources, psychologically and materially. The limits of that agency were primarily dictated by the close relationship between migrant farmworker employment and housing, by the importance of remittance wages for dependents, and by the naturalization of poverty and suffering for poor Latino immigrants. Along with differences in individual characteristics, this constellation of factors formed a structural nexus that was pivotal in determining the form and extent-and, consequently, the effects-of agentive acts by migrant farmworkers and their families to improve their living conditions. Accordingly, labor camp residents exercised agency through the pursuit of minor material improvements and symbolic appropriations of identity-based dignity. These expressions of agency resulted in minor but real improvements in everyday conditions for farmworkers and their families but cumulatively served to perpetuate substandard conditions in the long term by maintaining living areas at little to no cost to employers and without the intervention of regulatory authorities. --- Housing Maintenance, Life Management When putting up with harmful and unpleasant conditions, participants conserved and maximized their resources, psychological and material. Poor conditions were not unnoticed nor condoned; they were consciously denounced to interviewers, and actions were constantly taken to ameliorate their immediate manifestations in housing. Rather than struggling against seemingly inalterable structural forces and the individual employers and contractors in their lives that produced and reproduced those forces, putting up with it allowed farmworkers to focus on the care, improvement, and coping mechanisms that were most feasible. Therefore, putting up with these conditions was an agentive act with defined, albeit fluid, limits. Instances of putting up with it often began with participants weighing their health, comfort, dignity, and well-being against the usefulness of wages earned, the benefits of rent-free housing, and the possibility of dismissal from employment. This consideration reflected the fact that the domains of housing, work, and health overlapped and were mutually reinforcing for many participants: to work for a grower is to live in the housing the grower provides. Migrant farmworkers put up with instances of maltreatment, neglect, or substandard conditions in one domain because of a lack of agency in other domains. That is to say, some migrant farmworkers rationalized acceptance of poor housing because resistance could terminate their employment. Other participants put up with heat exhaustion and associated health issues because those very problems originated in the housing and work over which they had little control . Therefore, a consideration of agency within substandard housing must necessarily include a horizontal consideration of the broader domains of employment, health, and legal status that are linked to it through various institutional, informal, and legal arrangements. Oftentimes highlighting the value of remittances abroad, farmworkers presented the risks and hardships that accompanied their living and working conditions as trade-offs for their income. The preponderance of economic rationalizations for their suffering derives from poverty's and inequality's central, but not exclusive, roles in driving labor migration. Researchers highlight factors intimately related to employment, income, and purchasing power-namely "recurring Mexican economic crises, peso devaluations, and Mexican policies aimed at economic modernization"-when describing why supply-push factors became more important in Mexico-United States migration from the 1980s to the late 1990s . International poverty and inequality thus support and reinforce acquiescence to substandard housing and employment conditions. --- Scaling Agency When putting up with substandard housing, farmworkers were creative in the actions and behaviors they utilized to develop comfort, safety, and well-being: organized cleaning schedules, do-it-yourself repairs, and repurposing rooms and hallways were common. And in the one instance of outright resistance , his symbolic and physical refusal of insufficient and late transportation support from his employer did lead to better treatment for him. However, these actions-and other forms of resistance-were rarely organized past the household level and produced improved conditions that were limited to the individuals who personally instigated the complaint or resistance. In this way, farmworker agency encompassed informal, non-institutionalized means of resistance, primarily minor material improvements and symbolic appropriations of dignity. Prior research highlighted the importance of scale in analyzing agency among such "agency poor" populations. For example, Rogaly ) analyzed unorganized migrant farmworker agency in India and the United Kingdom and came to a similar characterization of their agency as the pursuit of "sometimes highly significant changes in the microspaces of work and living, albeit in a world dominated by capital." At a different analytical level, these agentive acts ended up as co-constitutive elements in some of the very structural inequalities to which they responded. By maintaining livable conditions without cost or legal penalty to the employer, residents' appropriations of marginal value in their living and working environments further incentivized that neglect and can thus contribute to the long-term, ongoing deterioration of those camps. In this way, the social, economic, and legal marginalization of migrant farmworkers shaped labor camp resident agency into forms that effectively undermined the long-term well-being of residents. Like the actions of "resilience" and "reworking" among the migrant garment workers of Carswell and De Neve's fieldwork, the agency of these workers and their adaptations in putting up with substandard conditions had no structurally transformative effect on their employment: "[such agency] does not challenge the basic status quo between capital and labor within the industry." The tendency of worker agency in highly-stratified employment contexts to have structurally-reinforcing effects is a well-documented phenomenon in economic anthropology. This field of inquiry highlights not just the material effects of agency but also the intentions behind it . There was one instance of explicitly articulated resistance, while the rest of the participants conceived of their actions simply as inevitable responses to obdurate, externally determined conditions: "What can we do?" . Though common , the naturalization and internalization of this suffering and structural inequality was incomplete; denunciations of the poor quality housing were far more common and salient than occasional rationalizations based on identity, nationality, or socioeconomic status. Labor camp residents' agency may reinforce its structural determinants, but these structures were not totalizing for these subjects. Farmworkers and their families articulated the shortcomings, hardship, and unfairness of these circumstances and even alluded to possible futures outside of these conditions, whether envisioned as a torched trailer, a refused ride from Walmart, or a decision not to commence another migrant round. The scarcity of instances of formal resistance suggests that migrant farmworkers felt their agency did not extend into institutionalized mechanisms of resistance and redress. This conclusion is supported by the fact that only sixty-seven postoccupancy inspections were conducted in 2014 among North Carolina farms , despite the presence of over 100,000 migrant farmworkers in the state and documentation of widespread housing violations throughout the state's labor camps . We contend that agency is differentially distributed among migrant farmworkers and their families, often on the basis of gender, employment, legal status, and language ability. Our data suggest that some women in North Carolina labor camps lack a presence in the very field that dictates their housing, an invisibility that inhibits their agency in advocating directly to employers and contractors but centers them in the work of maintaining poor conditions through cleaning. It also seems that the additional protections of the H-2A visa program buttress and act in the stead of agentive acts to improve migrant farmworker housing. And finally, the inability to speak English and communicate directly with most employers critically shapes agency by precluding multiple forms of self-advocacy on farms and labor camps. When put into conversation with prior ethnographic research on other migrant farmworker communities characterized by contrasting forms of housing tenure, by distinct histories of labor migration, and by differing production schemes and labor markets , our results contribute to a "dark anthropology" while also qualifying its more expansive claims. In true dark anthropological fashion, our findings do describe a migrant farmworker agency limited to "putting up with" poor, harmful, and unpleasant conditions, of individually mitigating their worst effects, and possibly reinforcing their everyday reproduction. Nonetheless, this limited agency's roots in local labor markets and housing tenure suggest that the self-defeating or self-exploiting nature of this farmworker agency is not inevitable nor coterminous with all of United States farm work. In contexts characterized by alternate labor market structures, social network formations, and longer histories of migrant settlement and community building, there is indeed evidence of immigrant farm laborer communities' exercising their agency to press new claims upon the public sphere and to "find more than just survival" . That is to say, it is primarily a question of context, rather than of content, that renders North Carolina labor camp residents' agency so futile in transforming their living and working conditions. --- Implications for Policy and Research For those looking to improve labor camp conditions, agency matters. Its current forms and distribution are important in bringing a hard-fought measure of dignity and comfort to the lives of individual residents but are structurally ineffective or reinforcing in their articulation with current institutional frameworks. Even those with H-2A visas, which should provide some protection, find that they have little influence on their housing and work conditions . A policy that provides migrant farmworkers with the opportunity to address housing conditions by reporting poor housing conditions or housing regulation infractions with the support of a third party could help address the limits on their agency. Such a mechanism was provided to farmworkers through the 2016 revision of the United States Environmental Protection Agency Worker Protection Standard . The Worker Protection Standard is a set of pesticide safety procedures. However, the strength of establishment agriculture has led to efforts through the legislative process to limit third party reporting . The vulnerability of migrant farmworkers argues for policy that does not require farmworker-initiated actions with an employer. Several grand level policy initiatives are needed to improve housing, health, and justice among migrant farmworkers. These include immigration reform, as well as amending agricultural workers' exclusions from extant labor protections, including the Occupational Safety Act of 1970, Fair Labor Standards Act, the National Labor Relations Act, and the Social Security Act. Several more specific, and potentially attainable, policy and research recommendations were presented in a recent multidisciplinary symposium on farmworker housing and health that included applied social scientists, public health scientists, health care providers, attorneys, architects, and advocates with expertise in farmworker housing issues . Their policy recommendations reflected needed enforcement, revision, and evaluation of housing regulations. Too often, current farmworker housing regulations are not adequately enforced . Therefore, state and federal agencies responsible for enforcement of current housing regulations should provide adequate funding and staffing for enforcement; educate farmworkers and growers about housing requirements; and train agency personnel, health care providers, and enforcement officials in the health risks of substandard housing. Farmworker housing regulations should be revised to require regular post-occupancy inspection and ensure anonymity of reporting through a third party. Revisions of the Migrant Housing Act are needed that are based on current empirical knowledge about the impact of housing characteristics on health. Definitions of adequate housing in these policies should take a more holistic view of housing needs to include social and community benefits in design decisions. Finally, these policy revisions should create positive incentives for local governments, agencies, and farmers to develop affordable, high-quality farmworker housing. Farmworker housing regulations, whether current or revised, should be evaluated to ensure that they are achieving the desired results. Evaluation should be based on the regular collection of information about the conditions in existing farmworker housing, health implications of the conditions, health care needed and received, and the nature of the work performed by farmworkers residing there. This information should be updated regularly with data that include an assessment of future housing needs. Finally, evaluation should be based on common measures addressing housing and health disparities. Research is needed to document farmworker housing characteristics. The last national survey of farmworker housing was conducted by the Housing Assistance Council in the late 1990s ; therefore, a new nationally valid survey of the current farmworker housing conditions is needed. A consortium of organizations to manage this effort and consult with local researchers addressing farmworker housing and health would enhance such a survey. New research should simultaneously assess housing conditions and the health status of resident farmworkers and their family members to inform farmworker housing policy. Researchers should develop a standard core instrument on housing issues to use in diverse studies conducted with farmworkers. This core instrument should include items on health outcomes directly affected by housing characteristics. Such core questionnaires have been developed for other health-related topics. --- Conclusion The pervasiveness of instances of putting up with substandard housing, employment, and health conditions among those living in migrant farmworker labor camps indicates that the structural barriers to agentive action are multiple and reinforcing. Future research would do well to analyze the specific ways in which agency in the domain of housing is undermined or weakened by other immigration and labor policies, making the structural nexus that migrant workers confront self-limiting and even self-defeating in improving their lives. Furthermore, based on limitations in our study sample, we suggest that future research should more deeply investigate how differences in gender and indigenous identity affect the development of migrant farmworker agency. --- Appendix I Community-based Participatory Approach to Farmworker Housing, Exposures, and Health: In-depth Interview Guide --- 1. Can you tell me about the different kinds of farm work you do in this area / North Carolina? --- 2. Do you migrate to do farm work in other parts of North Carolina? Of the country? --- 3. Tell me about the house in which you live in North Carolina. --- 4. Tell me about the people who live with you in North Carolina. --- 5. What rent do you pay for your housing? --- 6. What are some of the positive features of your house? --- 7. What are some of the problems? --- 8. Have you made any changes in your house? --- 9. How does this house compare to your previous houses?

Each year, tens of thousands of Latino migrant farmworkers return to the hills and fields of North Carolina to live in employer-provided labor camps that often fail to meet regulatory and ethical standards. Little attention has been paid to how these workers and their families perceive and respond to substandard housing conditions. This paper analyzes how far and in what ways Latino migrant farmworkers living in eastern North Carolina labor camps feel their agency extends in responding to substandard housing and work conditions, while also considering who among them has the power to exercise this agency. Based on qualitative analysis of interviews with twentythree migrant farmworkers or partners of farmworkers, our findings indicate that the limits of migrant farmworker agency are strongly dictated by the close relationship between migrant farmworker employment and housing; by the importance of remittance wages for dependents; and by individual immigration, gender, and education characteristics. In the current context, the informal and reparative nature of agentive acts in substandard labor camps creates real but minor improvements while also co-constructing the very structural neglect to which it responds. We conclude that current housing and labor policy frameworks do little to amplify and support differentially distributed migrant farmworker agency and its individualized successes.

Introduction Prison officers represent a key institutional resource. i They perform essential roles in keeping prisons safe, secure and humane. The prison officer pursues one of the most stressful occupations , and the difficulty associated with combining work with family responsibilities-the work-family conflict -is a key precursor to stress . Yet balancing work roles with family roles and responsibilities poses serious difficulties for many officers. The concept of WFC is embedded in the notion that individuals have finite psychological and physiological resources, and that work and family domains are in competition for those resources . WFC negatively affects job satisfaction, organisational commitment, support for offender rehabilitation, turnover intent and voluntary turnover in various prisons and secure institutions . This finding endorses the view that "most roles are greedy consumers" because of their competing demands for finite resources . Participation in one role thus diminishes personal resources, and hinders participation in another role . A further complication of WFC is that stress can be transferred from one role or domain to the other. Little is known about the nature, antecedents and impact of WFC among prison officers. The literature originates in England and North America, which limits our understanding of the scale and nature of WFC among prison officers in other parts of the world. In Africa, and particularly in Ghana, the traditional division of labour in the family has been eroded by the increasing presence of women in all occupations, including prison establishments . The change renders interference between work and family roles inevitable; however, no studies have yet investigated WFC among prison officers in Sub-Saharan Africa. This exclusion is important, because studying WFC from a different socio-cultural and economic standpoint would enable the development of "concepts and generalizations at a level between what is true of all societies and what is true of one society at one point in time and space" . It is also important because "new ideas or policy innovations uncovered in one society may have applicability in others" . The present study is an attempt to address the gap in our knowledge by investigating WFC among prison officers in Ghana. --- Previous research on work-family conflict WFC has evolved from its early conceptualization as a unidirectional construct, with the direction of interference running from work to family , to a bidirectional construct associated with unique domainspecific antecedents . WFC may thus take two forms: work-interference with family and family-interference with work . Greenhaus and Beutell argue that WFC is inherently non-directional and becomes directional only when the individual makes a conscious effort to resolve the conflict. WIF and FIW are thus "distinct but related constructs that form the overall concept of work-family conflict" . Greenhaus and Beutell propose that WFC may take three forms: time-basedin which excessive time-demands in the work domain make participation in family activities difficult; strain-basedin which tension, anxiety, or frustration in one's work frustrate successful participation in family activities; and behaviour-based-in which "specific patterns of inrole behaviour may be incompatible with expectations regarding behaviour in another role". Behaviour-based WFC thus occurs when there is conflict between work and family roles. For instance, Crawley found that as a result of their jobs some prison officers became suspicious and introduced routine and orderliness into their home. Since suspiciousness and routine are expected at work but not at home, this is a typical case of behaviour-based WFC. Time-based WFC has a high prevalence among prison employees. Triplett et al. claim that excessive working hours, frequent overtime and irregular shiftwork contribute to the experience of work-family conflict among prison officers . In the US and Israel, prison officers experience more timeand strain-based WFC than correctional staff with non-custodial roles in education, medical, business, industry, and administration . WFC may also cause work stress. In a medium-security correctional facility in the US, Triplett et al. found that WFC, in its behavioural form, was significantly correlated with work stress. By contrast, Lambert et al found no evidence of behaviour-based WIF and FIW effects on stress. Nor was time-based WFC associated with stress. Only strain-based WIF was significantly related to job stress. However, in a subsequent analysis of the same data, Lambert et al. found that after collapsing the time and strain sub-types into a single variable, WIF was the second most important correlate of work stress. This finding is thus an artefact of the measurement strategy. In a survey of correctional staff, Griffin found that, when operationalized by both WIF and FIW subscales, WFC was positively related to work stress. Also, correctional officers experiencing both WIF and FIW have been found to suffer depression but Shamir andDrory found no relationship between WFC and stress among Israeli prison personnel. Owing to the detrimental nature of WFC on both prison staff and the organisation itself, and in order to understand how WFC affects correctional staff, a small but growing number of studies has focused on the antecedents of WFC, identifying a number of personal characteristics and job-related variables. Triplett et al. found that WFC was a major issue for female staff, who experienced more stress than male staff. Griffin found that WFC was a more significant issue for female staff than for male staff. Lambert et al. found that employees with limited tenure and lower levels of education experienced more time-based WFC. Also, staff in supervisory roles experienced more time-and strain-based WFC . Young prison staff experience more time-based WFC than their older colleagues . On work-related antecedents, Lambert and Hogan found that the perceived dangerousness of prison work and role conflict was positively correlated with WFC while organisational justice was negatively correlated with WFC. Similarly, Obidoa et al. found that, while supervisory support reduced WFC, the perceived psychological demands of the job increased it. A recent replication of the Lambert and Hogan study by Lambert et al. found that the perceived dangerousness of prison work, role conflict and role overload contributed to the overall staff experience of WFC. These studies are not without limitations. There has been an overrepresentation of White prison staff, precluding generalization to other minority ethnic groups such as Blacks and Hispanics study). The study samples have included 'correctional staff', a term embracing both custodial and non-custodial staff, despite differences in tasks and routines. For instance, while custody officers work 24-hour shifts, non-custody staff do not. The empirical literature shows that correctional/prison staff experience higher levels of WFC than non-correctional staff . It may be argued that our understanding of WFC among prison officers as a distinct group of prison staff is limited. Moreover, the small number of empirical studies is mainly English or North American in origin. Without similar studies in African prisons, we cannot know whether African officers experience similar WFC and adopt similar coping strategies. It is far from clear whether policies based on the current literature will work in different contexts. Only through replication studies in under-researched contexts can prison managers be dissuaded from adopting 'best' practice from other countries without first taking local factors into consideration. Conspicuously missing from the literature is an exploration of specifically why, how and when work and family roles interfere with each other among prison officers in other cultures where 'work' and 'family' domains suggest different meanings and have broader and more nuanced implications . The present study attempts to address this challenge by focusing on Ghana. By the nature of the job of prison officers in Ghana, WFC assumes mainly a unidirectional fashion from work to family with very little interference from family to work . Thus, while the family 'domain' or 'boundary' of Ghanaian prison officers is easily permeable, that is not the case with the structure and nature of prison work. The cultural endorsement of work commitment as a sacrifice for the family and a means of family survival via income, security, and prestige provide the enabling environment for WFC to flourish. This therefore intensifies female officers' experience of WFC, compared to their male counterparts. --- The societal context of Ghana Ghana has diverse ethnic groups, each divided into different clans . Ghanaian families are "male-headed units of extended families consisting of one or several wives and their children and often extended with unmarried or elderly relatives" . With high sex segregation in the typical Ghanaian family structure, men are primarily breadwinners while women are tasked with housework and child-rearing. In traditional Ghanaian society, marriage indicates completeness of life. Unmarried adults are treated as second-class citizens. It is Akan practice, during family gatherings, to serve bachelors half-glasses of drinks as a depiction of society's disapproval of the single state, while married men are served with full glasses . The primary aim of marriage in Ghanaian society is procreation, and marriage is the only state in which sexual intercourse is sanctioned . Children are valued in Ghanaian society for their high economic, social, psychological and spiritual worth, as they ensure the continuation of the family lineage. For parents, children are a further source of social security in old age. Care of the elderly, entailing feeding, bathing, changing clothes, remittance, surveillance, healthcare and companionship, and organising funerals is a traditional responsibility for older children. It indicates respect and reciprocity towards elders. The inability of older children to provide such care attracts social disapproval and insults from family members, friends and society in general . Adult children's willingness to provide ongoing care for their elderly parents is not guaranteed but is contingent upon the quality of care those children received from their parents when they were young . In line with cultural norms, women undertake most care of the elderly. By contrast with Western individualism, this collectivist cultural ethos, the endorsement of marriage and procreation as the culturally dominant family arrangement , and the virtual absence of familyfriendly policies in government and private organizations, make it difficult for employees to balance work and family . The result is further gendered, with greater difficulty for women in balancing paid work and family responsibilities . Furthermore, despite weakening traditional patterns of family-related assistance owing partly to migration of kinsfolk and modernity, and with reduced observance of norms and values , cultural expectations and responsibilities remain entrenched. Support for extended family members and care for the elderly as culturally-mandated responsibilities are burdensome for working adults, because of the absence of care homes and limited welfare provisions . Thus, a prison officer need not be married in order to experience WFC in Ghana. Obligations towards the extended family take their toll on every officer's time, attention and energy. --- The work context of the Ghana Prisons Service Since 1993, the female officer contingent in the Ghana Prison Service has increased steadily from 19.9% to 31.7% in 2012. The increase in the female workforce has occurred mainly in the lower ranks. Of the 610 superior officers in 2012, female officers constituted 19.7% , compared to the subordinate officer corps where females constituted 33% . Thus, female officers are vertically and horizontally segregated in the GPS at both managerial and non-managerial levels, and this contributes to their greater experience of WFC-related stressors such as lack of autonomy, lower pay, and fewer incentives and privileges. At higher levels of the Ghana civil service, discrimination against women has its roots in the British colonial administration . It was not until 1967, a decade post-independence, that Ghana promulgated the Labour Decree granting women full employment rights. In the GPS paramilitary management structure, a consequence of its British colonial history, autonomy is key to balancing work and family demands. Owing to their subordinate status, female officers are more likely than their male counterparts to be transferred from one prison to another and to be allocated less living-space. Furthermore, with females in subordinate positions, their incomes are low, making it difficult to obtain help with housework, which is unpaid despite a long-standing discussion about unpaid housework and care work in Africa . By combining prison employment with unpaid household work, female officers appear to work twice as many hours as their male counterparts, and to experience more intense stress. To become eligible for maternity leave, female prison officers and their civil service counterparts are also required to have been in post for at least two years. Breaching this requirement by becoming pregnant during that period incurs repercussions such as suspension without pay and, at worst, dismissal from the service. Nevertheless, there are no paternity leave provisions for male prison officers. --- Methods Mainly qualitative methods, with interviews, observations and a survey, were employed to address the research questions. This ethnographic-led approach, involving sustained periods in prisons, helps with building trust, actively engaging with and participating in officers' working lives and helping to understand individual officers and institutional practices. The interview sample involved 78 officers, although only 59 responded directly to cues and prompts about WFC. The survey comprised a cross-sectional sample, with 1,062 officers submitting analysable data. Male and female officers comprised 65.4% and 34.6% respectively of the interview sample. This corresponds with the gender composition of the GPS, which out of a total of 5,898 officers in 2012 comprised 68.3% male officers and 31.7% female officers . Officers were based at 31 of the 43 establishments in all ten regions of Ghana . The interview sample was also representative of the officer hierarchy of the GPS . Interviews were recorded with the express consent of participants. Subsequently, the recorded interviews were transcribed verbatim, and the transcripts were analysed following a thematic approach . [Insert Table 1.0 about here] In most prisons, surveys were administered and collected after a brief staff presentation by the researcher to all prison officers at a full staff meeting. Efforts were also made to distribute surveys to officers at their various duty posts if they were unable to attend the meetings. Out of 1,490 surveys distributed, 1,117 completed surveys were returned, representing a response rate of 74.9%. However, only 1,062 surveys were analysable, representing 18% of the total officer workforce of the GPS. Male and female officers constituted respectively 65.9% and 34.1% of the survey sample. WFC was measured with six target statements , five of which were adapted from Netemeyer et al. . The sixth item was developed for the present study based on findings from a pilot study. The WFC scale proved to be highly reliable . In addition, the researcher was allowed to freely move around the prison, observe management meetings, officers at their duty posts, and prisoner-staff interactions. The ensuing sections outline the emerging themes from the qualitative data under two broad categories: time-and strain-based WFC. Findings from the survey are also presented. --- Time-based WFC --- Time-based job demands Twenty-four of 59 officers said that their work produced time-based conflicts which made it difficult to discharge their family duties satisfactorily. Officers specified the nature of the job, involving long hours on weekdays, weekends and holidays, and unpaid overtime, as interfering with family life: The work is so tedious and stressful that all your time is for the work. Even if you are on leave, you can be called upon at anytime. You have little or no time for your family. Recall from leave is stipulated in Section 25 of the Ghana Labour Act 651 , which allows "an employer, in case of urgent necessity … [to] require a worker to interrupt his or her leave and return to work". The interpretation of 'urgent necessity' was left to the discretion of the employer. The extended family, an important form of social security among Ghanaians , also suffers from work demands. This affects both married and single officers. One unmarried officer was distressed because: This job greatly interferes with my family. My father is bedridden for 11 years now due to stroke. I can't help. I have to go and help but I can't go due to the nature of this job. I am stuck and watch helplessly. Officers complained about their inability to participate in extended family activities such as out-dooring, ii or wedding and funeral ceremonies at weekends. These activities were important for people to learn of developments in the family, and to strengthen their bonds with the extended family and clan: As I speak to you now, I have lost my younger brother. Of course, I ought to be there but I can't. I have to give excuses in the sense that my job is so demanding that if I leave, what will happen next? Because anything that happens in this prison will affect me. For another senior officer, the time demands of his work interfered not only with performing family duties but also with discharging his traditional leadership role as a chief iii : Yes, it interferes with my family. It even goes beyond my family because, as I am speaking with you now, I am a Chief. But because of the work, I cannot go and perform the customary roles as required of meprison duties before anything else. The prison takes every bit of my time so my duty at this prison does not allow me to perform my customary duties as expected. Female officers, in particular, appeared to be much more distressed by the interference of work with their family duties. The Ghana Living Standards Survey found majority female participation in household chores, although with a gradual, steady percentage increase in men contributing to household tasks in urban settings. Thus, increased female officers' participation in the work domain, and the absence of family-friendly policies in the GPS, significantly hindered their participation in family-related duties. They were allowed only three months' statutory paid maternity leave and an unpaid extension. This provision pales in comparison with the 26-week ordinary maternity leave and a further 26-week additional maternity leave available to prison officers in England and Wales . As a coping strategy, therefore, female officers combined their maternity leave with their annual leave. Although three female officers in three different prisons were seen breastfeeding their babies while at work, they did not bring babies to work: the children were brought in by friends or neighbours to be breastfed during lunch breaks, as the prison barracks were located nearby. These were the only observed instances of family directly interfering with work . It may be argued that breastfeeding did not conflict with the job, since the officers used their legitimate lunch breaks for this purpose; rather, it was a vital coping strategy. Female officers also engaged in informal unpaid care provided by family, friends and neighboursa very common strategy in Ghana and the rest of the world . In the UK, for example, the Office of National Statistics estimated informal childcare to constitute between 19-25% of GDP in 1999 . Despite the high rate of informal unpaid care, it sometimes comes at very great cost to female officers: I was a nursing mother and my baby was only 7 months old. He had just started crawling and I left him in the house to be taken care of by a neighbour. […] When I returned from work, the woman I left the child with was nowhere to be found. A neighbour told me that my child has fallen into a pot of hot water and has been rushed to the hospital [by the informal caregiver]. If I were in the house to take care of my own child, this wouldn't have happened. This was my worst day. Apart from requiring assistance from informal caregivers from neighbours and extended family members for child support, female officers also believed that having understanding and supportive husbands was imperative for balancing work and family. Despite societal disapproval of the single estate, some female officers remained unmarried in order to avoid serious interference between their work and family lives: As for me, the interference is minimal because my husband is the understanding type. But we have had situations where officers had to divorce because of this work. That is why most female officers are not married. More often, it was noted that most female officers were married to members of other security service organisations such as the Fire, Immigration, Police and Customs services. It was widely acknowledged that such officers understood the demands of the job and understood the plight of women working in prisons. Nevertheless, WFC for female officers had a significant impact on marriage and relationships, owing mainly to the dominant Ghanaian traditional cultural role expectations of women. Although unsocial working hours are paid for in allowances, they have also been documented as disruptive of family life among prison officers in England and the US . However, women officers in Ghana face a cultural imperative in the form of guilt and extended family and clan affiliation risks associated with their inability to fulfil kinship demands such as elderly care and participation in the family members' rites of passage, owing to work demands and unpaid overtime. --- Lack of autonomy Job autonomy is the degree of freedom, independence and discretion that employees have in accomplishing their tasks or making job-related decisions . It helps employees to maintain a balance between work and family demands and to reduce work-stress . Prison officers claimed that, owing to lack of autonomy over their job demands and schedules, they were often restrained from participating in family activities as they wished. They reported that they often required permission to leave work in order to spend time with their families but were often turned down with the excuse of 'security first'. This was the case for male and female officers, and for junior and senior officers. Thus, officers lacked control over their ability to decide when, how, and over what periods to work, or to leave work when it was necessary to attend to family duties. This was stressful for these officers who felt disempowered or handicapped in making meaningful contributions to their family lives: In this job we don't have time for our families. I am here as the regional commander and anywhere I want to go, I have to seek permission from headquarters before I move and this is a bit worrying at times. My husband has always been complaining that 'it seems you are married to the work' and you don't have time to come home because there is always something to be done. So I can say confidently that this work is taking too much of my time. Officers relied on their superiors for permission but owing to staff shortages and other prison priorities such as security, even their entitlement to days off was denied. Similarly, despite officers' entitlement to 30 working days paid annual leave, how and when this statutory provision is to be taken or enjoyed is subject to the approval of superior prison officers. This was stressful and frustrating for the officers. The inconvenience was not limited to their participation in the nuclear family alone; it also affected the extended family during periods of crisis. Kaufman observed similar autonomy-related issues among her sample of prison officers, who were unable to negotiate their shift patterns and days off. As a result, officers' ability to socialize with their family and friends outside working hours was impaired. Moreover, since female officers predominantly occupied mainly subordinate positions they often lacked the autonomy to make flexibility decisions for their work schedules, although this strain was experienced by both men and women: You cannot always be with your family because at the time they need you, you must be at work. I remember that early this year, my grandfather who was so very kind to me died. I asked permission to go and it was denied. My uncle went to explain things to the prison boss in Kumasi Central that, as a result of the situation, the family wanted me to be there and I was still not allowed permission to go. I wept bitterly here while at work and it was a very difficult situation. . Although officers complained of sickness, it was difficult to get time off for rest and recuperation. They had to come to the prison in person and submit a doctor's note to a Governor before being granted time off work. If this rule were to be applied in England, absence by prison officers who feel entitled to sick leave would fall significantly. In 2012-13, the national average rate for sickness absence among prison staff in England and Wales stood at 10.5 days per person . High sickness levels were significantly related to staff attendance warnings and poor quality of life . --- Strain-based WFC --- Accommodation-based conflict Narratives of stress, and themes of frustration, fatigue, and sometimes anxiety dominated officers' accounts of their accommodation. Many officers described the barracks accommodation as 'small', 'tiny', 'cubicle', 'not fit for purpose' and 'overcrowded', which impaired their ability to contribute meaningfully to their family lives. This is mainly because barracks accommodation are built mainly to accommodate single officers and at best small nuclear families of a husband, wife and a child. This official expectation is at variance with the cultural preferences and pressures for marriage and large families associated with social approval and spiritual blessings. This reflects in the high fertility rates for Ghana compared to its western counterparts. iv The accommodation situation is thus, a strong source of strain and frustration for prison officers who feel a sense of 'collective squeeze' in their residential living space. Fourteen of 59 prison officers made this point: Our place of residence is very appalling. We are living in cubicle rooms with our families. Sometimes a family of seven will be sharing just a single room, which may be compared to someone's bathroom. Yet, it houses all these people. I would prefer that we get better accommodation because a sound mind thrives in a sound body. It is a common feature at the barracks to see officers' valuables such as freezers left outside their rooms at the mercy of the weather and at night, furniture rearranged with some brought outside to make room for sleeping spaces. Officers argued that they were unable to participate in family activities given the small size of their accommodation, to the extent that sex with one's partner is sometimes scarce. At a meeting at the Nsawam prison between the Prison Officers' Wives Association v and the Director of Prisons in charge of welfare, the prison officers' wives complained that, due to the size of their rooms, their husbands denied them prolonged sexual encounters for fear that a child might be observing secretly or that a child might wake up suddenly to use the bathroom. This situation, the women argued, starved them of sex and made child upbringing and discipline difficult . In officers' residential accommodation, amenities such as kitchens, bathrooms and toilets were also shared among families in the barracks. This was another area of tension, conflict and quarrel among some families residing in prison barracks. vi Furthermore, directives from management that only nuclear family members should occupy official accommodation precluded prison officers from receiving support from extended family relations or from house-helps who were readily available and willing to assist with household chores. Such directives particularly disadvantage female officers who have to combine unpaid housework without assistance with prison work: Our rules are also such that you cannot even get anybody to stay with you and help you with household chores. Even if the person comes, where would the person sleep? So this job makes combining work with family life difficult. Since extended families are a strong feature of Ghanaian social life owing to the social, cultural and psychological security it provides based mainly on the belief of collective responsibility , there is the expectation that those who travel to the city to find employment will offer support to their kin. This often involves the expectation of accommodating extended family members who travel to the city to search for jobs. Refusal by officers to accommodate their kin who often see the barracks as a transition zone to city life attracts opprobrium and social rejection. Thus officers are often caught between official prohibition to accommodate visitors and kin, and cultural pressures to fulfil such demands. Officers' prison barracks accommodation can best be understood through the notion of carceral collectivisma system of penal governance involving peer surveillance among prisoners housed en-masse in dormitories in post USSR penal establishments . Through a system of 'mutual peer surveillance', some form of carceral collectivism via the spatial and temporal structuring of the life of prison officers is manifested in officers' barracks. Through this mechanism, the GPS instils certain formal values and norms -such as respect for hierarchy and power, discipline, camaraderie, encouraging long working hours and unpaid overtime, observance of barracks rules and regulations, flexibility for easier and effective deployment -which tend to be incompatible with family values. It is not far fetched to think that some officers 'snitch' on colleagues who violate barracks rules such as accommodating extended family members and make life difficult for other officers, especially female officers who are often desperate for assistance with household chores post-partum. Thus, informal mechanisms through gossips and disapproval are employed to control life in the barracks. It is important to note that this form of mutual surveillance is not alien to Ghanaian culture. Indeed social control in 'traditional' Ghana was founded on mutual surveillance arising from the belief in collective responsibility for social order . --- Deployment-based conflict Officers noted that frequent transfers from one prison to another, and especially between different regions, disrupted family lives. They complained that frequent changes of school and location inhibited child development: Yes, this job interferes with my family. I was transferred from my previous station to this place. I realized that moving my family to this place would adversely affect the family and myself so I decided to leave them there. So the effect is that we are absent from each other. The children are missing me. We talk on phone and it's always 'daddy when are you coming?' . On transfer, officers had to move to their new station immediately, leaving their families behind. When, owing to the unavailability of staff accommodation, a senior male officer was put up in a hotel at his new station, he relied on female officers for his meals while his family still lived at his previous station in a different region. At weekends, he drove home to spend time with his family while they were waiting for official staff accommodation at his new posting. Other prison officers complained about their frustration with transfers: At times when you are transferred, you have to leave your family and depart for your new station immediately in case you are a senior officer. As a result, you would have to live as a single man in your new station for a very long time before your family is finally conveyed to your new station. Wives and female officers who had engaged in some form of trade or business to supplement the family's income also found their businesses disrupted by transfer: The most difficult experience in this job occurred when I persistently pleaded with my [Governor] and the transport unit at the Prisons headquarters to give me two weeks for my wife to dispose of her fermented corn dough for kenkey and fish at the barracks. This was turned down. Simply put, I should pack my belongings after a day's notice to my new station. Transfers were not merely distressing to an officer's family; they also impacted on spouse's incomes, and their living and business arrangements suffered significantly for a while, futher adding to the strain of relationships. On the whole, subordinate officers were transferred frequently from one prison to the other. With many female officers occupying subordinate officer positions, deployment-based conflict was more burdensome. One in five officers identified this as a problem. --- Work-family facilitation Despite the pervasiveness of WFC among prison officers in Ghana, three male officers and one female officer argued that the job made family life easier. Work improved overall quality of life and provided an income and other benefits which enabled them to discharge their primary responsibilities as breadwinners, and so their families had to adjust to their job demands: There is not much interference between my work and family. That is my daily work and that is what I do to fend for the family. That is where I earn my income. As for me, it has helped me to take care of my children because it is the salary that I use to pay my children's school fees. This female officer was the breadwinner as her husband had lost his job. Her work did not conflict with her family role, as she had effectively exchanged places with her husband. That notwithstanding, officers' assertions provide further evidence that participation in paid employment is construed as a sacrifice for the family rather than a sacrifice of the family, as is the case in other collectivist cultures of Asia . --- Prevalence of work-family conflict among prison officers [Insert Table 2.0 about here] One of this study's main aims was to assess the prevalence of work-family conflict among officers. As shown in Table 2.0, the prevalence of WFC among officers was high. A minimum of 53.6% and maximum of 85% of responding officers endorsed the statements identifying WFC. The 85% who agreed with statement 6, that "I often miss important family or social activities [e.g. spending time with the family, out-doorings, funerals, etc] because of my job", shows this item as the most frequently-experienced form of WFC. It further indicates the importance of such social activities in the socio-cultural lives of Ghanaian prison officers. Their inability to take part in such activities owing to job demands was frustrating. The above-neutral mean scores and narrow standard deviation values for all the WFC statements indicate a high prevalence of WFC. The survey findings are consistent with interviewees' accounts of WFC . --- Conclusion This study sought to explore WFC, its prevalence and manifestations among prison officers in Ghana. It examined contextual influences of WFC in a country where the interface between 'work' and 'family' is construed differently from the West. It was found that officers' conceptualization of WFC was mainly unidirectional, with the direction of influence from the work domain to the family, i.e. WIF. Employment demands stretched officers' resources, making it impossible for them to integrate work and family responsibilities. The finding in Ghana contrasts with the prevailing bi-directional thesis, informed mainly by quantitative research from the West, which compels respondents to make a distinction between WFC and FWC . This situation is reminiscent of the police organizations that "commit members to the occupational organization, to the occupational community, and to its norms of subordination and service to a degree where these commitments take precedence over extra-occupational ones to family and community" . The GPS is thus equivalent to Coser's "greedy institution", which makes "total claims" on officers for their "exclusive and undivided loyalty" in a quest to "reduce the claims of competing roles and status positions on those they wish to encompass within their boundaries". In being a greedy institution, it mostly affects female officers. The study found that, owing mainly to heavy demands of the unpaid domestic duties imposed by the dominant traditional or patriarchal gender roles entrenched in Ghanaian society, female officers experience more intense stress associated with WFC. The minority status of female officers, experienced at both lower and higher ranks in the masculinized organizational environment of the GPS, combined with unfavourable working policies and gender-specific societal expectations of marriage and large family size, overburden female officers and strain their ability to cope. The study found that while WFC permeates cultural boundaries its nature differs. WFC as experienced by prison officers in Ghana was mainly attributable to the absence of family-friendly policies in the GPS, beyond the constitutionallymandated maternity leave for female officers. Although taking part in extended family activities was stressful, it was also the sole support for officers adapting to competing work and family demands. There was no social support from the employer, exposing officers' family lives to interference through time-based work demands including both lack of autonomy and strain-based conflicts arising from inadequate residential accommodation and deployment-based interruptions. Gender was an important moderator of WFC. In the qualitative data, in line with previous research, female officers reported more intense stress than their male counterparts experienced from combining work and family duties . Again, the traditional gender roles emphasised by Ghana's collectivist ethos partly accounts for this phenomenon. This study is not without limitations. The mainly purposive and convenience sampling techniques employed for the interview and survey samples respectively do not permit the findings to be generalised. Despite the importance of qualitative methods in illuminating work-family conflict, a cause-effect relationship could not be established. Thus a longitudinal survey, as opposed to the cross-sectional method employed here, would help to identify the role demands within work and family domains that predict WFC. The paper could have explored a number of interesting themes that have not been addressed here , but those were not the primary focus of study, as the data were collected as part of a larger project on prison officer stress and well-being in Ghana . However, the study lays a foundation for more in-depth research on WFC and how it could be alleviated. It might be worth investigating how the experience of WFC in the GPS is embedded in the history of prison work, prison practices and politics by comparison with the Ghana Police Service, which appears to share similar organisational structures, work policies, employment practices and work-and family-role demands. The current data did not permit an investigation of WFC from a more gendered perspective, which is insufficiently studied and lacking empirical scrutiny in the developing world. Future enquiries might benefit from examining this gendered experience of WFC in greater depth. Despite these limitations, this paper shows that a predominantly collectivist culture held together via social activities such as birth, marriage and death, as well as the traditional conceptualization of the role of work as sacrifice for the family, makes WFC a widespread occurrence in the GPS. The paper has made an empirical and theoretical contribution to the literature on WFC among prison officers by showing that "work and family issues are intricately related to cultural beliefs, values and norms" . It has also revealed a spatial dimension of WFC --- literature on WFC has yet to embrace this concept, which merits further exploration and development in various prison and national contexts.

This paper documents the experience of work-family conflict (WFC) among prison officers in Ghana. Although the term WFC has been used in relation to prison officers in the UK (Crawley, 2002) and the US (Triplett et al., 1999), the context of WFC in Ghana is unusual. In this predominantly collectivist culture, family responsibilities include obligations to the extended family. WFC is mainly unidirectional, with interference running from work to the family. Officers are thus impaired in fulfilling their family responsibilities, which consequently impairs their daily work and mental well-being. The 'crisis controlling' or 'paramilitary' organisational structure of the Ghana Prisons Service (GPS) makes it very difficult for the work domain of prison officers to accommodate family responsibilities. Female officers appear to bear a heavier WFC burden than male officers, mainly on account of their traditionally unpaid housekeeping role in addition to their paid employment in a masculine organisational culture. The findings are significant, as they show that the promulgation of family-friendly policies to alleviate WFC-associated stress lies in the hands of the GPS, since WFC emanates solely from the work domain.

INTRODUCTION According to the United Nations Program on HIV and AIDS in 2018, statistical data obtained about 37.9 million people in the world living with HIV with the number of AIDSrelated deaths around 770,000 . 000) cases and there are about 1.7 million new cases of HIV infection worldwide. 1 In Southeast Asia there are 3.8 million people living with HIV as measured by all ages, followed by 170,000 new cases of HIV infection . 2,3 Based on the Ministry of Health of the Republic of Indonesia, from October to December, 46,659 people were infected with HIV and there were 10,190 people with AIDS. The presentation of HIV cases based on gender was 29,287 in the male and 16,872 in the female. The number of reported HIV cases in North Sumatra province has increased every year, there were 1,891 cases in 2016, 1,914 cases in 2017 and 1,999 cases in 2018. Based on data from the Ministry of Health of the Republic of Indonesia, there were 160 cases of HIV infection and 279 cases of AIDS. The number of people with HIV/AIDS in Deli Serdang Regency has increased every year. 4 Stigma is the main concept in sociological deviance, by labeling someone as a sign that the person deviates both morally, membership in a group or physically. 5 Stigma associated with HIV, can be interpreted as a negative feeling of prejudice, attitudes and beliefs towards PLWHA and people who have a high risk of HIV infection or are called HIV key populations. 6 Stigma against PLWHA can trigger acts of discrimination from various sectors of society, such as the living environment, work environment, including the school environment. Stigma and discrimination that occurs can make people living with HIV and key populations reluctant to access health services, such as knowing HIV prevention methods, checking HIV status, getting treatment and following treatment. 7 A study conducted by --- RESEARCH METHODS This research uses an analytical cross-sectional study. The sample of this study amounted to 80 samples from four public junior high schools in Lubuk Pakam District. The sampling technique used was total sampling. The inclusion criteria in this study were teachers who were willing to be respondents and the exclusion criteria were teachers who were unable to attend when conducting the research. Data was collected using an instrument in the form of a questionnaire to teachers who teach from the Lubuk Pakam sub-district. This questionnaire has been validated, contains 45 questions to assess knowledge, attitudes, and stigma towards HIV. This instrument uses 35 questions to assess the respondent's knowledge of HIV/AIDS using a two-point Guttman scale and 5 questions each to assess attitudes and stigma towards HIV/AIDS using a Likert scale with a choice of statements whose assessment points are determined based on favorable or unfavorable statement items, for answers with favorable statements, strongly agree, get the highest points, namely 4 points, 3 points for agreeing answers, 2 points for disagreeing answers, and 1 point for strongly disagreeing answers, while the points for unfavorable statements are the opposite. The independent variable in this study is the level of knowledge and attitudes about HIV/AIDS with the dependent variable in this study is the stigma of HIV/AIDS. Univariate analysis was used to describe the characteristics of the sample and the characteristics of the variables studied in the study. Bivariate analysis was used to analyze the relationship between the dependent variable and the independent variable. Bivariate analysis test is Chi Square test and Fisher's Exact test. --- RESULTS AND DISCUSSIONS The description of the variable frequency distribution can be seen in table 1. The largest number of respondents in this study were women, as many as 64 respondents, the age group of respondents was mostly at the age of 51-60 years as many as 30 and the least was at age category 21 -30 years as many as 5 . Most of the respondents' formal education was in the Bachelor's level formal education category, which was 68 respondents and the least respondents were in the Diploma level formal education category as many as 5 respondents. The most respondents have a level of knowledge in the good category as many as 53 respondents , with the most attitudes being in the has low stigma category, as many as 53 respondents . The results of the statistical test in table 2 shows that there is no relationship between the level of teacher knowledge about HIV/AIDS and the stigma of HIV/AIDS based on the Chi Square Test, the p value of 0.400 between the level of knowledge with stigmas. The relation between attitudes and stigma levels based on The Fisher's exact test, obtained a p value of 1,000 which means no there is a relationship between the level of teacher knowledge about HIV/AIDS and the stigma of HIV/AIDS. In this study, out of 80 respondents, 53 respondents had a good level of knowledge about HIV/AIDS, and 27 respondents had a poor level of knowledge, based on the table above, it can be seen that 85% of respondents have the highest level of education at the bachelor's level. The respondents' knowledge of HIV/AIDS was obtained from health officers or service providers, health education about HIV/AIDS and social media. The results of this study are in line with research conducted by Simanjuntak NH, of 49 respondents who are academics and are educators at the HKBP Nommensen University Medan, as many as 43 respondents have a good level of knowledge about HIV/AIDS and 6 have a good level of knowledge about HIV/AIDS. 12.2%) respondents have a low level of knowledge. 9 from these results it can be seen that there are more teachers who have good knowledge than educators who have low knowledge. This can be caused by the education level of respondents who are already high, most of them are has bachelor degree. There are 74 respondents who have a good attitude and there are 6 respondents who have a bad attitude. Based on research conducted by Nirindah R and Prastiwi W, on AIDS-aware Citizens in Surakarta City from 99 respondents, there are 77 respondents who have a good attitude and there are 22 respondents who have a bad attitude. 10 Most respondents have low stigma, as many as 51 respondents and 29 respondents have high stigma. Another study conducted by Lubis L, et al about "The Relationship of Stigma, Depression, and Fatigue with Quality of Life of HIV/AIDS Patients at the Medan Veterans Clinic" stated that there is a relationship between stigma and quality of life and there is a significant correlation between the two variables with a strong correlation strength and a negative correlation direction. This shows that the higher the stigma felt by PLWHA, the lower the quality of life of PLWHA. 11 --- Table 1. Variable Frequency Distribution In this study, there was no relationship between the level of teacher knowledge about HIV/AIDS and the stigma of HIV/AIDS. This is in line with research conducted by Simanjuntak NH, which states that there is no relationship between the level of knowledge and stigma. Other research results obtained by Finnajakh A, and study conducted by Nirindah R and Prastiwi W, states the same result. 9,10,12 --- Table 2. Relationship between Knowledge Level and Attitude with Stigma --- CONCLUSION There is no relationship between the level of teacher knowledge about HIV/AIDS with stigma against HIV/AIDS and there is no relationship between teacher attitudes and stigma towards HIV/AIDS. --- REFERENCES

Stigma is an attribute either socially or physically that highly discredits a person, thus the person has a bad identity in social interaction in society. Stigma and discrimination are the main barriers to prevention, treatment, care, and support for people with HIV / AIDS. This study aims to determine whether there is a relationship between the level of knowledge and attitudes of teachers about HIV / AIDS with the stigma against HIV / AIDS in Lubuk Pakam District Junior High School. This research was an analytic study with cross sectional design. The population in this study were the teachers of Lubuk Pakam District Junior High School, Samples were selected using a total sampling method and 80 teachers were assessed using a questionnaire to measure the level of knowledge, attitudes and stigma of the respondents. Data collection methods in this study used primary data and data were analyzed univariately and bivariately. The results showed 66.3% of respondents had a good level of knowledge about HIV / AIDS, 92.5% of respondents had a good attitude towards HIV / AIDS and 63.7% had a good stigma about PLWHA. The results of data that was analysed using chi square test found that there was no relationship between the level of knowledge with the stigma of HIV / AIDS with p value = 0.400 (p> 0.05) and the results of data which was analysed using the fisher exact test, obtained p value = 1,000 (0.05) which meaning there was no relationship between attitude and stigma against HIV / AIDS. T here was no relationship between the level of knowledge and attitudes of teachers about HIV / AIDS with the stigma against HIV / AIDS in Lubuk Pakam District Junior High School.

Introduction There has been a marked shift in policy concerned with obesity in recent years from a focus on individual treatments towards a greater emphasis on public health interventions . In the UK this change in focus is illustrated by the 2004 House of Commons Health Committee report on Obesity which concluded that, "While we accept that individuals have the right and responsibility to make choices about their own health and lifestyle, we accept the importance of health education in enabling them to do so, we believe that to tackle obesity successfully education must be supported by a wider range of measures designed to remove the key barriers to choosing a healthy diet" . The Government's Foresight Programme was subsequently asked to identify the nature of these barriers, and their report, Tackling Obesities: Future Choices, highlighted the importance of the total 'obesogenic environment' , defined as the 'sum of the influences that the surroundings, opportunities or conditions of life have on promoting obesity in individuals and populations' . The term embraces the entire range of social, cultural and environmental conditions that influence an individual's ability to adopt a healthy lifestyle. These might include factors within the environment which shape the availability and consumption of particular foods or the levels of physical activity undertaken, as well as a range of social variables. People exist in a number of different environments including physical, social, and cultural micro environments such as school home and workplaces, and they are also influenced by macro level environments such as education and health systems, and related government policy . Exactly how these different factors and environments interact and contribute to obesity is still poorly understood and some have argued that environmental effects are relatively small in comparison to genetic influences . Despite this continuing debate about the relative influence of different causal factors, 'obesogenic environments' are widely accepted as a driving force behind the escalating levels of obesity , and the concept has gained considerable prominence in both the research and policy communities . The increasing availability and consumption of food outside the home, and in particular fast food, is thought to be one of the key factors within the obesogenic environment contributing to rising levels of obesity. Fast food tends to be more energy dense and has a higher fat content than meals prepared at home , and the frequency with which it is consumed has been shown to be linked to increased body weight and obesity . Consumption of fast food has been growing as lifestyle trends in the UK tend to support a 'convenience food culture '. In the UK one in six meals is now consumed outside the home , fuelling an increase in fast food outlets . Not surprisingly therefore, the Foresight Report identified "Controlling the availability of/exposure to obesogenic foods and drinks" as one the top five policy responses likely to have the greatest impact on levels of obesity. . Until recently the multi-national fast food chains, such as McDonalds, have received most of the blame for popularising fast food , and have been a key target of government initiatives designed to encourage healthier fast food options. Whilst the major chains dominate in terms of market share, the fast food sector as a whole is characterised by a greater number of small independent businesses. In many areas these independent outlets form a significant proportion of the local food offering and are growing in number, and by 2008 accounted for 46.8% of the market . The proliferation of fast food outlets in local neighbourhoods, particularly in close proximity to schools, has led to calls to curb their expansion . In 2008 the government's Healthy Weight Healthy Lives strategy drew attention to the large number of fast food outlets operating in local neighbourhoods noting that, "One of the challenges that we face in promoting healthy eating is the availability of foods high in fat, salt and sugar in local neighbourhoods, including the prevalence of fast food restaurants and takeaways in some communities" . It recommended that local authorities should," use existing planning powers to control more carefully the number and location of fast food outlets in their local areas." . This policy is now being actively considered by a number of local authorities with two in London having developed specific supplementary planning guidance, . However such policies are rarely based on a detailed assessment of the local area and its communities. This paper aims to demonstrate the importance of such an analysis. The paper draws on a case study of the independent fast food sector in the London Borough of Tower Hamlets designed to develop a more detailed understanding of the nature of the local fast food environment, and to identify the factors within the local neighbourhood that might be contributing to the proliferation of fast food outlets and fast food consumption. Previous research which has explored potential neighbourhood determinants of obesity has been largely quantitative in nature and has sought to determine a statistically significant relationship between just one or two risk factors rather than develop a comprehensive model which incorporates the range of influential variables . By adopting a more qualitative approach this study aims to highlight the wider range of issues involved, including, in particular, the influence of local cultural and religious values on provision and consumption patterns. In so doing it illustrates the importance of micro-level analysis as a means of understanding the local sensitivities which may affect a particular fast food environment. Given this detailed local analysis the paper sets out to question the extent to which planning proposals designed to curb the proliferation of fast food outlets are likely to contribute to a reduction in the risk of obesity in areas such as Tower Hamlets. Tower Hamlets provides an ideal setting for such a study. It is home to some of the most deprived communities, with the highest levels of childhood obesity in the UK , and reportedly has one of the greatest densities of fast food outlets in the country . Tower Hamlet's role as one of the host boroughs for the 2012 Olympics adds additional impetus to the focus on encouraging healthier lifestyles. These factors contributed to the borough receiving £4m in November 2008 under the Department of Heath's new 'Healthy Towns' initiative. This funding, awarded to 9 areas in England, is to support initiatives, designed to "test and evaluate different approaches to making regular physical activity and healthy food choices easier for local communities". In Tower Hamlets part of this grant is being used to fund the introduction of a healthy food award scheme for those fast food outlets able to demonstrate that they provide a range of healthy foods. This paper draws on research commissioned by the Tower Hamlets Primary Care Trust in 2009 to inform the development of this scheme The paper is structured as follows: The first section explores the debates around food environments in deprived urban areas. The second section sets the context by providing some background information on Tower Hamlets and its demographic and socio-economic characteristics. It then moves on to discuss the study and its findings. These highlight the crucial interplay of socio-economic and cultural issues which have influenced the development of the local fast food environment and which need to be clearly understood when developing initiatives designed to tackle obesity. In so doing the paper suggests that the current policy focus on tackling the physical aspects of the obesogenic environment tends to neglect many of the underlying causes and economic inequalities which tend to lead to obesity in the first place. --- Food environments in deprived urban areas A key factor which has shaped the policy debate is the tendency for obesity to be more prevalent within poorer communities. The link between obesity and deprivation was most recently highlighted in the Marmot Strategic Review of Health Inequalities in England Post 2010 which noted that, "Obesity is associated with social and economic deprivation across all age ranges and is becoming increasingly common" . In attempting to understand what it is about the environments in which poorer communities live that may make them more obesogenic, attention has been drawn to the nature of the neighbourhood food environment. A number of studies have explored the potential link between deprivation, health, and access to healthy food for a review of the evidence), and in the 1990s the concept of 'food deserts' became a popular policy concept to describe areas which lacked access to healthy food . Initial evidence seemed to confirm the existence of food deserts and suggested that consumption of fruit and vegetables increased amongst those with the poorest diets when access to healthier food improved, in particular through the provision of a new supermarket within easy walking distance . Recent US research tends to support this finding. For example Moore et al in a study of multi-ethnic populations found that in neighbourhoods with poor access to supermarkets people were less likely to have healthier diets. In the UK and New Zealand the evidence seems less clear cut with better access to supermarkets not being found to be significantly related to increased fruit and vegetable consumption . However differences in the design and quality of different studies make comparisons difficult suggesting that this counterargument has yet to be proved. More recently attention has turned to the increasing prevalence of 'unhealthy food', and in particular fast food, and the extent to which this is contributing to rising levels of obesity, particularly within poorer communities and amongst children. Numerous studies have sought to ascertain whether fast food is more readily available in deprived areas and may thus be contributing to the heightened levels of obesity found within deprived communities . The results here are conflicting with several researchers finding a greater density of fast food outlets in more deprived areas and others suggesting that outlets are more likely to be concentrated in the more prosperous commercial and business districts and along arterial roads rather than in the poorer residential areas , or in the case of one US study, in both deprived and more affluent black neighbourhoods . It has also been suggested that the greater level of demand for fast food in deprived areas might be due to the higher population densities generally found in these locations, and the greater number of outlets due to fast food businesses choosing to locate in areas with lower land prices and building rental costs . Furthermore it is not clear whether living near fast-food restaurants actually results in greater levels of obesity. In the US studies have shown that the close proximity of fast food outlets to schools leads to greater fast food consumption and higher levels of obesity amongst children , but research elsewhere has found little support for the concept that exposure to fast food outlets in the local neighbourhood increases risk of obesity . Others have suggested that the socio-economic characteristics of individuals and their households have a greater influence over purchasing behaviour than the nature of the takeaway food environment . One Australian study found that the type of fast food chosen was significantly different across different socioeconomic groups; whilst the better educated group had greater takeaway consumption the lowest educated group were more likely to have consumed "less-healthy" takeaway choices . Similarly in the UK the heaviest users of chicken and burger restaurants are less affluent consumers in socio economic groups C2 and D . Given the conflicting nature of much of the evidence, and the complex mix of variables involved, it would appear that a more detailed understanding of the spatial and socio-economic characteristics of local areas and fast food consumers is necessary to fully appreciate the causal relationships operating in different localities. This was a key objective of this study of Tower Hamlets. Research and policy to date has also been predominantly concerned with the physical and economic access to food, whilst more complex social and cultural influences which tend to influence choice, opinion or satisfaction, have been largely overlooked . Access to affordable healthy food that meets with cultural or religious dietary requirements may be a key issue in deprived urban areas with large ethnic minority populations. Independent outlets may feature more frequently in such neighbourhoods for a number of reasons. Urban neglect, high levels of crime, perceived or real poor market demand and poverty has often meant that major multiples have by-passed or deserted these areas. This has resulted in a dearth of enterprise in some areas and has led to a raft government initiatives designed to stimulate business growth as well as campaigns to encourage business investment and retail-led regeneration to redress this imbalance . In some cases the tightening of planning restrictions in more attractive locations has forced major retailers to reconsider policies towards such areas. Lowe and Wrigley cite the example of Tesco's decision to locate in under-served markets as part of a deliberate strategy for gaining entry to the US market. In many inner city areas, however, these voids have often been filled by local independent ethnic minority firms, who attracted by low rents, and a plentiful supply of cheap, ethnic labour, are able to set up businesses catering to the specific needs of the local community. A weak retail climate and a surplus of low-wage labour, both make the proliferation of fast food outlets probable in such environments . --- Tower Hamlets: Setting the Scene The London Borough of Tower Hamlets lies on the northern banks of the river Thames adjacent to the City of London . With an area of just 21.5 sq km and an estimated population of 234,800 the borough is one of the smallest and the 5th most densely populated boroughs in the UK . It also has one of the youngest populations with 24% of the population under 20 years of age . Data from the last census suggested that 33.4% of the borough's population were of Bangladeshi origin . This community accounts for over 25% of the entire UK Bangladeshi population and as such is the most concentrated ethnic minority community in the UK . The strong presence of the Bangladeshi community is particularly evident in the north of the borough. The area is home to Brick Lane, known as the 'curry capital of the UK', with around 50 Asian restaurants now serving a largely white clientele of city workers and tourists . It is also the location of a growing night-time economy and a hub for the creative industries. Nearby on Whitechapel High Street a thriving street market sells cheap food and clothes to local residents, whilst the East London Mosque, the largest mosque in Europe, provides a focus for the borough's 80,000 Muslims as well as a large number of non-local visitors from outside the area . --- Fig 1. East London Mosque on the Whitechapel High Road Tower Hamlets is also an area of stark contrasts. Its proximity to the City has attracted an increasing number of professionals who have brought and restored many of the old Georgian terraces in the borough or moved into new apartments in the re-developed docklands area. But these pockets of gentrification lie adjacent to some of the most deprived wards in the country. Over 50% of the borough's population live in areas which are amongst the 10% most deprived in the UK . In November 2009 20.3% of residents of working age were in receipt of some sort of out of work benefitover 4% more than the average for London . Poor levels of proficiency in English amongst the first generation, coupled with poor educational achievements, a limited network of contacts in different sectors, and discrimination, have impacted on the Bangladeshi community's employment prospects . Although Bangladeshi children have seen rapid advancement in recent years the community as a whole still has one of the lowest levels of educational attainment and the highest rates of unemployment in the UK. In 2009 20.5% of 16-24 year olds had no qualifications . Unsurprisingly this high level of deprivation is coupled with equally high levels of poor health and nutrition. Tower Hamlets has one of the worst childhood obesity rates in England. Results from the National Child Measurement Programme, for 2008-9, show that 13.4% of 4-5 year olds, and 25.69% of 10-11 year old children, were obese . The borough ranks 5th highest in London for prevalence in 4-5 year olds, and 2nd highest in London, , for obesity prevalence in 10-11 year olds . Data for adult obesity is less readily available but public health records suggest that one in 5 adults are currently likely to be obese . In 2008 Tower Hamlets launched its own local Healthy Weight, Healthy Lives strategy which set out plans for a multi-agency approach towards tackling the continuing rise in overweight and obesity . The subsequent award of Healthy Towns funding, later in 2008, provided an opportunity to supplement this work and trial some new approaches. This includes a Healthy Food strand which aims to "improve access to healthier food options in the local community and organisations, and where possible, reduce the availability of foods that are high in fat, sugar and salt." . It incorporates the Food4Health Award scheme which is being led by the local authority's environmental health team in conjunction with the THPCT's public health dietician, and which is inviting local food businesses to apply for a Bronze, Silver or Gold award depending on the extent to which they can demonstrate that their menus include a healthy range of food options. Free nutrition and/or business advice is provided to applicants working towards the award. Businesses that achieve the award receive a certificate to display and free local publicity that it is hoped will encourage customers to opt for healthier food choices. In tandem with this initiative the THPCT staff are also working with the borough's planning department to explore how planning policy can be used to control the proliferation of fast food outlets. --- Methods The data for the case study was collected using a range of methods. The local authority's Environmental Health Department provided access to the Food Register which lists all premises offering food to the public and this enabled some quantitative data analysis of the fast food businesses by type to be undertaken, and for the location of business premises to be mapped using GIS mapping software. Face to face interviews were conducted with 30 fast food businesses and 51 of their customers. Resource constraints limited the number of interviews that could be undertaken, but the business sample was designed to be broadly representative of the range of independent fast food outlets by type of food sold and location across the borough. The interviews were conducted on the business premises during trading hours and were based on a semi-structured questionnaire. This sought to gather demographic data on the owners, their motivation for entering the fast food sector, the nature of food sold by the business, and attitudes and barriers towards the sale of healthier food. They were also asked to identify the principle demographic characteristics of their customers. Healthier food was defined here as including dishes incorporating items such as fruit, vegetables, salad, bread or rice instead of chips, and fresh fruit juices, as well as healthier cooking methods including the roasting or grilling of meat instead of deep fat frying. The customer survey was undertaken by approaching customers eating in or purchasing food for takeaway consumption from a sample of the outlets chosen for the business interviews. This research was conducted at lunchtime and during the early evening to allow for the range of customer types to be covered. It sought to provide a snapshot of the type of customers frequenting the outlets , the nature and frequency of their fast food consumption, average spend, and attitudes towards healthier menus. In line with other research which has found fast food consumption to be greatest amongst young men our sample comprised of 74.5% males, 96% under 40 years of age, 67% under 25 years of age, and 43% school or college students. These interviews were supplemented by a number of field visits to strategically chosen locations around the borough, where the outlets were known to be concentrated, to observe the businesses, their customers and the local area. These visits were designed to develop a deeper understanding of the nature of the fast food environment and local consumption patterns. This field work was carried out over a two month period in the summer of 2009. In addition semi-structured interviews were held with key personnel from the local agencies working with the businesses. These included three key staff from the Tower Hamlets Primary Care Trust, the local authority environmental health officer, the borough planning officer responsible for developing local planning policy regarding the fast food sector, a business advisor from the local business advice agency, and the main supplier to the local fast food industry. The data collected during these interviews provided additional information on the nature of the food being sold, business support, regulation practice, and policy developments. The case study approach clearly limits the transferability of the findings, but did allow for a more detailed qualitative analysis of local influences on business development and fast food consumption which is frequently absent from larger scale surveys. --- Fast food businesses in Tower Hamlets: an important source of local ethnic minority employment The Local Authority's Food Register lists a total of 296 businesses which are registered as takeaways, of which 10 were found to be branches of national chains. However inspection of the list of businesses registered as restaurants/cafes/canteens revealed that this contained at least 80 businesses whose sales were known to also include the sale of takeaway food. This included, for example all 5 of the McDonalds branches in the borough as well as the 3 Kentucky Fried Chicken outlets, where although substantial seating is provided, takeaway sales are also offered. Of these 80 businesses 29 were identified as chain businesses. Thus out of a total of 376 businesses known to be selling takeaway food 39 were part of chains and the remaining 337 were independent outlets . This would appear to be a much greater proportion of independent outlets than the national average of 46.8% reported by the Local Data Company . Thus national initiatives being undertaken by the Food Standards Agency with the major fast food chains are likely to have a limited impact on boroughs such as Tower Hamlets where independent outlets dominate the local offering. In common with the UK catering industry as a whole, the independent fast food sector in Tower Hamlets is an important source of employment for local ethnic minority communities, with the local Bangladeshi community in particular making up a large proportion of owners and staff. The last Census reported that 46.86% of self-employed Bangladeshis in the UK were operating in the hotel and restaurant sector . The sample of 30 businesses interviewed for this study included businesses whose owners were Bangladeshi, Chinese, Cypriot, French, Iranian, Jamaican, Lebanese, Pakistani and Turkish. Their businesses included pizza outlets, fish and chip shops, Chinese takeaways, outlets selling Caribbean, Lebanese and Indian food, kebabs, and a large number offering primarily chicken and chips or burgers and chips, sometimes with a few curry dishes as well. Seventeen of the 30 owners interviewed were of Bangladeshi origin reflecting the high proportion of Bangladeshis both resident in the area and engaged in the catering trade. The interviews with owner managers revealed that most had entered the fast food sector because they had previous experience of the industry. For those with limited educational qualifications, or contacts beyond their own community, working for catering establishments owned by family and friends was often the only available source of employment. Catering skills were usually acquired by working as employees in restaurants or other fast food outlets, until they and their family amassed enough capital to start their own business. Businesses other than catering had generally not been considered by the interview sample as they had little or no work experience of other sectors. Basic fast food outlets require limited skills and a capital investment in equipment of only £15,000 to £20,000 and thus entry thresholds are low. The cash based nature of the business is a further attraction allowing for declared turnover to be kept below the VAT threshold. In common with the ethnic minority catering sector in general there is a heavy reliance on informal immigrant labour willing to work flexible hours for minimal wages -a practice described by businesses in other studies of the ethnic minority catering trade as "an absolute necessity for competitive survival" in what is an increasingly crowded and competitive marketplace. A typical outlet employed 3-4 staff including the owner manager, and on this basis it is estimated that fast food outlets in the borough are probably providing over 1,000 local jobsnot an insignificant source of local employmentalbeit poorly paid. --- Fast food outlets: Providing culturally acceptable food and a social meeting place The independent fast food sector in Tower Hamlets, particularly those businesses clustered around the mosque and close to the more deprived and predominantly Bangladeshi residential areas, has clearly adapted its offering to suit the local community's tastes. Many outlets trade under the brand name Perfect Fried Chicken; we counted 23 outlets on the food register with this name and a further 36 with similar names such as Halal Fried Chicken, Docklands Fried Chicken, Limehouse Fried Chicken and Spice, London Fried Chicken, Shah Jalal Fried Chicken, and Fab Fried Chicken. These businesses are not franchises, but their name has come to symbolise a particular type of fried chickensimilar to the familiar Kentucky Fried Chicken variety, but with meat sourced from halal suppliers and cooked to a spicier recipe . In the wake of Islamic revivalism, halal markets have been expanding on a global scale, and London has emerged as a centre for halal production, trade and consumption. . In Tower Hamlets, with its high Muslim population, this is particularly evident. A large number of outlets clearly use their halal status as a key marketing tool, displaying their halal certification notices prominently on widows and menus . This growth in demand for halal food has not gone unnoticed by the fast food chains, several of which have been piloting halal only menus in some of their premises. According to the McDonalds sponsored blog, Make up your own mind, McDonalds ran a short pilot scheme in September 2009 selling halal food at an outlet in Southall, west London, but then dropped the idea concluding that it lacked the necessary infrastructure in its supply chain to guarantee the requirements of halal . KFC and Subway are currently offering halal-only menus at a limited number of their restaurants and takeaways . However as the local independent fast food owners explained, "Muslims have confidence that Muslim owned outlets can be trusted to serve halal meats that have been slaughtered according to religious laws" . This trust rarely extends to takeaways not owned by Muslims or businesses who attempt to sell both halal and non-halal food, even if not from the same premises. This powerful relationship of trust has been shown to influence the shopping practices of Muslim migrants elsewhere. . Ahmed , for example, noted that Muslims prefer to shop at small shops run by Muslims rather than at supermarkets and that this practice is changing the shape of many urban areas as a result as this specialist provision replaces other more mainstream retail outlets. In a similar fashion independent fast food outlets offering halal food now dominate a number of Tower Hamlet's high streets. Fast food was also described by business owners and their customers as providing a welcome alternative to the type of food normally eaten at home. "They mainly buy fried chicken because they can't cook it at home." A fast food meal was therefore something a bit different -a bit of a treat -albeit a regular one for some customers. In addition to providing an alternative to traditional home cooking it has been suggested that fast food may also be performing an additional function by providing young people with a medium for expressing a youthful self and/or lifestyle image . Notably Ram et al found that the younger generation of Asians were more likely to be found in fast food restaurants than curry houses. Thus by serving halal fast food the outlets allow Muslim young people to consume food that is culturally acceptable whilst also being able to express their identity as young westernised British Asians. The fast food businesses in Tower Hamlets also perform an important social function within the community. The majority of outlets, whether they are registered as takeaways or restaurants/cafes, have a small seating area and thus provide a useful meeting place, particularly for young men and women. This role is particularly important in areas such as Whitechapel, Stepney and Mile End where the businesses are catering primarily for the local Muslim community who are expected to abstain from drinking alcohol and are therefore less likely to use pubs, clubs, or restaurants as a social meeting space. Almost 25% of the customers interviewed suggested that the main reason they had decided to buy fast food from the outlet they were in was, "because friends were going." Whilst from discussions with owners and observation of consumption patterns it was noted that after midday Friday prayers at the Mosque the outlets are particularly busy as worshippers stop to chat with family and friends over a lunch of fried chicken and chips or a more traditional biryani. --- Fast Food: Affordability and Access to Healthy Fast Food Alternatives For those on low incomes the fast food outlets provide an opportunity for a quick and tasty hot meal on the go or an affordable venue for 'dining out'. A large plate of chicken and chips, or burger and chips, can be brought in most outlets in the more disadvantaged areas of the borough around Whitechapel, Stepney and Mile End, for just £2.49. Those interviewed spent an average of £2-3 on their meal. Even cheaper deals were targeted directly at children. £1 buys two chicken wings with a generous portion of chips and several were observed purchasing a 50p snack of two chicken wings. Price was also cited as the key factor influencing the purchasing decision of the customers interviewed. Thirty-four percent reported that the cheap price of the food was the main reason for buying fast food from the outlet they were in. The impact of the economic recession on the UK's dining out habits has been widely reported , and was very evident here as the owner of one Indian takeaway, who had been trading for over 15 years, explained, "Previously there was more income from Asian food but now there is more from fried chicken and chips. A year ago I changed to also offering fried chicken and chips. There is less profit from chicken and chipsapproximately 30% compared with 50% on an Asian meal, but now people can't afford it. Customers who used to buy an Indian takeaway once a week have cut back on this treat to just once a month." Thus the sale of these unhealthy fried food items was seen as crucial for business survival. Owners were very conscious of the highly competitive and price sensitive nature of the market and felt unable to pass on recent price rises in food supplies to the customer. "Profit margins are decliningthe chicken comes from Holland and euro exchange rate is poor". "The chicken price has gone up in the last couple of months but I can't pass it on to the customer" Several businesses highlighted the competitive nature of the environment by commenting that, "There is too much competition." As a result most of the owners felt that the only viable business strategy was to keep prices low by using cheap ingredients including poor quality meat and the cheapest of vegetable oils, high in saturated fats. They were often well aware of the unhealthy nature of the food they were selling. Indeed one owner reported that, "I buy vegetable oil for the business but I use rapeseed oil at home" Lack of access to affordable healthier alternatives that met with religious and cultural dietary requirements was also clearly an issue for a number of customers. Two in particular commented that, "Healthy food is not halal", whilst fifty-six percent claimed that they would have brought healthier food if more were available, either from the outlet they were in or others in the area. Clearly some respondents may have responded in this way because they felt this was the more acceptable answer and further research is needed to ascertain to what extent this was a factor here. The fast food business owners had no objection to selling a healthier range of food, in principal, although many cited barriers including the need for equipment such as grills, juicing machines and salad cabinets, or lack of space or staff with the necessary skills. "I have no space to grill chicken. It takes more time -most people don't want to wait" "I would need more staff -can't do too may varieties" The main barrier, however, was perceived to be cost. Healthier foodfor example grilled chicken, cost considerably more, and many owners felt that customers would not be prepared to pay the higher prices involved. As one explained, "We do roast chicken but it is not as popular because it's more expensive." There was a notable absence of 'healthier' fast food operators in the poorer parts of the borough. GIS mapping of fried chicken outlets, those known to be selling 'healthier' alternatives, and levels of deprivation show that the 'healthier' outlets tend to be located close to the City of London and the newly developed commercial centres in Docklands, in particular Canary Warf, where wealthier working commuters are found . Average spend at these 'healthier' outlets is closer to £5double the price of a chicken and chips meal. Notably in the more affluent commercial centre at Canary Warf, in the docklands area of the borough, fast food outlets were specifically prohibited from selling fried foods under the terms of their lease in an attempt to maintain a more prosperous image for the development and higher rental levels . The corporate companies controlling this type of large scale development contrast sharply with the myriad of individual landlords leasing to outlets in less affluent areas of the borough. The GIS mapping also illustrated how fast food outlets tended to be particularly clustered along the main arterial and commercial routes, and close to key venues such as the Mosque and local college . A high concentration was also noted at the northern end of Brick Lane in Spitalfieldsan area which has become an important cultural and tourist attraction. Most of the customers interviewed were well aware that the food they were eating was not particularly healthy. Seventy percent claimed to be concerned about the health implications of eating a lot of fried food or food with high sugar or salt content, yet 45% of these were eating fried chicken and chips or a burger and chips when interviewed. The interviews suggested that eating habits were not necessarily due to lack of information regarding healthy diets, but were governed by affordability and choice. This supports other research that has shown that viewing unhealthy food as a treat places the choice of such food in a positive light, thereby justifying consumption . Thus purchasing decisions are dependent not only on factors such as the relative accessibility of fast food but a more complex mix of factors including cultural and religious influences, price, taste, and personal choice. , 2007;Food Register, London Borough of Tower Hamlets, 2009) 8. Influencing and 'Controlling' the fast food environment: Viable Interventions Given the range of potential influences on fast food provision and consumption attempting to 'control' fast food outlets within the obesogenic environment in areas such as Tower Hamlets presents a particular challenge. UK government policy advocating the use of the planning system to control the increasing proliferation of outlets, particularly around schools, is attracting a significant amount of interest. The scope for doing this was made possible by an amendment in 2005 to the Town and Country Planning Use Classes Order which created a new class -A 5 for Hot Food Takeaway shops. The distinction between this and the A3 class: Restaurants and Cafes is based on the proportion of space designated for food preparation and the number of tables or chairs. The change was introduced largely in response to concerns regarding environmental rather than health issues, with the planning guidance explaining that, "Takeaways are differentiated from restaurants because they raise different environmental issues such as litter, longer opening house, extra traffic and pedestrian activity, from those generally raised by A3 Restaurants and cafe uses" . The change does however provide local authorities with the power to limit the growth in the number of outlets as both new outlets and restaurants converting to fast food outlets need planning permission. The London Borough of Waltham Forest has been the first London borough to introduce supplementary planning guidance on hot food takeaways. This seeks to restrict the number of new units that would be allowed in certain key areas, including banning new units within 400m of the boundary of an existing school, youth centre or park. Other boroughs are beginning to follow suit with similar bans and some such as the London Borough of Barking and Dagenham are proposing to charge a levy if planning approval is granted for takeaway use; the fee to be used to fund measures designed to target childhood obesity . In a recent landmark legal ruling, planning permission granted for a takeaway in Tower Hamlets was overturned by the High Court. The judge ruled that the outlet was too close to a secondary school and that the local authority had failed to take account of the health and well being of the local communityparticularly school children, in considering the planning application . Such measures may be effective in restricting the opening of new outlets, but do little to limit the sale of fast food from existing outlets or indeed the many restaurants who also sell unhealthy fast food. In Tower Hamlets the small geographical size of the borough coupled with its high population density and large number of existing fast food outlets means that every resident is no more than a 10 minute walk away from an outlet . In these circumstances most children will pass at least one on their journey to and from school. Furthermore, as a member of the planning department pointed out, it is very difficult to monitor the sale of fast food by restaurants; as long as they have a table and chair it is almost impossible to enforce exclusive A3 use . Monitoring also requires resources which are not available and it is only if complaints are received from the public that action may be taken . National campaigns to encourage healthier menus are also much harder to promote in areas such as Tower Hamlets where independent outlets dominate the fast food landscape. Current proposals for food labelling, designed to provide customers with information on the food content and calorific value of menu items, are much easier for major chains offering standardised menus and with in-house nutritionists, to comply with. Introducing and monitoring such initiatives in the independent sector where menus may look very similar but in practice are unique to each outlet is a much greater challenge. As a result these national initiatives are largely bypassing the areas that most need them. The UK Food Standards Agency has recognised this problem and is introducing guidance sheets targeted at specific types of smaller catering businesses which are designed to encourage salt and saturated fat reduction. However effective promotion requires, as the FSA itself has found, a significant amount of face to face work with businesses . In the short-term initiatives designed to encourage fast food outlets in more deprived areas to sell a healthier range of food need to identify strategies for encouraging outlet managers to modify their menus in ways that will not impact on costs and result in higher prices for the customer. In Tower Hamlets this has meant workshops and outreach work to persuade fast food owners to fry food in healthier oil at higher temperatures, sell fatter chips which absorb less fat, shake the fat off chips coming from the fryer, grill rather than frying food where possible, and introduce a wider range of salads and fresh juices. These are relatively small-scale changes, but changes that will, if widely adopted, have a significant impact on local health. For example it has been suggested that the use of fatter cut chips and better frying practices such as the triple shake' techniqueencouraging three shakes of chips coming from the fryercould result in up to a 20% reduction in the fat content of the chips . --- Conclusions This paper has identified key local influences within one deprived inner city area which are contributing to the proliferation of fast food outlets and encouraging fast food consumption. The research findings suggest that a wide range of variables are at play leading to a more complex pattern of causal relationships than many previous studies have suggested. Like other studies it has highlighted the potential link between deprivation, the spatial concentration of fast food outlets and the price of food sold. However it has also shown that in areas such as Tower Hamlets local economic disadvantage can fuel not just a demand for cheap fast food. It also contributes to business start-up in the sector as those with limited qualifications or contacts have few alternative employment options. The spatial distribution of outlets and the type of food sold is also influenced by the nature of local or passing trade as well as the availability of suitable premises, the ownership structure of the businesses and the characteristics of the local property market. Thus in a mixed inner city area such as Tower Hamlets healthier and often more expensive fast food options are found in more affluent residential or commercial areas or along commuting routes, whilst cheaper less healthy fast food tends to be concentrated in more deprived areas. The study has also shown that cultural, social and religious norms, as well as notions of identity, also play a key role in influencing business start-up and consumption patterns. For some communities these norms may serve to define or restrict choice. Valentine notes for example, how the Muslim community's culture of abstention shapes its members access to and use of space with many leisure facilities being off limits to the community because they serve alcohol. The findings from this study suggest that in Tower Hamlets the independent halal fast food outlets have stepped in to fill this gap in provision and are providing the local community with an affordable and culturally acceptable meeting space and opportunity for dining out, accounting for much of the growth in the provision and popularity of fast food outlets in areas where the Muslim community is concentrated. These issues are all likely to influence the viability of intervention strategies designed to combat obesity. In Tower Hamlets the increasingly popular policy approach, advised by central government, of restricting the opening of new outlets in certain areas may be of limited effect. It may help to reinforce the public health message that fast food is bad for you, but in many deprived urban areas such as Tower Hamlets this type of intervention is inappropriate, the area is already saturated with existing outlets and the small geographical size of the borough means that these are accessible to all. Thus whilst not for one moment suggesting that such a policy should not be considered, it also needs to be recognised that its impact on obesity reduction is likely to be limited. Furthermore it is not necessarily hot food takeaways , which are the main culprits here, as many restaurants and cafes registered as A3 are serving equally unhealthy fried food. Such policies also ignore the fact that not all fast food outlets sell unhealthy food. It is not fast food per se that is the problem, but the type of food sold. There is no planning use class order for 'healthy' as opposed to 'unhealthy' food establishments. Current lifestyle patterns suggest that demand for takeaway food is likely to go on rising. Thus rather than focusing primarily on banning outlets more emphasis needs to be placed on promoting a healthier variety of takeaway foods that meets different dietary requirements. At the same time it needs to be recognised that whilst such measures may have some impact on local communities, along with the host of other initiatives now on offer in many deprived areas, such as encouraging local communities to grow their own food, learn to cook healthy menus, and buy more fruit and vegetables, these fail to tackle the structural inequalities that are the underlying cause of deprivation. There is a tendency, as Herrick noted in her study of Austin, Texas , for intervention designed to tackle obesity to result in the continued neglect of the basic underlying causes and structural inequalities which lead to obesity in the first place. Only by addressing the structural disadvantages, and increasing educational and employment opportunities which offer alternatives to catering, will this vicious cycle be broken in areas such as Tower Hamlets. Banning fast food outlets is not likely to impact significantly on health inequalities, and without ensuring that alternative business or employment opportunities are available for their owners and employees, it may actually contribute to increasing levels of deprivation. The case of Tower Hamlets may not be typical, and more extensive research across different communities throughout the capital and other areas is needed. However this unique case study does serve to illustrate the complex mix of variables involved in understanding the way in which fast food outlets contribute to the 'obesogenic environment'. In this case culture and religious values were found to play a particular role in mediating wider socio-economic factors affecting the growth and popularity of fast food in the area. Elsewhere other influences may be of greater importance in the overall mix. Thus the study particularly demonstrates the importance of local micro-level analysis as a means of understanding the nature of local fast food provision and consumption. Only though such detailed analysis can a more comprehensive appreciation of these different contributory factors be developed and a hopefully more viable and holistic approach to intervention be considered.

In the battle to curb rising levels of obesity the focus has moved from individual interventions towards addressing factors within the 'obesogenic environment.' Fast food outlets are argued to be an increasingly important aspect of this environment and recent UK government policy has suggested that local authorities should use their planning powers to restrict the opening of new fast food outlets in local neighbourhoods. This policy has a particular impact on the growing number of independent fast food outlets that characterise deprived inner city areas. Focusing on a case study of the London Borough of Tower Hamlets and its independent fast food sector this paper explores the range of local variables, and in particular the crucial mix of socio-economic and cultural factors, which have influenced its development. The role played by fast food outlets in providing local employment and a culturally acceptable social space for the local Muslim community, through the provision of halal food in an alcohol-free environment, is particularly highlighted. In uncovering these particular local sensitivities the paper aims to contribute to debates regarding the nature of 'obesogenic environments' and demonstrate that local level research is essential for enabling better understanding for developing policy interventions.

Introduction As a result of the growing aging population, there has been an increase in the prevalence of chronic medical conditions . In lieu of system-level difficulties in managing the chronic conditions of this population, the emergency department has taken on a significant role in providing care for this population [1]. Older adults account for 14.4% of the 20 million ED visits that occur each year in the US [2]. Based on the current rate of growth in healthcare utilization, by 2050 it is expected that the US health system would not be able to meet the demands of the older population who would need ED [3]. Therefore, there is a need for additional information that can potentially reduce unnecessary ED utilization in populations with CMCs. Emergency Departments deliver important healthcare services and commonly serve as the point of entry to the hospital system or as a means of connecting patients to after-hours care, particularly those without a primary care provider . ED utilization can also be a pathway to long-term care settings. While EDs will always play a vital role, ED over-utilization remains a concern and could be reduced through better downstream management of chronic conditions. Therefore, further investigation of the health status of older adults who utilize the ED is warranted and may yield insight into the factors that contribute to ED utilization and hospitalization [4]. The delivery of care to the growing aging population is a pressing challenge facing the United States health care system [1]. Technological innovations, healthcare provider incentives, higher inpatient costs, and managed care policies have contributed to the shift in decreasing inpatient hospital admissions, leading to increased ED visits. Older adults with multiple comorbidities, cognitive and functional impairments, and polypharmacy may present to the ED with symptoms and signs of an acute physical and/or mental illness, requiring rapid triage and diagnosis [5][6][7][8]. Providing high quality and safe care for underserved older adults with multiple and complex CMCs remains challenging for ED providers. Older adults often experience adverse events after an ED visit [9]. Low neighborhood income and high comorbidity status are contributing risk factors for in-patient hospitalization within 30 days of an ED visit among older adults [10]. ED visits are more common for older adults, minorities, and uninsured individuals [11]. The Nationwide Inpatient Sample , a large, nationally-representative dataset of hospital discharges, shows that the odds of unscheduled ED-to-hospital admission is 39% higher for African Americans compared to their white counterparts [11]. The ED is responsible for four out of five unscheduled hospitalizations. Furthermore, in the last decade, unscheduled hospitalization in the United States has grown in complexity, with more individuals presenting in the ED with a variety of acute clinical conditions. Traditional approaches to assessing healthcare delivery have focused mainly on primary care services, delivery of care through patient-centered homes, managed care, and accountable care organizations [12], but the issue of ED utilization has gone relatively unexamined. The healthcare system is overwhelmed with the challenges of providing care for older adults, specifically AAs who suffer from poorer health outcomes. The 2014 National Hospital Ambulatory Medical Care Survey determined that AA older adults had higher ED visit rates than their white counterparts, including higher rates of non-urgent visits, thus indicating a trend of ED over-utilization by this population [2]. Multiple studies have established that AAs have higher ED visit rates than other ethnicities [13,14]. Interestingly, AAs who have a usual source of care or commercial health maintenance organization coverage are more likely to use the ED compared to other people [15]. In addition, compared to whites, AAs are more likely to select the ED as their usual place for healthcare [16,17]. These disparities have been attributed to a number of factors; however, the underlying reasons remain poorly understood [18]. One study found that low income AAs preferred the ED over ambulatory care because it was more accessible, less expensive, and provided a higher quality of care [19]. Disparities within the ED exist for AAs who are less likely to receive non-invasive cardiac diagnostic tests for chest pain [20], adequate pain management [21], post-ED follow-up care for psychiatric and behavioral disorders including substance use [22], as well as cancer care [23]. --- Theoretical Framework This study uses the Andersen's Behavioral Model of Health Services Use [24] to examine ED visits among older AAs. The model has been used extensively to examine the use of health services by various populations, including minority groups [25][26][27][28][29][30][31]. The aim of this model is to discover potential determinants that either facilitate or impede the utilization of health services among various populations by observing the interactions of relevant features of health care, including individual, social, and contextual factors [32]. Specifically, the model examines the inequities of healthcare access and has the potential to inform health-service policy and delivery. Initially developed in the 1960s, the model has undergone five phases of transformation, with a specific focus on contextual and individual determinants. The major components of both contextual and individual determinants are, first, predisposing characteristics; second, enabling factors; and, third, perceived and evaluated need as potential predictors for health service utilization. The fourth component is health-seeking behavior, which is comprised of individual health practices including health service use [24,33]. Predisposing factors encompass demographic and socio-cultural characteristics, including personal health beliefs, social structure, and political perspectives [34]. Enabling factors include decision-making factors to use a healthcare resource, incorporating community and organizational values, and system/structural elements [35]. Need factors are subdivided into perceived need and evaluated need [36]. Perceived need refers to the patient's own sense of need for health services, including self-judged severity of health and illness symptoms. Evaluated need refers to objective measures of the patient's medical and healthcare needs as determined by healthcare providers or other professionals [25]. Both health and functional status can be measured within the perceived need and evaluated need constructs. Incidence and progression of CMCs are both impacted by the quantity and quality of health service utilization [37]. Comprehensive, multilevel assessment of individual and contextual differences in health-services utilization on the part of older AAs remains an urgent need. --- Aims Using Andersen's Behavioral Model of Health Services Use [24], we conducted this study to explore social, behavioral, and health factors that are associated with ED utilization on the part of underserved populations. We were particularly interested in populations with multiple risk factors such as older age, economic disadvantage, and minority racial status. As a result, we studied determinants of ED use among underserved AA older adults in one of the most economically disadvantaged urban areas in south Los Angeles, California-Service Planning Area 6 . --- Materials and Methods --- Design and Setting This cross-sectional study included a convenience sample of 609 non-institutionalized underserved AA older adults, aged 65 years and older. Participants were recruited from 11 senior housing units, 16 predominantly AA churches, and one public housing project located in SPA6 in Los Angeles County. --- Participants and Samling Participants were eligible if they were AA, 65 years or older, and were able to complete an interview in English. Participation in any other clinical trials, being institutionalized in a health care setting, and considerable cognitive impairment were exclusion criteria. --- Comparability of Our Sample Our participants were comparable to AA older adults in South LA. About 35% of our sample had a high school diploma. Similar to our sample, in the California Health Interview Survey data, 37% of AAs aged 65 years or older in South Los Angeles had a high school diploma [38]. Regarding the health status of the participants, one third of the participants described their self-rated health as fair or poor . Similar to our data, CHIS data has shown that 38% of AA older adults living in South Los Angeles report their SRH as fair or poor [38]. --- Measurements The study employed face-to-face structured interviews. The survey toolkit was a collection of several validated measures from various sources [26,[39][40][41][42][43]. --- Enabling Factors Financial Difficulty. Financial difficulty was measured using a five-item measure with items that were on a five-point Likert scale [44]. Participants were asked in the last 12 months how frequently they were unable to: buy the amount of food their family should have; buy the clothes they feel their family should have; pay their rent/mortgage; pay their monthly bills; and make ends meet. A higher score was indicative of less financial difficulty within the last 12 months . Continuity of Medical Care. Continuity of medical care was measured using three items. Participants were asked to report what type of place they usually visit to receive medical care ; whether they usually go to the same place for medical care; and whether they are usually seen by the same health provider when they receive medical care. The answer to the second and third items were yes = 1, no = 0, and the response to the first question was coded private office = 1 versus any other place = 0. The total score ranged from 0-3, where a higher score indicated more continuity of care. Accessibility of Medical Care. Accessibility of medical care was measured using three items. Participants were asked how difficult it is for them to "visit a doctor when you need medical care"; "get a routine physical examination if you wanted one"; and "travel to medical appointments". A high score indicates less difficulty. Satisfaction with Medical Care. Satisfaction with Medical Care was measured using three items. They were asked "Overall, how satisfied are you with the medical care you currently receive; how available medical care is for you; and your access to preventative services . A high score indicates less satisfaction. Demographic Factors. Age and gender were the demographic factors. --- Perceived and Evaluated Need for Care Characteristics Intensity of Pain. Pain intensity was captured using the Short-Form McGill Pain Questionnaire 2 [41]. In structured interviews, participants reported the extent to which they experienced each type of pain. Overall, 22 pain items measured experience of bodily pain in the past week. All items were on an 11-point scale . The SF-MPQ-2 provides a total pain score that is an average of all questions [41]. A high score is indicative of greater pain intensity . We categorized level of pain based on the tertile of the variable . Depressive Symptoms. We used the Geriatric Depression Scale to measure severity and frequency of depressive symptoms. This measure uses 15 items that are on a "yes"/ "no" response scale [43]. The scale provides a total score which varies between 0 and 15, where a higher score indicates more depressive symptomatology. The GDS-SF has excellent reliability and validity. It is widely used to measure depressive symptoms in older adults in the community, as well as acute and long-term care settings [45]. Self-Rated Health . This study measured SRH by the following single question: "In general, would you say your health is Excellent; Very good; Good; Fair; and Poor?" This single item has been frequently used in health research, national surveys, and longitudinal cohorts [42]. --- Chronic Medical Conditions . Number of CMCs was measured by asking whether participants have been diagnosed with the following conditions: asthma or bronchitis; arthritis; high blood pressure; heart problems; --- Outcome Emergency Department Visits. Participants were asked how many times they had utilized ED in the last 12 months. Responses were coded as 0, 1, or 2+. --- Statistical Analysis Univariate, bivariate, and multivariable statistical methods were applied using the Statistical Package for Social Sciences version 22 . We reported frequencies and means for univariate analysis. At bivariate level, chi-square test, independent samples t-test, and analysis of variance were employed to explore correlation between ED service utilizations and predisposing; enabling; and need-for-care characteristics. In addition, multinomial logistic regressions were applied to test the association between independent variables and ED utilization. We used the "Enter" rather than the "Forward" or "Backward" method. Thus, independent variables were kept in the model even if they were not statistically significant. Our independent variables were selected based on an extensive literature review informed by our theoretical framework. Andersen's Behavioral Model of Health Services Use mainly focuses on: predisposing; enabling; and need-for-care characteristics. --- Ethics The current investigation was approved by the Institutional Review Board at the Charles R. Drew University of Medicine and Science . Written informed consent was received from all the participants. Data were kept confidential at all stages of research. --- Results --- Sample Table 1 provides a descriptive view of the sample. The current analysis included 609 AA individuals who were 65 years or older . About 35% of the participants were male. About 16% of the sample was currently married or lived with a partner/significant other. Table 2 shows the CMCs in our sample. The overall number of CMCs ranged between 1 and 10 . Thirty-four percent of participants reported being diagnosed with diabetes mellitus and 58% reported having chronic back pain. Almost one out of four participants reported that they suffer from asthma or chronic bronchitis. Three out of four older AAs received care from two or more physicians. In addition, 68% reported that they accessed medical care at a private doctor's office or private medical group. --- Emergency Department Visits About 41% of our participants were treated at an ED during the last 12 months prior to the interviews. Approximately 27% and 14% of participants were treated once and at least twice in the ED during the last 12 months, respectively . More than 19% of men and 11% of women were treated at least twice in the last 12 months in the ED. In addition, 50% and 25% with 6+ CMC received care at the ED once or at least twice within last 12 months, respectively . --- Bivariate Correlates of ED Utilization Tables 1 and 2 report bivariate correlations between independent variables and ED utilization. Table 1 indicates that gender, continuity and accessibility of medical care, number of CMCs, SRH, intensity of pain, and depressive symptoms all are significantly associated with ED utilization. Among various CMCs, asthma, heart disease, depression, stroke, migraine headache, and GI-related problems were positively associated with ED utilization . --- Multivariate Correlates of ED Utilization Table 3 shows the results of multinomial logistic regression models on the effects of predisposing; enabling; and need-for-care characteristics on ED utilization. The table shows odds ratios and 95% confidence intervals for each independent variable. Table 3 indicates that gender is the only predisposing variable that is associated with ED utilization. Compared to women, men had 2.0 and 2.2 times less odds of "no ED" or "one ED visit" as opposed to two or more ED visits during the last 12 months. Controlling for predisposing variables, the enabling factor of "continuity of care" was found to be a significant correlate of ED utilization. Respondents who indicated a lower level of continuity of care had 1.55 higher odds of reporting at least two ED visits compared to no ED visits within the last 12 months. After controlling for predisposing and enabling characteristics, the multinomial logistic regression showed two of the need-for-care variables as significant predictors of ED utilization: number of CMCs and severity of pain. Depressive symptoms and SRH were not significant in multivariate analysis, even though they were significant in bivariate analysis. Respondents with a higher number of CMCs had 2.61 and 1.87 higher odds of two ED and one ED visits within the last 12 months, compared to participants with one or no CMC. Table 3 indicates that after controlling for other predisposing, enabling, and need-for-care factors, pain intensity was associated with ED visit. Respondents in first or second tertile of pain intensity had 2.8 and 1.76 lower odds of having no ED visit or only one ED visit compared to those who were in the highest tertile of pain intensity. Odds of having at least two ED visits increased 2.42 times among participants who were in the highest tertile of pain intensity. --- Discussion We found that 27% and 14% of older AAs in SPA6, one of the most economically disadvantaged areas in South Los Angeles, had at least one or two ED visits within a one-year period. Previous studies have found that AAs are more likely than whites to have ED visits [14,[46][47][48]. Indeed, AAs have two-times higher odds of having the ED as their usual healthcare source than their white counterparts [16]. Furthermore, frequent ED use is associated with poor quality of life [49]. Frequent ED use disrupts the continuum of care among older adults [50]. As a result, frequent ED users are at an increased risk of subsequent hospitalization, readmission, nursing home placement, and death [50]. In our study, the multivariate analyses introduced a range of predisposing, enabling, and need-for-care determinants of ED utilization. First, AA men were more likely to visit the ED frequently than AA women. One hypothesis is that men are less likely to engage in preventive and continuity of care than women, leading to increased visits to the ED. A considerable literature has shown the central role of gender in shaping attitudes and behaviors related to health care use. For example, a study documented that compared to AA men, AA women have lower rates of early diagnostic cardiac testing in the ED when reporting chest pain [20]. Furthermore, AA women with private insurance have fewer inpatient hospitalizations and ED visits, compared to those with public insurance [51]. This suggests that women with public insurance may have a greater financial burden and may perceive less benefit in a visit to the ED, leading to delays in accessing care. Culturally, older AA women are identified as the matriarchs of their families and usually provide care and significant resources to their immediate families and extended relatives. AA women may be burdened with so many responsibilities that their health takes a "back seat" to others in the family. With regard to enabling factors, this study found that continuity of care was associated with fewer ED visits among this sample of older AAs. One recent study showed that patients who live in medically underserved areas have a higher incidence of preventable ED utilization [52]. Other studies have documented that low versus high continuity of care is associated with greater risk of frequent ED utilization [50]. Additionally, it has been established that AAs have a longer wait time in ED when compared to whites [53][54][55]. A recent study found a 30% longer wait time for AAs as opposed to whites [56]. In addition, the ED length of stay is longer for this target population [57]. Therefore, it is reasonable to assume that lack of continuity of care, a higher rate of frequent ED visits, combined with a longer waiting time at ED are significant factors contributing to health disparities among AA older adults. Current research shows that continuity of care improves quality of care, improves patient outcomes, reduces hospitalizations and decreases ED visits, particularly among older adults with CMCs that need self-management [58]. Improving and maintaining continuity of care among older adults have the potential to reduce health disparities. Programs that increase continuity of care should be developed and implemented among undeserved AA older adults. Pain severity and number of CMCs were revealed as the significant "need-for-care" variables for ED utilization. It is not surprising that higher pain severity leads to ED utilization. Overall, AAs report a higher pain level than whites when appearing in the ED [59] and a lower rate of being prescribed pain analgesics. Systematic reviews on race/ethnicity and chronic pain in the United States show that AAs are more likely to suffer untreated pain compared to whites [60]. Additionally, it is well known that minority patients are often under-evaluated and under-treated for their painful conditions in the ED [61,62]. Moreover, racial/ethnic disparities in pain management persist due to limited access to health care, inappropriate analgesic use and limited access to or under-utilization of pain specialists [63]. Empirical evidence suggest that older AAs have a higher risk for severe pain compared with whites [64]. Management of both multiple CMCs and psychological distress are strong predictors of severe pain which may explain racial/ethnic differences in pain severity [64]. In addition to an accurate diagnosis and high psychosocial functioning, pain management of AA older adults with multiple CMCs and comorbid cognitive decline benefits from a multidisciplinary approach that involves pharmacotherapy, physical rehabilitation, psychological support, and various other interventions [65]. Our findings are similar to other studies that have called for robust translational research programs on the causes and consequences of chronic pain that are specifically designed to reduce and eliminate disparities in management of care [61]. Successful case management of AA older adults with multiple CMCs requires development and implementation of well-designed care-continuum that prioritises patient safety. Delivery of health care in an interrupted, fragmented, or uncoordinated manner contributes to increased healthcare costs and inappropriate healthcare use [66]. Improving continuity of care in this context requires careful coordination across the healthcare continuum [67]. Successful communication between EDs and primary care providers in acute and community settings is an essential component of such interventions and programs [67]. A recent systematic review that evaluated the effectiveness of several types of interventions and programs targeting adult frequent ED users, showed that some interventions are effective in reducing visit frequency and improving patient outcomes [68]. While not directly addressed by our study, ED case management may be an appropriate intervention to decrease frequent ED utilization among underserved AA older adult patients, particularly those with multiple CMCs who suffer from pain. Case management programs have been adopted at multiple health systems as a method of improving care and decreasing costs [69]. Moreover, specialized ED case management programs are currently being implemented with a centered interdisciplinary team of physicians, nurses, social workers, and other healthcare team members. A specific focus is placed on coordination of care and communication, cross-cultural competencies, counseling, and specialized care referrals [49]. Limited available data demonstrate that frequent ED users who utilize case management programs services report a higher quality of life and improved biopsychosocial health [49]. Multiple studies also report substantial reductions in rates for ED visits and Medicare spending after implementation of a case management program [70][71][72]. One study found that individuals who accept community case management services have 55% fewer ED visits and 61% fewer hospitalizations compared to those who refused case management services [73]. Another study found that ED visits were lowered by 49% and ED length of stay decreased by 39%, attributed to the case management programs at their hospital [74]. ED-CM programs for older adults, especially AAs, must consider customized cultural and aging interventions to assist with disease management and pain treatment. Given the limited resources available in underserved areas , priority should be given to minority older adult patients with multiple CMCs with severe pain, especially those with limited access to and lack of continuity of care. In the bivariate analyses, we also observed a statistically significant relationship between depressive symptoms and ED utilization. However, multivariate analysis showed no association between these two variables. The association between depressive symptoms and ED utilization was subsumed by including the number of CMCs in the analysis. African American older adults with multiple CMCs may experience depressive symptoms and be at risk for major depressive disorder. Management of multiple CMCs may also lead to higher depressive symptoms [72,[75][76][77][78]. A systematic review of 40 articles by Read and colleagues found that individuals with multimorbidity are twice as likely to have a depressive disorder and there is a 45% increased odds of depression with each additional CMC [79]. African American older adults have multiple CMCs, so it is imperative that health care providers screen AA older adults, especially frequent ED users, for comorbid CMCs and depression. Implementing a routine self-administered questionnaire in the ED can identify older adults at high risk of depression. Therefore, our health care system is facing two notable phenomena regarding ED utilization among AA older adults, both clearly pointing to major health disparities. First, AA older adults may have higher utilization of the ED compared to whites, particularly for non-urgent care. Second, AAs are at increased risk of non-scheduled hospital admissions, compared to their white counterparts. Indeed, little attention has been paid to ED utilization in AA older adults. A greater understanding of the factors that account for acute care utilization in the AA population will provide healthcare providers with strategies that can facilitate more effective utilization of ED services, promote the proper utilization of health services, improve their health status, and decrease over-crowded EDs. The current study was conducted in SPA6 south LA. Being home to over 1,000,000 residents, SPA6 is disproportionately affected by various health disparities relative to the rest of Los Angeles County [80]. For example, the age-adjusted diabetes death rate in SPA6 is 37.6 per 100,000 population which is almost five times higher than that in West Los Angeles . Similarly, age-adjusted coronary heart disease death rate in the SPA6 is 147.5 per 100,000 population, compared to 87.7 in SPA5. While 32.5% of adults in SPA6 reported difficulty with accessing medical care, only 13.1% of SPA5 residents experienced these issues [80]. This evidence supports a pressing need for in-depth analyses of health care access and utilization among underserved older AAs in SPA6. This study has implications. Providing proper and safe care for minority older adult patients with uncontrolled chronic conditions remains challenging for ED providers. There is a need for additional research to identify older adults who "frequently use ED" in the under-served and under-resourced urban community in order to design culturally appropriate interventions that can ultimately reduce unnecessary ED visits and improve the quality of care for this segment of our population. --- Limitations Several limitations of this study should be considered before the results are interpreted and applied by clinicians, program planners, and policy-makers. First of all, the research did not collect data from participants' medical records. We relied on self-reported data. Second, information regarding health services utilization was limited. We could collect data on nuances and details such as cause of ED use. Third, the study did not use a random sampling. We recruited a convenience sample that limits the generalizability of our results. To compare with other estimates, given the non-probability sampling methods of this study, future research may conduct comparisons by probability samples to describe the methods used to draw the sample, collect the data, and make inferences. However, our study generated comparable estimates to estimates obtained from other surveys that used probability sampling methods. This is the reason we compared our estimates of education and SRH to the estimates obtained from the CHIS survey. We did not have data on ED visits using the CMS Medicare database. Other data sets like the CMS Medicare database could provide more confidence in the accuracy of the estimates given in this study. Fourth, the study did not have any whites or any other racial and ethnic groups. As a result, we do not know if ED is higher or lower, and whether unique processes impact ED utilization in AA older adults. Fifth, some differences may exist between AA men and women regarding determinants of ED use. This study did not explore gender differences in causes and consequences of ED utilization. Finally, design of this study was cross-sectional, which limits any causal inferences. More research is needed with data collected beyond a single point in time. Nevertheless, this study is a unique community-based research that employed face-to-face interviews with a relatively large sample of AA older adults that was conducted in an underserved and under-resourced area of South Los Angeles. Still, the results should be regarded as preliminary. --- Conclusions Despite its limitations, the results of the current study suggest avenues for future research on ED utilization patterns of low income urban older AAs with multiple CMCs. Future longitudinal research should examine the relationship between ED visit and ED-to-hospital inpatient admission. Future research may also study predisposing, enabling and need-for-care characteristics over time to understand the temporal causality between these factors. We encourage further research to explore causes and patterns of frequent ED utilization among older AAs. We also encourage a shift toward personalized patient-centered delivered systems. This research demonstrates a need to explore ways to enhance the communication between primary care and ED providers to support patients in engaging in continuity of care settings. A key factor in these patient-centered delivery systems is socially and culturally effective and responsive care that engages patients to meet their needs. Given the impact of pain and CMCs on ED utilization, improvement of disease management and control of pain may be regarded as key components in reducing ED over-utilization in lower income urban older AAs with multiple CMCs. --- Author Contributions: M.B.: conceptualization of the study, study design, funding acquisition, data analysis, first draft, overseeing the study, approval of the final draft; J.S.: data collection, analysis of data, contribution to the manuscript, approval of the final draft; S.C.: data Analysis, contribution to the manuscript, approval of the final draft; L.B.: contribution to the manuscript, approval of the final draft; C.W.: data management, contribution to the manuscript, approval of the final draft; E.N.: managing data, contribution to the manuscript, approval of the final draft; R.J.T.: contribution to the manuscript, approval of the final draft; S.A.: contribution to the manuscript, revision of the paper, approval of the final draft. ---

Objectives: Using the Andersen's Behavioral Model of Health Services Use, we explored social, behavioral, and health factors that are associated with emergency department (ED) utilization among underserved African American (AA) older adults in one of the most economically disadvantaged urban areas in South Los Angeles, California. Methods: This cross-sectional study recruited a convenience sample of 609 non-institutionalized AA older adults (age ≥ 65 years) from South Los Angeles, California. Participants were interviewed for demographic factors, self-rated health, chronic medication conditions (CMCs), pain, depressive symptoms, access to care, and continuity of care. Outcomes included 1 or 2+ ED visits in the last 12 months. Polynomial regression was used for data analysis. Results: Almost 41% of participants were treated at an ED during the last 12 months. In all, 27% of participants attended an ED once and 14% two or more times. Half of those with 6+ chronic conditions reported being treated at an ED once; one quarter at least twice. Factors that predicted no ED visit were male gender (OR = 0.50, 95% CI = 0.29-0.85), higher continuity of medical care (OR = 1.55, 95% CI = 1.04-2.31), individuals with two CMCs or less (OR = 2.61 (1.03-6.59), second tertile of pain severity (OR = 2.80, 95% CI = 1.36-5.73). Factors that predicted only one ED visit were male gender (OR = 0.45, 95% CI = 0.25-0.82), higher continuity of medical care (OR = 1.39, 95% CI = 1.01-2.15) and second tertile of pain severity (OR = 2.42, 95% CI = 1. 13-5.19). Conclusions: This study documented that a lack of continuity of care for individuals with multiple chronic conditions leads to a higher rate of ED presentations. The results are significant given that ED visits may contribute to health disparities among AA older adults. Future research should examine whether case management decreases ED utilization among underserved AA older adults with multiple chronic conditions and/or severe pain. To explore the generalizability of these findings, the study should be repeated in other settings.

About the study The UCL-Penn Global Covid Study 1 launched in April 2020 is a 12-month longitudinal study of the impact of Covid-19 on social trust, mental health and physical health. in collaboration with six institutions from italy, Singapore, the United States, China and the United Kingdom, 2 the study looks at the short-and longer-term effects of Covid-19 on an individuals' mental health and social relationships with others. Survey data were collected at three time-points: 17 April-14 July 2020 , 17 October-31 January 2021 and 17 April-31 July 2021 . --- About the webinar Held online between 2 June and 28 July 2021, the study group presented research data at five online webinars, as part of the UCL Global Engagement Fund sponsorship, to discuss the lessons learned and invited policy makers and other subject experts to speak on the policy relevance and implications of the study findings. The recorded comments from these discussions focusing on the policy relevance and implications of each academic article were recorded as discussant articles and are published in this journal to be read alongside the research article being discussed. These discussant articles are reviewed by members of the Editorial Board before being published. it is hoped that these discussant articles, read alongside the academic articles, will provide more holistic understanding of the issues at hand, how findings may inform policies in the coming months and/or assist in future crisis management strategies and aid decisionmaking, in an open and transparent manner. The study was pre-registered https://osf.io/4nj3g/ on 17 May 2021) and ethical approval was obtained from the University College London institute of Education Ethics and Review Committee on 8 April 2020 . 1 --- Linked research article --- Introduction The UCL-Penn Global Covid Study Family Life: Stress, relationship conflict and child adjustment, focuses on studying family relationships during the coronavirus pandemic. it specifically seeks to contribute to our understanding of the bi-directional relationship dynamics that occur between parents and their children, by focusing on the effects of child behaviour on parental outcomes. --- ( The research team collected data on relationships, trust, empathy, conflict, parenting and stress using parents' reports on a 30-minute online survey. Data were collected in eight languages between April 2020 and January 2021, which reflect three UK lockdowns and easing . The current study reported on two waves of data. Data collection efforts were ongoing for the third wave as the findings were presented. Studies conducted early-on during the pandemic period reported that hardships attributed to Covid-19 were associated with worsened parental mood and increased uncooperativeness by children [1]. This study adds to the body of knowledge by measuring changes in parental outcome across two waves of data. More specifically, the study evaluated the extent to which parental depression and stress changed from Wave 1 to Wave 2 as a function of child emotional problems, conduct problems and hyperactivity at Wave 1. Results were presented as part of a series of online webinars between June and July 2021. The webinar for the study under consideration was presented on 30 June 2021. Adverse child behaviours are linked to marital relationship conflicts, poor parental mental health and reduced quality of parenting practices. This study adds to the body of knowledge by exploring these complicated family dynamics under unique social, relational and emotional conditions: a global pandemic. These conditions are such that children's behaviours are worsening just as parents' mental and emotional capacities to deal with their disruptive behaviour are waning. --- Discussant comments The discussion surrounding this paper aroused a great deal of questions and recommendations for future studies, particularly surrounding the finding that child emotional problems, conduct problems and hyperactivity were not associated with relational conflict. This article focuses on highlighting some of the questions and future research recommendations that came up during those conversations. --- Sustained effects The findings presented were based on two waves of data collected between April 2020 and January 2021, allowing us to evaluate the change in parental outcomes during the initial stages of the pandemic and six months later. There is also a need to investigate and report on the sustained effects of child adjustment on parental outcomes as families and other social institutions learn to navigate the challenges associated with the Covid-19 pandemic. Governmental restrictions have varied over time and across nations, but overall, they have softened. Schools and businesses have re-opened. Parents have gone back to work. How have these shifts affected family dynamics? --- Parental relational conflict The study reported no significant effect of child behaviour on martial conflict, which was unexpected. Several points related to this finding were discussed including a potential measurement issue and a limitation of the sample. As it pertains to the measurement issue, relationship conflict was measured using items that reflected general areas of relationship conflict, as opposed to conflicts directly related to parenting practices, which are more likely to be linked to child misbehaviour. The issue related to the sample revolved around the fact that it was relatively normative. Child level of misbehaviour may not have reached the threshold necessary to see significant results we expected to it. Family systems theory [2], and in particular the concept of 'scapegoating', was brought up as a potential theoretical framework for understanding why child emotional problems, conduct problems and hyperactivity did not predict parental relationship conflict. The paradoxical idea that once a family member is 'scapegoated' difficult family dynamics become stabilised may be at play here. We are living in a time of global uncertainty and anxiety, which uniquely impacts families with children. These data were collected at the very early stages of the pandemic. it is possible that as these data were collected, the families represented in this study were operating under survival mode and would resort to any and all psychological processes in order to find stability in the most uncertain of times. --- Controls versus categories A good portion of the ideas shared during the discussion involved looking at differences by subcategories of the sample such as parents' sex and age. These variables were introduced as controls in the regression models and therefore held constant. it is possible, however, that when looking at these findings overall and/or if we artificially hold something as constant that in fact varies, we are unintentionally concealing interesting and/or significant results by sub-categories of the sample. it may be worth trying to figure out if the effect of child behaviour on parental relationship conflict depends on the parents' age, sex or profession. --- Self-report data versus official statistics One of the strengths of this study is that it is based on self-report data. in the United States, reports of suspected child abuse and neglect decreased substantially during quarantine. At first glance, this may seem like a positive social change perhaps attributed to families spending more time together. On the other hand, it could also reflect time away from mandated reporters due to school closures. The importance of self-report data to complement the official statistics cannot be overstated. --- A call to unify One of the most important take-away points from this study, and in particular the transactional model framework that it employs, is that family relationship dynamics are complex and multidirectional. Responses to family conflict and stress, particularly during times of heightened anxiety and uncertainty need to be multi-faceted to have a positive effect on our families and children. We cannot operate in silos and we cannot do it alone. Social service agencies often have more work to do than they have resources to do it with. Forming collaborative relationships across agencies is more important than ever. --- Declarations and conflicts of interest --- Research ethics statement The author declares that research ethics approval for the study was provided by the University College London institute of Education Ethics and Review Committee in April 2020 . --- Consent for publication statement The author declares that research participants' informed consent to publication of findings -including photos, videos and any personal or identifiable information -was secured prior to publication. --- Conflicts of interest statement No conflicts of interest or funding related to this. The author provides consent for the publication of the views expressed here and during the webinar. --- Extra information UCL Open: Environment is an open scholarship publication ucl.scienceopen.com

The main objective of this article is to comment on the findings presented during the UCL-Penn Global Covid Study webinar, 'Family Life: Stress, Relationship Conflict and Child Adjustment' by Portnoy and colleagues. The study examined the ways in which family stress conflict has been affected by the coronavirus (Covid-19) pandemic. informed by the transactional models of parentchild behaviour, the authors are specifically interested in exploring the effect of child adjustment on parental outcomes. The study, currently under consideration for publication, found that child emotional and conduct problems predicted changes in parental depression and stress during the early stages of the Covid-19 pandemic. Child hyperactivity predicted parental stress, but not depression. None of the child behaviour problems (emotional problems, conduct problems and hyperactivity) predicted parental relational conflict. This article discusses reasons why the study under consideration did not find a significant effect on relational conflict and posts questions that can be addressed in future studies.

INTRODUCTION Canadian youth have one of the highest rates of cannabis use worldwide 1 2 with prevalence rates that are almost double that of adults. Recent data indicate that 44% of youth aged 16-19, and 51% of youth aged 20-24, report past year cannabis use versus 21% of adults over the age of 25. 3 Daily cannabis use is linked to more significant adverse effects 4 and is reported by 16% of youth aged 16-19 and 23% of those aged 20-24. 3 In Canada, recreational cannabis use was legalised in 2018 across the 10 provinces and 3 territories, and one of the key objectives of the Cannabis Act is to protect the health and safety of youth. 5 Despite this objective, cannabis use among youth continues to be a public health concern linked with physical and mental health concerns including anxiety, 6 depression, 7 8 psychosis, [9][10][11] attention deficit hyperactivity disorder, respiratory problems and use of other illicit substances. 12 Youth with mental health concerns have higher rates of cannabis use, 13 and early initiation of cannabis use is linked with increased mental health concerns. 12 Earlier initiation and higher frequency of cannabis use are associated with more adverse effects 14 15 and a higher risk of dependence. 16 Studies show that health knowledge and awareness of perceived risks influence the rates of cannabis consumption with higher rates when a substance is not perceived as harmful, which makes education and prevention a public health priority. 15 17 Canadian youth have a wide range of beliefs about recreational cannabis use, and many have misconceptions and lack information about the risks and harms associated with cannabis use. 18 19 Service providers also have gaps in knowledge about cannabis use in youth and need more education and training. 20 21 A recent survey conducted in the USA where 11 states have legalised recreational cannabis use found variation and gaps in service providers' knowledge and beliefs on cannabis. Many service providers described discomfort discussing cannabis use with clients. 22 Parents and caregivers also need more information --- STRENGTHS AND LIMITATIONS OF THIS STUDY Open access on the harms associated with cannabis use in youth, and studies have shown that parents have concerns about the adverse effects of cannabis use on mental health. 23 24 There is inadequate support for families and insufficient knowledge that is drawn from the lived experiences of diverse families. 25 Parents have been described as 'invisible experts' as they are often the core support for youth with substance use disorders, but their perspectives are often excluded from research, services, and policies. 23 26 Despite widespread use and robust evidence of associations between cannabis use and adverse effects on youth, 7 8 27 there are several gaps in the current literature on youth cannabis use. These knowledge gaps are even wider with marginalised populations such as sexual and gender minority youth, immigrant and refugee youth, and Black, Indigenous and youth of colour. [28][29][30] There is some research showing gender differences in cannabis use, 31 as well as ethnoracial and immigrant group differences, 32 but more research is needed to increase our understanding of factors that may be contributing to these differences. Given the gaps in our understanding of risks and harms to youth, updated research can inform policies, public education strategies and evidence-based programming. 28 33 Hawke and colleagues 27 state that there is a need for research that is youth-specific and cannabis-specific with a focus on concurrent disorders. Others argue that research and educational interventions do not consider contextual and cultural factors 34 and have not engaged youth and parents adequately. 25 35 Few studies have examined parents' perspectives on addressing cannabis use, resulting in substance use programmes that are not evidence-based and developed without the lived experiences and expertise of parents and youth. 25 Furthermore, this underlines the importance of research that focuses on diversity within the Canadian youth population and the influence of social determinants of health. 5 28 36 Early intervention is essential to mitigate the adverse effects of cannabis use on youth. However, the lack of service continuity contributes to 52% of youth dropping out of treatment, resulting in untreated mental health and substance use concerns, poorer long-term mental health outcomes and increased risk of severe mental illness, homelessness, high school dropout and unemployment. 13 In Canada, several factors contribute to high rates of untreated mental health and substance use concerns, 37 including a lack of a national strategy and pan-Canadian policy, unresponsive and fragmented services that do not meet the needs of youth with concurrent disorders, 38 and inadequate engagement of youth and families in service development and delivery. 39 40 Moreover, LGBTQ youth, 41 immigrant and refugee youth, as well as Black, Indigenous, youth of colour 28 have additional barriers to accessing mental health and addiction services related to inequities, discrimination and systemic racism. [42][43][44] The objectives of this protocol are to scope the literature on cannabis use among youth under 25 in Canada; examine the relationship between cannabis use and physical and mental health in Canadian youth; examine the relationship of cannabis use and co-occurring use of other substances in Canadian youth; identify prevention strategies, interventions and programmes to address cannabis use in Canada; explore how the Canadian literature considers the influence of social determinants of health on youth cannabis use; and identify gaps and recommendations in the literature. --- METHODS AND ANALYSIS Scoping reviews are widely used approaches for mapping the literature on topics that have insufficient evidence, and they provide greater breadth and depth than other reviews. 45 Due to the limited research on cannabis use among Canadian youth, this form of knowledge synthesis will be valuable in providing an overview of existing literature, and it will identify knowledge gaps and generate recommendations to inform policies and services. 46 47 This scoping review will follow the methodological framework introduced by Arksey and O'Malley, 48 which includes the following five phases: identification of the research question, identification of relevant studies, selection of relevant articles, charting of the data, and identification, synthesisation and reporting on study findings. 48 This scoping review is also following guidelines described in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews . 49 See online online supplemental appendix A for the PRISMA-ScR checklist. --- Stage 1: identifying the research question As highlighted in our preliminary review of the literature, there are significant gaps in the literature on youth cannabis use in Canada. Using a population, concept and context framework, the research team consisting of four social work faculty members , one research coordinator who is a doctoral candidate , one social sciences librarian and two graduatelevel research assistants developed these research questions: what type of literature is available describing youth cannabis use in Canada? how does the literature describe the relationship between cannabis use and physical and mental health? how does the literature describe the relationship between Canadian youth cannabis use and use of other substances? what are the programmes and prevention/intervention strategies used to address youth cannabis use in the Canadian literature? how does the literature consider the influence of the social determinants of health on cannabis use in Canadian youth? what are the gaps and recommendations identified in the literature on cannabis use in Canadian youth? Step 2: identifying relevant studies A social sciences librarian will design a comprehensive search of the published literature. Text words and controlled vocabulary relating to youth, Canada Open access and cannabis will be adapted from reputable published search terms. [50][51][52][53] Table 1 presents a draft search strategy developed for MEDLINE . The search strategy will be peer-reviewed by an independent librarian and translated into Embase , APA PsycInfo , CINAHL and Web of Science Core Collection, which includes Science Citation Index Expanded, Social Sciences Citation Index, Arts & Humanities Citation Index, Emerging Sources Citation Index, Conference Proceedings Citation Index and Book Citation Index. A date limit of 2000+ will be applied. Search results will be downloaded on a single day and then uploaded to Covidence for deduplication and screening. Step 3: study selection We will include articles meeting the following inclusion criteria: written in English or French; published on or after 2000; focus on Canada and/or any of its provinces, territories or other jurisdictions; focus on youth aged 24 and younger; focus on cannabis and/ or cannabis use; and empirical studies using any type of method . Papers that examine substance use are eligible if cannabis is specifically identified. Likewise, papers that focus on adult populations and include young adults aged 18-24 are also eligible if the young adult age group is distinguished. We will also include papers that compare findings related to cannabis use and youth across countries, provided Canada is included as one of the countries of comparison. We will exclude reviews, theoretical or conceptual papers, books, book chapters, book reviews, dissertations, commentaries and editorials. After selecting a list of articles from our search strategy, we will use an iterative two-stage peer review screening process and include two independent screeners at each stage. In the first stage, two independent reviewers will screen articles for suitability based on title, abstract and keywords. In the second stage, the reviewers will conduct an independent full-text review of articles selected in phase I. The first author will resolve discrepant findings between the first and second reviewers and hold debriefing meetings to clarify questions around eligibility. To support the process, we will use Covidence, a web-based software for systematic and scoping reviews that facilitates screening, study selection and data extraction. 54 Step 4: charting the data Based on the objectives and research questions, the research team has developed charting categories, and these will be used to extract relevant information from the selected papers. A data charting form will be prepared that includes the following charting categories: authors; year of publication; name of journal; type of paper ; method ; sample size and characteristics; physical and mental health outcomes; use of other substances; programmes, prevention and intervention strategies; influence of social determinants ; youth and/or family engagement in research; and gaps and recommendations for policy, practice, education and research. To enhance reliability, two reviewers will independently chart the first five articles meeting the inclusion criteria, and the lead author will review charting completed by the reviewers and resolve conflicts. Any discrepancies --- Open access will be discussed in a debriefing meeting and charting categories will be refined if necessary. Full data abstraction will only begin after sufficient agreement has been achieved on charting . Findings will be organised and presented on a data extraction spreadsheet. Step 5: identification, synthesis and report of study findings Findings on the data extraction form will be synthesised and analysed using descriptive numerical summaries and thematic analysis. 55 There will be multiple forms of knowledge translation used to report study findings, including an open-access publication in a peer-reviewed journal, presentation at a relevant addiction and/or mental health conference, and an infographic that presents the findings in a more engaging manner. Patient and public involvement While we will not involve patients or members of the public in this review, we will disseminate the results of this scoping review to mental health and substance use community organisations serving youth and families. --- ETHICS AND DISSEMINATION The scoping review protocol outlined in this paper will advance knowledge of cannabis use in Canadian youth. The information gathered for this paper and the outlined scoping review were retrieved from publicly available sources; therefore, ethics approval is not required for this project. The results will be disseminated through a peerreviewed journal and reported at national and international conferences on mental health and addictions. Twitter Toula Kourgiantakis @DrKourgiantakis Contributors All authors made substantive intellectual contributions to the development of this protocol. TK, EL, SLC, and CCW contributed to the conceptualisation of the protocol. TK developed, wrote, and edited the initial protocol. TE, RV and MS-T contributed to editing and writing parts of the protocol. JL developed the search strategy and contributed to the writing of the protocol. All authors critically reviewed and revised the final version prior to submission. Competing interests None declared. Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research. --- Patient consent for publication Not applicable. Ethics approval Not applicable. Provenance and peer review Not commissioned; externally peer reviewed. Supplemental material This content has been supplied by the author. It has not been vetted by BMJ Publishing Group Limited and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations , and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Introduction Canadian youth (aged 15-24) have the highest rates of cannabis use globally. There are increasing concerns about the adverse effects of cannabis use on youth physical and mental health. However, there are gaps in our understanding of risks and harms to youth. This scoping review will synthesise the literature related to youth cannabis use in Canada. We will examine the relationship between youth cannabis use and physical and mental health, and the relationship with use of other substances. We will also examine prevention strategies for youth cannabis use in Canada and how the literature addresses social determinants of health. Methods and analysis Using a scoping review framework developed by Arksey and O'Malley, we will conduct our search in five academic databases: MEDLINE, Embase, APA PsycInfo, CINAHL and Web of Science's Core Collection. We will include articles published between 2000 and 2021, and articles meeting the inclusion criteria will be charted to extract relevant themes and analysed using a qualitative thematic analysis approach. Ethics and dissemination This review will provide relevant information about youth cannabis use and generate recommendations and gaps in the literature. Updated research will inform policies, public education strategies and evidence-based programming. Results will be disseminated through an infographic, peer-reviewed publication and presentation at a mental health and addiction conference. Ethics approval is not required for this scoping review. ⇒ This is the first comprehensive review examining youth cannabis use in Canada. ⇒ The search strategy has been developed by a research team with expertise in the methodology and subject area. ⇒ This scoping review will include all article types and methodologies. ⇒ Due to the nature of the scoping review framework, the studies included in the review will not be appraised for quality.

Population aging in the United States, as in most countries around the world, poses many challenges to our major private, public, and societal institutions. At the same time, however, it creates great opportunities for building a more resilient, successful, and cohesive society. For several reasons, work organization and labor-force participation are central to both opportunities and challenges posed by an aging population. First, for all but a very few, working remains the main source of income for consumption and for savings. Work also has an anchoring function in society, bringing multiple bene½ts to our physical, social, and emotional well-being. Given the continuing increase in life expectancy in aging societies, it is more necessary than ever before to revisit the role of work in older ages and the opportunities that longer working lives can bring to aging societies. We argue in this essay that individual expectations about old age have not suf½ciently adjusted to the new reality of an aging society. Reflecting this, our private, public, and societal institutions suffer from the same disconnect and now need to adapt considerably to face the challenges and embrace the opportunities of demographic change. Although this adaptation must focus es pecially on work patterns among the "younger elderly," our entire society has to adapt its attitudes toward work organization and labor-force participation, and in the process rethink its education and health care policies and expectations. In this essay, we will describe some of the often understated bene½ts that working and remaining active may have for health and well-being in older populations. Furthermore, it is often claimed that an older workforce has negative implications for general productivity and displaces younger workers from positions they would otherwise occupy. We will show that this is a misconception emerging from the "lump-ofla bor" fallacy: the idea that the amount of work available to laborers is ½xed. We maintain that it is not demographic transitions per se that will shape our future, but instead how our institutions and policies respond and adapt to them. It is our choice. Individual expectations about old age tend to be formed by looking to history. Our intuition about health and workability at older ages stems to a large extent from the experiences of our parents and grandparents. Our expectations about what Social Security should deliver to us stem from what Social Security delivered to previous generations. We have a hard time imagining, however, what our lives will look like in twenty-½ve or ½fty years: how long we will live, how healthy we will be by then, and under how much pressure Social Security will be. We will address all of these issues, beginning with the salient points about Social Security At the time Social Security was established in 1935, sixty-½ve was commonly the age U.S. citizens received other government bene½ts. Table 1 shows the dramatic changes in life ex pectancy that have taken place since thenchanges that have been reported in many articles and studies, but that appear not to have entered our collective wealth of selfevident facts on which individual actions and general policy are based. In the mid-twentieth century, just over half of all Americans who reached the age of twenty-one could expect to reach sixty-½ve. Many workers paying into Social Security would never live long enough to receive bene½ts, especially African Americans, whose life expectancy was just over ½fty in 1935. Today, however, men who retire at age sixty-½ve can expect to live for an additional seventeen years; women live even longer. 1 Average life expectancies have risen so sharply since 1935 that it is no surprise that Social Security-and our expectations about what old age looks like-have been unable to keep up. Signi½cantly, there is also evidence for a substantial increase in healthy and disability-free life expectancy; that is, the num ber of years men and women can expect to live without major functional limitations. Despite some controversy, the gen eral trend appears clear: over the past decades, as life expectancy has improved in the United States, so too has healthy or disability-free life expectancy. When life expectancy improves and morbidity and disability decease, this leads to "compression of morbidity," or a longer part of life spent in good health and a shorter number of years lived in ill health. Most studies show that a signi½cant compression of mor bidity has occurred over the last decades in the United States, making it clear that many-although not all-Americans are able to work until after current retirement ages or eligibility for Social Security. 2 Yet some evidence suggests that compression of morbidity has stagnated in recent years. Even more concerning are current hints that the middle-aged and the "young old" are showing increases in morbidity and functional limitations not evident in earlier cohorts that grew old between 1990 and the present. A closer look at the evidence shows that good health at older ages is strongly predicted by educational attainment and other indicators of socioeconomic status. In a recent study, pub lic health researcher David Rehkopf and colleagues projected the future employment potential for the near-elderly workforce based on demographic trends and trends in mobility and functional status from the 1982-2004 National Long Term Care Study and the 1992-2010 Health and Retirement Study. 3 Rehkopf's projections through 2050 focus speci½c attention on educational differences. His group concludes-based on multiple sce-narios of population trends in disabilitythat those with a high school diploma have generally a high and consistent potential to work productively between the ages of ½fty-½ve and seventy-four. The estimates suggest that under most scenarios, about 70 percent of those with at least a high school diploma will be able to work in this age range. These ½ndings are generally in line with studies done by economist David M. Cutler and others, who assume no huge upward turns in disability prevalence, even with modest or no improvements in active life expectancy. 4 But the same is not true for those with low levels of education. Rehkopf's team sug gests that if less favorable trends continue for men and women without a high school diploma, this group could have depressed levels of ability to work at older ages. 5 These estimates vary much more depending on different assumptions about trends, but they range from just over 0 percent to 60 percent in the most optimistic circumstances. Of course, new medical tech nologies could alter these trends by providing more support to those with limitations. Furthermore, patterns of immi gration bringing in more or less educated young and middle-aged adults could also change these estimates. But, assuming that 43 Lisa F. Berkman The dramatic demographic changes depicted in Table 1 imply that most men, and even more women, will now survive early childhood and mid-adulthood to reach old er ages, meaning that larger and larger numbers of older men and women will reach eligibility for Social Security. When most retirement and other age-related policies were established, they made sense in terms of the current life expectancy. Today they do not. These demographic changes point to the necessity of institutional adaptation. Among industrialized nations, the United States does not have the largest disparity between life expectancy and retirement age. France and Italy, for example, set the age of early retirement earlier than the United States' , even though they enjoy greater life expectancy than the United States does. Thus, the time from retirement to death is decades long in France and Italy, creating a large imbalance between work pat-terns and demographic reality. In response, France and Italy have joined the growing number of European countries adapting their statutory retirement ages, although only for future cohorts and with many exceptions. Germany, for instance, has confronted this challenge through a variety of policy adjustments: First, it introduced penalties for early retirement that were gradually phased in between 2000 and 2010. Second, it abolished some special tracks for early retirement, including the differential treatment of men and women . Third, the stat utory retirement age began gradually shifting in 2013 by about a month per year to reach a target of age sixty-seven by 2029 -roughly in line with the expected increase in the length of life. These adaptations have met much resistance. Accommodating pop ular opinion, the German gov ernment reintroduced a pathway to early re tire ment at age sixty-three. France and Italy have seen similar policy reversals, backtracking from modest increases in statutory retire ment ages to early retirement op tions. Pub lic anxiety about increased retirement age clearly shows the need to overcome pop ular misconceptions about actual and fu ture life expectancy and health in older ages. The United States has done part of its home work and is ahead of Europe in its preparations. The Social Security Amendments of 1983 legislated a gradual shift in the eligibility age for normal Social Security bene½ts. It will not be suf½cient, however, to bring the U.S. Social Security system back into actu arial balance. Thus, a steeper path to changing the average worker's eligibility ages is still necessary. In addition, given the reality of a divide in the health of America's older population, it is critical to create differentiated paths to retirement and labor-force exits depending upon health . This life-course fork in long-term trajectories suggests that options for early retirement based on work disability are critical to maintaining intragen erational equity between those who can work longer and those who cannot. A onesize-½ts-all Social Security reform will not do; only a differentiated approach will avoid the political backlashes observed in Europe. This may mean the implementation of both a general retirement age that is indexed in some way to life expectancy and an early-retirement option based on the ability to work. For older workers in poor health, it is obviously better for their health and well-being not to have to work. This may mean that certain groups within the population-such as the less educated and those with very physically demanding jobs-may need the option to take an early path to retirement. Current Social Security policies enabling people to start collecting bene½ts around age sixty-two may be a good model for workers in this group. Those with better functioning or with limitations that do not interfere with their performance could have longer working lives and continue to make substantial contributions to productivity in U.S. society. Adapting the retirement age to a longer life has many implications. We note, as others in this volume have, that these do not exclusively affect the elderly. First, the ability of future generations to work longer hinges on their education and health throughout life. Additionally, since the United States will have a larger number of retirees in the future than ever before, the younger and middle-aged will have to be better-educated and in better health than they are today in order to carry this additional actuarial burden . To optimize our chances of providing a healthier workforce at older ages, we must adopt a life-course approach that focuses on the social experiences and behaviors of men and women across all ages. For example, the men and women who will be in their sixties and seventies in 2030 to 2050 are in early to mid-adulthood now. Their current health and social conditions are therefore shaping their capacity and the opportunities for employment options they will have at age sixty, seventy, or eighty. The point is even stronger for education because of the many implications a poor education has on opportunities over the entire course of life. We have no time to lose if we hope to change trajectories of work and retirement. Debates about our aging society too quickly end up foundering on the issue of Social Security reform and the well-being of the elderly, while tending to ignore the necessity of creating better starting positions for the young. Macroeconomic an alyses show that education and health care reform have more leverage in shaping our society's ability to create new jobs, foster better working conditions, and encourage labor-force participation than an adapted retirement age and disability insurance reform. 6 Other studies suggest that recessions have differential impacts on health in the long run, with disadvantages accruing to particularly vulnerable age groups. 7 In any argument about retirement, it is important to discuss the implications of working at older ages. In general, being em ployed is positively associated with health. There are many reasons for this relationship, the most obvious being health selection; that is, healthy people are more likely to be able to work. However, there is growing evidence that employment itself actively yields both physical and mental health bene½ts. Here, we will explore ½ndings suggesting that employment may im -prove health and well-being by increasing social engagement; developing and maintaining intellectual and interpersonal skills; and, importantly, continuing to earn and delaying the use of savings, pensions, and other bene½ts. This phenomenon has obvious implications for retirement, since neg ative effects of retirement are often ignored in the cost-bene½t analysis that is done when a state attempts to determine the optimal retirement age. Retirement introduces large changes to an individual's life. While there is little doubt that poorer health is associated with early retirement, studies on the health impact of retirement have so far reached no consensus on whether retirement promotes or harms health. Differentiation is essential in conducting these studies: phys ically and psychologically strenuous work conditions are unlikely to be good for a person's health, while working in a rewarding and healthy environment may be better for mental and physical health than leaving the workforce. Many of the existing studies have faced methodological problems: they do not distinguish the effects of aging from those of retirement and they often do not distinguish the effects of retirement from those of previous life experiences or conditions that themselves influence retirement decisions. Many of the descriptive studies lack an ad equate control or comparison group. This is important because the decision to retire is not random: there is a self-selection issue. For instance, those who are ill are more likely to retire early. Those with great wealth may also retire early. However, illness may be caused by environmental conditions and wealth by enhanced educational opportunities. Thus, in both cases, retirement is not causally linked to health or wealth but to previous life experiences and conditions. In addition, the effects of retirement on health may depend on many contextual factors, including the adequa-cy of retirement bene½ts, as well as individual factors such as occupation, socioeconomic status, and marital status. Retirement may also have different effects on physical and mental health, requiring the need for further differentiation in the methodology of studies on retirement. The literature on this question has focused on understanding how reforms on the age of retirement-namely, the age of compulsory retirement or the minimum age of retirement-might impact health. These laws have a strong effect on retirement decisions: while a substantial proportion of workers retire before the statutory age of retirement, a higher statutory retirement age encourages individuals to work longer. Here we focus on a small set of studies that have attempted to establish causality between the statutory retirement age and retirement's effects on health. We will examine two pieces of evidence that are important in this respect: First, we discuss some of the studies using longitudinal data to assess how retirement influences health, taking into account the complex set of factors that lead individuals into retiring early. Sec ond, we discuss the evidence of the health impact of retirement age-related policy reforms. Longitudinal studies follow the health of workers during the years prior, during, and after retirement and compare it to the health trajectories of workers who continue to work. Using data from employees from the French companies Électricité de France and Gaz de France , epidemiologist and gerontologist Hugo Westerlund and colleagues found that between the year before and the year after retirement, the prevalence of poor self-rated health fell from 19 percent to 14 percent. 8 These health improvements were stronger for workers with a poor work environment before re - --- Daedalus, the Journal of the American Academy of Arts & Sciences --- Labor-Force Participation, Policies & Practices in an Aging America tirement. Using data from the Whitehall II study of British civil servants, epidemiologist Gill Mein and colleagues found that although mental health improved after re tirement, physical functioning did not appear to change. 9 Mental health improvements, however, were con½ned to highgrade employees. In a reexamination of the data, epidemiologist and social scientist Markus Jokela and colleagues found that compulsory retirement at age sixty and early voluntary retirement were associated with improvements in mental health and physical functioning. 10 In contrast, retirement due to ill health was associated with poorer mental health and physical functioning. Their ½ndings highlight the important role of health-related selection as a potential explanation of the negative association between retirement and health. Many of the studies referenced above, which are con½ned to European populations, cast doubt on the notion that retirement is bad for health overall: the prevailing ½nding appears to be that in the short term, retirement is associated with an improvement in mental health and little or no change to physical health; though there is no doubt that the effect of retirement depends on the nature of the worker's oc cupa tion and health prior to retirement. More recently, studies have used differences across cohorts in eligibility for retirement bene½ts to iso late the effects of retirement on health. Economist Kerwin Ko½ Charles has used policy variation in mandatory retirement and Social Security bene½ts that influence retirement incentives by age and cohort in the United States to examine the impact of retirement on depression. 11 He found that retirement leads to better mental health and well-being. Other studies ex ploit variations across countries in the age of eligibility for early and full retirement bene -½ts. Based on these variations, economists Norma B. Coe and Gema Za mar ro found that retirement leads to a short-term decrease in the probability of reporting poor health, and a long-lasting improvement in the overall health in dex. 12 It is nonetheless important to distinguish general health from cognitive abilities, which appear to bene½t from working at older ages. The impact of retirement on cognitive function is of particular interest in view of the "use it or lose it" hypothesis, which suggests that age-related cognitive decline can be lessened through engagement in cognitively demanding activities. 13 This suggests that individuals whose work is cognitively demanding may bene½t from later retirement. Using data from the Whitehall II study, B. A. Roberts and colleagues found that workers who retired experienced smaller improvements in mean cognitive test scores than continuously employed workers, although these differences were not signi½cant for most cognitive test scores. 14 A seminal paper by psychologist Stéphane Adam and colleagues, based on the Study of Health, Aging, and Retirement in Europe , found that cognition-measured mainly by memory abilities such as delayed word recall-declined during retirement. 15 This ½nding has given rise to an entire literature. Economists Susann Rohwedder and Robert J. Willis compared studies of retirement from Europe, the United States, and Britain and found that early retirement has a negative effect on cognitive ability. 16 Their ½ndings, however, are not corroborated by other studies. Based on data from the U.S. Health and Retirement Study , Coe and colleagues ex amined employers that offered early retire ment windows and found that time in re tirement was unrelated to cognitive function among white-collar workers, but may have a positive effect on cognitive function among blue-collar workers. 17 More recent studies based on European data, however, show that early retirees suffer from fast er cognitive decline than later retirees. 18 Based on the hrs, sociologist Esteban Calvo and colleagues found that the effect of retirement depends on the timing: those retiring before age sixty-two seem to fare worse than those who continue to work; yet, retirement at age sixty-two or older is not associated with worse physical and mental health. 19 Again, these studies have generally not found any negative effects of retirement on physical or mental health. Other studies have also found that retirement has no impact on mortality. 20 The studies discussed above reflect the variety of approaches to determining the health effects of retirement. What do these results tell us about the potential impact of recently enacted policies to increase retirement age for future generations across many countries? On the one hand, there seems to be little evidence that retirement harms physical health or increases the risk of dying. Although some studies do suggest that retirement may be bene½cial to mental health, distinguishing between different cohorts is again paramount: several studies suggest that the mental health consequences of retirement depend on the working environment and type of job the retiree had. While retirement does ap pear to bene½t the mental health of many working in strenuous conditions and performing manual labor, this is less clear for work ers in white-collar positions and with healthy work environments. Finally, the more recent evidence tends to ½nd some support for the hypothesis that retiring later helps individuals maintain better cognitive function. What effects will delayed retirement have on the greater population and the young in particular? Higher rates of laborforce participation in older individuals is often said to have negative side effects for the economy as a whole. For many years, common sense suggested that the number of jobs in the economy is ½nite, and that a new population entering the labor force would therefore push other workers out. This so-called lump-of-labor fallacy has been invoked at moments in history when women's labor-force participation increased, because it was thought that they would take "good jobs" away from men. Immigrants to the United States con tinue to be accused of stealing jobs from other, native lower-wage workers. Likewise, many older people who wish to continue working today are accused of taking jobs from younger workers, creating intergenerational conflict. The lump-of-labor fallacy is one of the most damaging myths in economics. 21 It is deeply rooted in the belief that the economy resembles a small enterprise with a small, ½xed number of clients and a ½xed demand for its product. Such an enterprise has a set amount of output based on demand, and therefore can only use a certain amount of labor. This is a poor analogue to a suf½ciently large and complex economy. This is shown most clearly in the United States, where the sharp increase in female labor-force participation not only did not cause mass un employment for men, but actually correlated with a rise in male employment rates. More speci½cally, recent ½ndings from cross-national comparisons show that high er employment of older individuals is actually positively correlated with higher em ployment of the young; that is, countries with a high prevalence of early re tire ment tend to have higher unemployment rates and lower employment of the young. 22 Figure 1 shows a correlation between early retirement and youth unemployment in oecd countries. These ½ndings may be challenged, however, as many confounding factors operate at the same time in the 48 --- Daedalus, the Journal of the American Academy of Arts & Sciences Labor-Force Participation, Policies & Practices in an Aging America aggregate data. Strong and isolated re forms are more suitable for empirically identifying the effects of pension policies on labormarket outcomes for the young. It is therefore instructive to examine the impact of speci½c pension reforms on employment rates at different ages. Germany provides a particularly neat case, since strong and isolated reforms in the years 1972, 1984, and 1998 dramatically changed retirement incentives. 23 Figure 2 depicts the labor-force participation rates for four age groups in Germany, and Figure 3 presents the corresponding unemployment rates. These ½gures reveal three important facts. First, the 1972 reform dramatically reduced retirement age, laborforce participation, and employment of old er individuals. In spite of this, youth em ployment did not increase. Second, the "bridge to retirement" legislation intro-duced in 1984 substantially increased the unemployment rate of those aged ½fty-½ve to ½fty-nine, since unemployment insurance bene½ts were used as substitutes for early retirement pen sions. Yet youth employment did not rise in response. The phasing-in of "actuarial" adjustments after 1998 re versed the trend of early retirement. Em ployment in creased from 30 percent to 40 percent in those aged sixty to sixtyfour. There is a very slight concurrent decrease in employ ment of the young. The ½rst two cases are clear-cut: employment of the young and the old moved in tandem. But the third case may appear to contradict this relationship. Axel Boersch-Supan and economist Reinhold Schnabel, however, have shown in their regression analysis of the third case that the slight decrease in employment of the young is in fact a re flection of the business cycle and --- Percent of Elderly Out of Labor Force --- Percent of Youth Unemployed The R-squared of the correlation is 18 percent. This is not driven by the outliers . If these crisis-affected economies are omitted, the positive correlation is actually stronger (with an R-squared of 21 percent --- Labor-Force Participation, Policies & Practices in an Aging America not a response to the introduction of actuarial adjustments. 24 The German analysis is part of the work by an international team that used pensiondesign changes in eleven countries to identify how changes in the employment of old er individuals has affected the employment of the young. The results vary considerably across speci½cations, but in these studies there are many more cases that refute the lump-of-labor hypothesis than cases that support it. As economists Jonathan Gruber and David Wise have written: The overwhelming weight of the evidence, as well as the evidence from each of the several different methods of estimation, is contrary to the "boxed economy" proposition. We ½nd no evidence that increasing the employment of older persons will reduce the employment opportunities of youth and no evidence that increasing the em ployment of older persons will increase the unemployment of youth. 25 Countries have large multifaceted economies that cannot be likened to small companies with ½xed, "boxed" labor needs. National economies can grow, increasing the demand for all goods and services and therefore also the demand for labor. As with women and immigrants, the increasing entry of older workers into the workforce contributes to a vital and productive economy. Moreover, unless a pension system is fully funded, there is a tax cost for retirement-whether early or not-that must be spread over the entire economy. This raises the total labor compensation em ployers must pay for all workers, including the young. The greater the number of older workers that leave the workforce, therefore, the more likely it is that the employment prospects of the young will worsen. Increases in life expectancy and compression of morbidity, funding de½cits in Social Security, possible cognitive bene½ts to work ing at older ages, and the potential for economic vitalization are some of the factors that support increasing the number of individuals who work past today's statutory retirement age. It would be naive to expect that this will occur only through Social Security reform and legislation encouraging workers to work longer; we also need structural policy changes that gen erate a healthier and more productive America. These include policies that invest in human capital throughout individuals' lives, thus enabling them to work longer; such as policies on early childhood, education, employment protection, work flex ibility, income support, poverty reduction, and health care access. 26 Most individuals should not ex perience deterioration of men tal and phys ical health from working longer; rather, the goal is to support healthy aging in such a way that working will be more feasible and potentially flexible for older cohorts. Only through policies that promote life trajectories leading to healthy aging will we be able to create a workforce able to work longer, and only then will we be able to accrue the true societal bene½ts of So cial Security re form. How we adapt the major U.S. institutions related to work organization and laborforce participation will shape our future as we move through this demographic transition. Although shaping public and pri vate policies is of paramount importance, this adaptation must take place on all levels, such as with more in formal workplace practices. These policies and practices shape patterns of labor-force participation for older men and women and determine how they will pursue retire ment. These policies must also take into account that each older generation is a diverse set of men and women with different life-course patterns of education, skills, family constellations, and health conditions, and that this in turn affects the em ployment op -Lisa F. Berkman, Axel Boersch-Supan & Mauricio Avendano portunities they will have at older ages. Also determining outcomes of labor policy changes are the social, economic, and health capital of Americans in the labor force, currently and in the future. These two sets of conditions-one at the labor-policy level and the other at the population level -will determine whether we remain a resilient and successful society as we experience the aging of our population. --- Authors' Note: This work was supported by the MacArthur Foundation Research Network on an Aging Society and by a grant from the National Institute on Aging . Mauricio Avendano is also supported by the European Research Council . This paper includes sections adapted from Axel Boersch-Supan, "Myths, Scienti½c Evidence, and Economic Policy in an Aging World," The Journal of the Economics of Ageing 1-2 : 3-15; and from Mauricio Avendano and Lisa F. Berkman

Population aging in the United States poses challenges to societal institutions while simultaneously creating opportunities to build a more resilient, successful, and cohesive society. Work organization and labor-force participation are central to both the opportunities and challenges posed by our aging society. We argue that expectations about old age have not suf½ciently adapted to the reality of aging today. Our institutions need more adaptation in order to successfully face the consequences of demographic change. Although this adaptation needs to focus especially on work patterns among the "younger elderly," our society has to change its general attitudes toward work organization and labor-force participation, which will have implications for education and health care. We also show that work's bene½cial effects on well-being in older ages are often neglected, while the idea that older workers displace younger workers is a misconception emerging from the "lump-of-labor" fallacy. We conclude, therefore, that working at older ages can lead to better quality of life for older people and to a more productive and resilient society overall.

and this area of public health remains limited or marginal within mainstream policy practice. Surveillance of social determinants of health is therefore a neglected but essential and challenging public health issue. The WHO Commission on S ocial Determinants of Health strongly recommended the creation of National Health Equity Surveillance Systems, with routine data collection on the social determinants of health and health equity, and investment in training of policy-makers and health practitioners in equity monitoring and health equity impact assessment. 1 Reduction of Health Inequalities' as a top priority for the Spanish presidency of the European Union in the first semester of 2010. 4 In spite of these initiatives, however, today there is no comprehensive surveillance system capable of globally or nationally monitoring social determinants and their relationship with health inequalities. Employment and working conditions are key areas of social determinants of health research agenda. 5 This paper describes the importance, gaps and challenges in developing employment-healthequity surveillance systems. --- MONITORING EMPLOYMENT CONDITIONS Employment and working conditions are prominent determinants of health because work and employment relations play a major role in most people's lives. 6 Because they are unequally distributed by social class, gender, ethnicity/race or migrant status, these conditions also make a substantial contribution to health inequalities. 7 While working conditions have received considerable attention as social determinants of health and, to a lesser extent, of health inequalities, 8 researchers have rarely focused on employment relations, that is, the power relations between employers and workers and the level of social protection that workers can count on. The employment conditions knowledge network of the Commission on Social Determinants of Health classi fied employment conditions into six main categories: unemployment, precarious employment, informal employment, child labour, slavery/bonded labour, and fulltime permanent employment, and found that countries whose governments favour fair employment and decent work policies also tend to have better health indicators and lower health inequalities. 7 However, knowledge on t hose employment-related health inequalities remains limited or is neglected, and their monitoring over time is limited to a few indicators of employment conditions and for economic -not public health-purposes. --- Surveillance gaps: precarious employment, informal employment and migration as examples Precarious employment encompasses unstable employment, individual-level bargaining over employment conditions , low wages and economic deprivation, limited workplace rights and social protection, vulnerability and unbalanced workplace power relations, and powerlessness to exercise workplace rights. 9 Today there is a lack of indicators for precarious employment surveillance beyond standard indicators, such as the percentage of temporary contracts, 10 which can turn out to be highly misleading. In fact, during the current economic crisis the share of temporary employment in the Spanish labour market has fallen rapidly and is currently at its lowest in 20 years , reflecting a deep deterioration in labour market conditions rather than a decrease in the prevalence of precarious employment. The recent development of an Employment Precariousness Scale, a theory-based multidimensional questionnaire specifically devised for epidemiological studies 11 Organization gathers data on i nformal employment, but this information is limited. 13 First, countries can insert their questions on informal employment within one of multiple survey types, rendering data not necessarily comparable between countries. Second, countries can differ in their definitions of informal employment, geographical areas and branches of economic activity covered, informal enterprise cut-off size, registration criteria, criteria for inclusion or exclusion of secondary jobs in the informal sector, age limits, and whether or not to include professional or technical activities, among other criteria. 14 So, while it is estimated that informal employment represents 41% of total employment, meaning nearly two billion people only in low income countries, 15 there is a lack of reliable prevalence estimations concerning informal employment and its impact on he alth and health inequalities, especially in rural settings and poor countries. Efforts must be made to standardise an informal employment definition in low-income countries, capable of capturing informal jobs within and outside the informal economy. Indeed, the development of informal employment health-related inequalities surveillance systems will require extensive collaboration between national statistical offices, researchers and policy makers. 13 Employment and working conditions faced by most migrant workers are dangerous to their health. With 'globalisation', investments, and labour has been increasingly mobile. 6 International migrants were estimated at 214 m illion in 2010, of which about 90% are migrant workers and their families. 16 In 2009, 40% of EU residents were born outside the European Union. 17 The 'pull' of workers trying to access higher wage labour markets is compounded by t he 'push' factor of corporations seeking less regulated labour markets. 18 A major gap in international migration research is the lack of consistent, comparable data across countries. More global health surveillance and socio-epidemiologic analyses of migration are needed to render employment conditions prominent in migration policy. 19 A key challenge concerns the need for high quality data and information systems, including reliable estimates of international migration flows, national level data on the incidence and magnitude of migration, migrant workers' employment and working conditions and health status. 14 Oversampling of migrant workers and data-pooling are two approaches that can be used to address the statistical instability introduced by the small numbers of migrant workers in many studies. Overcoming these data problems will help to obtain better estimates of injuries, mortality and morbidity, and to properly monitor the evolution of immigrants, their health inequalities and health policy and prevention programmes. Researchers and policy makers would therefore be able to know whether the country's health averages are concealing large health differences between groups, and be able to formulate policies and interventions that are more equitable. 6 Moreover, researchers would be able to study the differential exposure to socially patterned risks and differential vulnerability-effect modification to these factors across groups. Both, differential exposure and effect modification are the main hypotheses to explain health inequalities. 20 Second, and when possible, interactions between social strata as determined by s ocial class, gender, ethnicity, age and migration status could be analysed, as intersectional approaches provide an excellent insight on health inequalities. 21 22 Specifically, for analysing gender inequalities due to employment conditions, family characteristics and household chores should be taken into account, as health inequalities cannot be understood without a work-family interaction insight. 6 23 Third, surveillance should be inclusive and cover all the population connected to employment: from standard employees to persons exclusively dedicated to reproductive labour and disabled persons. Fourth, some attention should be paid to choosing the most appropriate measures for health inequalities. 24 Finally, the most appropriate definitions and indicators for employment conditions such as precarious and informal employment have to be formulated, based on a theoretical conceptualisation for health purposes, and not exclusively for economic purposes. Today knowledge of what constitutes the most appropriate indicators for monitoring employment conditions in different social contexts, and the methods to choose them, remain limited to traditional labour market outcomes such as unemployment 25 but are very scarce for other employment conditions. To date, only a few countries have introduced data on e mployment conditions -mainly unemployment -in their public health surveillance systems, Sweden and the UK, being two good e xamples. 25 In spite of these limitations, ongoing employment and working conditions-related surveys are promising tools to monitor employment-related health inequalities. However, to be useful, these surveys must be representative of the population and large enough to analyse inequalities by ge nder, socioeconomic position and immigration status, they must be repeated over time in order to evaluate time trends and include health indicators to analyse the impact of employment conditions. However, the greatest challenge regarding ongoing surveys is being able to apply valid and conceptually sound measures, as exemplified above with the employment precariousness concept. 11 12 CONCLUDING REMARKS The surveillance of health-related employment conditions is necessary to determine their magnitude, evolution over time and most affected populations. The monitoring of employment healthrelated inequalities is important for two main reasons: to identify potential policy entry-points, and to assess the impact of policies Contributors JB and VP-B wrote the first draft. AV, GT and CM reviewed the draft and made substantial contributions to the paper.

Social determinants of health are major factors responsible for a population's health and health inequalities. 1 Public health surveillance was originally developed for the control of infectious diseases, but today its principles have also been applied to other public health problems such as chronic diseases and occupational and traffic injuries. 2 Many countries already collect data on social determinants of health, dispersed across different information systems typically designed for other purposes. However, most social determinants of health remain outside surveillance systems

Introduction Women are able to give birth! We as midwives know that-still. Midwives know that women have the capacities to give birth qua nature [von Natur aus]. And they trust in [women's] success. Because midwives know that a spontaneous birthing process [spontaner Geburtsverlauf] cannot be improved. Obstetricians tend to consider birth as something that can be made safer with the help of technological interventions than it would be without those interventions. The dominance of the medical concept of risk has led to high intervention rates and to increasingly pathologizing an originally healthy and natural vital process. -Martina Klenk, President of the German Association for Midwives At the triennial conference for midwives in 2016, Martina Klenk, then President of the German Association for Midwives , to which the majority of German midwives are affiliated, proclaimed a programmatic position for midwifery in Germany. She warned against an obstetric-technological authority redefining birth as essentially risky and marginalizing midwives, who, by contrast, simply allow 'natural' birth to happen. Klenk voiced a common juxtaposition: obstetricians argue for prioritizing the medical surveillance of and the corporeal changes that pregnancy, birthing, and postpartum stages bring, while midwives emphasize that those events form a 'normal' or 'natural' life phase, a definition that infers ''both a task and a way of belonging for midwifery, a profoundly normative claim'' . The German ''Law for Midwives'' sets forth these tasks that are also a claim: midwives are trained ''to give advice to women in pregnancy, during birth, and the postpartum phase; to provide the care necessary; to guide normal birth; to recognize any complication during birth early on; to take care of neonates; to surveil the postpartum period; and to document the course of birth.'' Giving advice, providing care, monitoring, and diagnosing are, however, activities that midwives share with their medical colleagues. As the German healthcare landscape is dominated by obstetric care, it is difficult for midwives to find their niche: irrespective of whether women are actually in need of obstetric treatment, most women are attended to by gynecologists and obstetricians during pregnancy. Then they give birth in hospitals. On clinical labor wards they are accompanied by both midwives and obstetricians, but obstetricians are the ones in charge. That birth in Germany is a 'medicalized event' is also indicated by the high cesarean section rates , twice as high as those assessed in Finland or Norway, for comparison . German midwives provide nonclinical assistance during birth and accompany women independently from obstetricians and throughout the whole trajectory, including pregnancy, birthing, and postpartum stages. But this is decreasingly the case: in 2015 it was estimated that in Germany no more than 1.3% of all births took place outside the hospital 1 ; in comparison, 30% of all births in the Netherlands were extra-clinical . Discourses that juxtapose obstetrics and midwifery easily camouflage that in practice, the tasks, approaches, and responsibilities of midwives and obstetricians overlap, and their respective remits are less well defined, as is repeatedly suggested . I therefore propose to understand midwifery care neither as a counterpoint nor as an accessory to obstetric care, but as a set of practices in which different interventions and ideals are combined in ways that fit the environment in which midwives attend women and their children, partners, and families during pregnancy, birth, and the weeks and months following birth. In this paper, I suggest revising the discursive position that designates midwives as being in charge of 'natural' or physiological births, in opposition to obstetricians who deal with pathological births; the practical realities of midwifery care emerge from a syncretic combination of repertoires that are informed by both discourses, creating a sociotechnical practice of skilled midwifery. I argue that it is important for midwives, and the women, children, and families they attend to, not to conjure the notion of a 'natural' birth, thus insinuating that giving birth is an innate physical event that requires as little intervention as possible, since this makes invisible both the work that needs to be done for giving birth, and the environment in which certain skills and knowledge repertoires are indispensable. Instead, I suggest concentrating on the specificities of midwifery practices in order to improve the understanding of what they actually involve. Against the backdrop of contested ''global and local politics of birth'' , encompassing rising obstetric intervention rates all over the globe as well as high rates of perinatal deaths of women and children in 'Southern' parts of the world , midwives and women could position themselves more strongly in relation to midwifery care and the alternatives articulated therein, particularly regarding births at home and in homelike environments. --- Anthropological Background Martina Klenk argues for a certain midwifery culture that allows 'natural' birth ''to run its course'' . In her conceptualization of 'natural' birth she stages birthing bodies as needing undisturbed-equated to un-technological and unmedical-surroundings in order for their 'natural' potential to unfold. Conceiving bodies as not only naturally universal, as suggested by midwifery discourses, but also as socioculturally specific, is done in anthropological inquiries of reproduction whose starting point is marked by Brigitte Jordan's 1978ethnography Birth in Four Cultures . Jordan states that birth is an universal event that is ''everywhere socially marked and shaped.'' This sociocultural notion of birth has been taken up fruitfully, resulting in a rich corpus of intercultural comparisons of birth practices . In early works in particular, 'modern' obstetrics were vehemently criticized for disrupting ''viable, healthy and culturally embedded indigenous systems'' , an idea that has also informed research questions of studies undertaken in so-called Western countries . 'Western', and more precisely American, obstetrics have been shown for producing medicalized or ''technocratic bodies'' . Similarly to bodies in ''traditional non-Western birthing system[s]'' , bodies giving birth in non-clinical midwife-led environments in the US have been conceived as 'natural ', knowing, and self-sufficient . Conceiving of births and birthing bodies as either belonging to a midwifery model that figures as natural or physiological, or to the medical model that makes giving birth risky if not pathological , not only keeps social and technological practices separate from one another, but also situates birth within birthing bodies. In this paper, I seek to pay attention to birthing bodies' physicalities without essentializing them or making them universal, but instead by situating birthing bodies in specific social and technological practices. To do so, I build on insights from science and technology studies -inspired anthropological research on new reproductive technologies. Anthropological scholars engaged with STS in order to account for the birth of the first child produced by in vitro fertilization, Louise Brown.2 They were and are still interested in studying new technologies' impacts not only on human reproduction but also on classical binary categories, such as nature-culture3 or, indeed, human-nonhuman, as well as on many domains of social life , including kinship and gender . Including technologies in non-binary ways in their analysis is also what Madeleine Akrich and Bernike Pasveer have done in their studies of obstetrical practices, which combine anthropology with STS. They argue that '''[n]atural' birth is not something that occurs all by itself'' but is a sociotechnical achievement. Their work shows that homebirth bodies are neither pre-existent nor unchangeable, and that their emergences are, among others, mediated by technical devices . I especially take inspiration from Akrich and Pasveer's work on obstetrical monitoring technologies that coordinate obstetrical practices and their participants with each other and with the system or 'paradigm' of Obstetrics. In this way, Akrich and Pasveer argue, technologies of surveillance multiply their objects-fetuses among others-and thereby also what Obstetrics are . Following Akrich and Pasveer , I would like to argue that interventions aimed at surveilling women and children, which midwifery shares with obstetrics, are not ''static and monolithic'' but are combined with other interventions and ideals in midwifery practices,4 and are thereby transformed into midwifery techniques.5 Considering that objects of obstetric technologies differ from one practice to another , I show how women's bodies are shaped by midwifery techniques in homebirth environments. I give answers to following questions: What techniques do midwives and women use in homebirth practices in order to make homebirth work? Which homebirth bodies emerge from those midwifery attendance techniques? --- Studying Midwifery Practices: A Praxiographic Approach The findings I present are based on praxiographic fieldwork conducted between February 2015 and March 2016 in most of the various settings in which midwives in Germany work. Interested in their care practices , I accompanied midwives in birthing centers, hospitals, and people's homes as they took care of women, children, and families during pregnancy, during birth, and during the weeks and month after birth. In addition to participant observation, I conducted twenty semistructured interviews with the women and midwives I accompanied. These interviews were initiated by asking my interviewees to describe in their own terms a particular attendance situation in which we had both participated. The praxiographic position I take allows for considering materialities and techniques, in this case those that are involved in homebirth midwifery practices and that allow for bringing homebirth bodies into being. The aim is to present complex situations in which social and material entities and events are not separated. In order to do so, it is necessary to decide not only who but also what is part of the situation in which ways. Homebirth bodies are not presumed to precede the practices of giving birth in which they are involved, but they are enacted in those practices in specific ways. For mapping out midwifery practices that are specific to the environments in which they take place, I moved between different sites and care phases in order to draw out similarities and contrasts of ''local knowledges'' . The midwifery practices that I describe are situated: they are related to specific times and spaces; to the women and midwives I encountered; to the German healthcare system; and not least to my own position as a researcher and a trained midwife. Engaging with different sites allowed me to discern overlapping and differing ways of attending births, to pay attention to overarching or recurring discourses, and to be sensitive to how these discourses relate to different practices. These sensitivities also inform this article, in which I focus on homebirth practices. My focus on homebirth practices should not distract from the larger context in which, not only through sharing certain devices and techniques but also through partitioning tasks and responsibilities, homebirth practices are tightly associated with clinical birthing practices. --- Midwifery Techniques: Routinizing and Multiplying Obstetrical Interventions Midwife Jana explains that when she started to provide homebirth assistance after her midwifery training6 she appreciated having received a ''good medical training'' that prepared her well. Like medical doctors in German spa settings who provide orthodox medical remedies next to alternative therapies , Jana emphasizes that she ''expanded her repertoire'' by taking courses in acupuncture and homeopathy, and using obstetrical as well as alternative diagnostics and therapies in her work. Jana's midwifery practices are syncretic as they combine different knowledge and skills in specific ways. But how do midwives do that? How are practices that do not necessarily cohere made to fit together in more or less coherent ways? Midwife Julia illustrates the broad scope of the work done by midwives: My work is quite diverse: I attend women during pregnancy, which means I am always there for them if they have questions, worries, or fears. But I also do prenatal screening, this specialist field. And I do births from A to Z. This means I attend to women at home or in the birthing place. And I also attend to couples in their being-parents. And I make sure that the baby and the mother and the father are well. I am a temporary friend, I do lactation counseling, I do nutrition counseling, I do partnership counseling, I do how-to-handle-yourmother-in-law counseling. So there is a lot more to it than just checking. Julia makes clear that when accompanying women during pregnancy and birth, interventions carried out for surveilling pregnancy and birth are important, even pivotal activities. However, they belong to ''this specialist field'' that midwifery shares with obstetrics, which is surveillance. She emphasizes that ''there is a lot more'' to her work ''than just checking,'' a statement that decenters monitoring as the main activity in her homebirth attendance. Instead, monitoring is one of many other ''diverse'' activities: assuring women, assisting birth, accompanying 'becoming family', giving professional advice and becoming a ''temporary friend'' who is ''always there.''7 According to Julia, ''attending to'' women, couples, and families encompasses taking care of the physical, emotional, and social dimensions of being pregnant, giving birth, and taking care of a newborn. ''Checking'' women's and children's physical state becomes merged into the multifaceted attendance repertoire Julia uses in order to configure pregnancy as a physical, emotional, and social becoming. Tina, who has been attended to by Julia during both of her pregnancies and births in the midwife-led birthing center and at home, appreciates how Julia handles the surveillance of her pregnancy: What I like about Julia is that she … focuses on how I feel: ''If you feel well then usually your child is well, too.'' And she still checks urine and blood values where you can also find out a lot. You can have a relaxed pregnancy. You don't have to worry too much even if the urine is perhaps not so good sometimes. While Tina knows well the screening procedures and what they aim at, she is not supposed to ''worry too much'' about the results as they generally do not affect her pregnancy. Tina can be ''relaxed'' because her midwife guides her in focusing on how she feels, while Julia assumes the tasks of evaluating the measurements and acting upon them if necessary. Tina's statement shows that this is a mutual, codependent endeavor: she needs to collaborate by following Julia's guidance to focus on feeling well independently of what obstetric markers might tell. Both Tina and midwife Julia emphasize that they do not do away with ''screening'' or ''checking.'' They assert that obstetrical surveillance is an important ingredient of homebirth care, by, paradoxically, simultaneously separating it from and inserting it into the larger picture of homebirth attendance. Monitoring gets separated from homebirth care by becoming articulated with a ''specialist field'' and with ''worries,'' and it is inserted into care, becoming one of many interventions within the homebirth midwifery repertoire and the monitoring of obstetric markers that is delegated to the midwife. The midwife-led birthplaces that I observed and the procedures of prenatal care visits that took place mostly in these places, indicate ambiguous ways of handling devices and interventions that also belong to obstetrics. Obstetrical devices are kept apart, but they are also introduced in self-evident and incidental ways. These birthplaces are warmly colored rooms with carpets, curtains, cozy couches and beds, wooden cupboards and tables, plants and candles. They are not arranged in accordance with practical or hygienic criteria only, but evoke an atmosphere of comfort and personal privacy. They are rooms to live in: the famous gynecological chair often placed prominently in the middle of the rooms in ob-gyn practices is missing. Syringes and other frequently used utensils are hidden away in drawers. The contact between women and midwives, especially at the beginning of visits, resembles more that of close friends than how patients and health care professionals interact. Midwives and women8 hug each other in greeting, saying things like ''Good to see you. Come in. How are you?'' The answers women give to that initial question are often elaborated, women working through their daily life in detail: how tired and moody they feel, how they wake up regularly in the middle of the night from the hiccups of the child in their belly, how they had a silly dispute with their partner about the stroller they wanted to buy, how they are desperately searching for a bigger flat. These conversations continue throughout the visits, frequently turning into consultations and then again becoming mundane causeries. In the course of these chats, not only do midwives and women get to know each other but midwives also 'use' them to get what they call a ''general impression'' of how the woman feelstired, anxious, or relaxed-and her physical condition: her gain in weight, the growing of her belly, her back pain or swollen ankles. Prenatal screening tests 9 are introduced in ways that are nearly unnoticeable : chatting does not need to stop to measure a woman's blood pressure. She can stay seated while Julia grabs the blood pressure monitor from one of the wooden shelves next to the couch. As she knows the procedure, there is no need to advise her to roll up her left sleeve so that Julia can strap the cuff around her upper arm. After a moment of silence in which Julia listens to the pulse and checks the monitor's dial, she comments: ''120 to 85. Everything is fine.'' Then she continues to inquire about the couple's last vacation. Whenever blood pressure values are slightly higher than usual or higher for the first time, Julia reads them as the body's expression of the current events in women's lives, remarking: ''No wonder that your blood pressure is high with all the things you have to deal with at the moment.'' As Julia gets to know the woman, she can articulate a high blood pressure reading with women's mundane troubles. Since high blood pressure does not have to become the object of medical treatment, it can be addressed along with all those other onerous affairs that need to be dealt with anyway. 10 By placing screening interventions into a home-like environment, by making them part of the visit routines without singling them out, screening interventions get routinized. In this case, routinizing is a technique that makes surveilling practices fit into women's daily lives, letting obstetric markers become signs of more or less ordinary events. This creates a precarious and provisional coherence: as they are being related to the daily activities and concerns, obstetric markers and screening interventions become non-specialized and non-singular. Not all surveilling practices are routinized, however. Differently from measuring blood pressure or body weight, feeling women's bellies in prenatal care requires the attention of both women and midwives to the intervention itself. In obstetrical terms, feeling a woman's belly is defined as abdominal palpation, carried out in order to determine the height of the uterus, fetal growth, and the amount of amniotic fluid, as well as the position of the fetus in the womb. In homebirth care, midwife Clara suggests, the practice allows palpation of both the woman's belly and the 'child-in-the-belly', and allows the two women to become familiar with each other: If you attend women in the beginning, you can hardly approach them. This means you cannot feel a lot [when touching the belly]. As the pregnancy progresses and the woman gets more open, because she knows you better, the easier it gets to feel how the child lies in the belly. The more you get the feeling that women open themselves up to you and allow you to approach. 9 Those screening tests are mandatory elements of prenatal care visits and encompass taking blood samples, analyzing urine, measuring blood pressure, palpating the woman's belly, and listening to the child's heart. 10 However, the situation changes if blood pressure is extraordinarily high the third time in three days, and the woman's urine contains proteins as well. Midwife Julia reads these as possible signs of a severe pregnancy disorder and sends the woman to the hospital where she gets continual surveillance and medical treatment. Monitoring then comes to the fore and obstetric markers become decisive. At that point ''also doing those medical-specialist things'' turns into the promise that as a midwife Julia knows when to act upon pathological processes, and that making sure that woman and child are well in obstetric or medical terms is also part of her work and the ideal of building trustworthy relations. Palpating becomes a midwifery technique by being multiplied into feeling. Feeling the woman's belly does not only serve to determine the child's growth and position but also to build up trust. Through feeling the belly, both midwife and woman turn towards each other and meet each other with mutual attentiveness and interest. Feeling 'more' or 'better' is an achievement that results from mutually engaging in the procedure. The different purposes and results of the feeling practice, surveilling and establishing trusting relations, do not exclude one another, but become mutually stabilizing. The midwife's and the woman's bodies, but also the midwife's and the child's bodies, get increasingly familiar with each other through touching and feeling, and becoming familiar authorizes a more detailed obstetrical examination of the child. When diagnosing is combined with getting to know each other, relating personally and intimately helps to do better diagnostics. Similar to what Tina explained, midwife Clara makes clear that both she and the woman she works with need to engage with the midwifery technique when she says: ''the more open the woman gets, because she knows you better, the easier it gets to feels how the child lies in the belly.'' In multiplying, different goals and findings get aligned by means of amplifying one another. Obstetrical procedures such as monitoring or diagnosing are interwoven with homebirth midwifery practices. In the midwifery techniques of routinizing and multiplying, monitoring and diagnosing are articulated with a wider set of goals, namely attending to events that are physical, emotional, and social. --- Co-responding Homebirth Bodies The bodies of midwives and pregnant women need to learn to ''open themselves up'' to each other, to the techniques that are mobilized, and to the surrounding that indicates intimacy. Their bodies learn to be receptive to the specific repertoire that is offered and they learn to respond to it adequately. The following excerpt from my field notes illustrates how bodies learn to co-respond in the procedure of feeling the woman's belly in prenatal care: During her first prenatal care visit, midwife Lisa asks Angelique to uncover her belly and to lie down on the sofa. ''I would like to feel the child [nach dem Kind tasten],'' Lisa says, kneeling beside the couch. She puts both hands on Angelique's belly and remains in this position for quite a while. Tim, Angelique's partner, squats next to Lisa and stretches his head to observe her hands on Angelique's belly. Angelique lies still and straight, while Lisa starts to move her hands slowly, centimeter by centimeter. One hand rests while the other slides further. ''The child's head is here and the back…there.'' She invites first Angelique and then Tim to touch as well. To do so, she puts her own hands on theirs to guide them. Yes, they could feel something, both say with surprise. Both pregnant woman and midwife need to align their bodies in order to make feeling the woman's belly and the child-in-the-belly work. Angelique lies on her back and relaxes and Lisa gets close in a way that allows her to touch Angelique's belly with both hands. Midwife Lisa makes Angelique and Tim feel what she feels, thus acquainting them with her technique of feeling. Lisa's and Angelique's bodies learn to co-respond with the help of touching and feeling and with the help of the verbal cues Lisa gives. Midwife Clara explains further: It was our first real prenatal care visit today, the first time of seeing and touching Laura's belly, and being physical with that woman at all. I have to get more familiar with her. One can notice easily that we met only twice. It is really good that I can do the prenatal care visits regularly now so that I get a feeling for her and for her way of being physical. In prenatal care you get references of how the woman wants to be attended to during birth. Actually, in prenatal care we are working towards birth. Becoming increasingly familiar with Laura allows midwife Clara's body to refine her skills of adjusting to Laura's body. Laura's body gets used to midwife Clara's touching and learns to co-respond to it. Laura's ''way of being physical'' does not merely get disclosed by ''seeing and touching and being physical'' with each other, however, but is also being shaped with the help of Clara's recurrent interventions, as well as through the particular attention and interest she directs towards Laura's body. When Clara gets ''references of how the woman wants to be attended to during birth,'' the two women's bodies are enacted as co-responsive, which is what matters when they ''are working towards birth,'' as Clara characterizes the aim of prenatal care visits in homebirth midwifery. Midwives' and women's bodies learning to become co-responsive is considered necessary for succeeding in giving birth in homelike environments, as homebirth bodies need to permit being guided during birth. That guidance is distributed across the midwife, the pregnant woman, other attendants, and the homebirth environment that favors certain positions and interventions and hinders others. 11Who or what is taking the lead shifts frequently in the course of homebirths. The story of Ruth's homebirth illustrates that those dynamics build on bodies' facility with co-responding. Ruth has given birth to three children at home, and all three births have been attended by midwife Agnes. I describe an interaction I observed between Ruth and Agnes in the following excerpt from my field notes: It is late in the evening when I meet midwife Agnes at Ruth's home. Ruth called Agnes earlier to tell her that she was having contractions every ten minutes and they decided that Agnes and I should set off now. Agnes does not want to be helped when she carries three heavy bags as we enter the flat. We join Ruth in her bedroom, which is well prepared with different items on which to sit, lie, and stand: the bed, chairs, a mat, and a gymnastic ball. Agnes settles in one of the dark corners of the room. She unpacks some papers, her Doppler ultrasound, a pen, and even a small headlamp that she uses when writing. Ruth breathes lightly and moves her pelvis in circles. Ruth's partner Paul serves us some tea and Agnes and Ruth chat about someone they both know in between Ruth's contractions. Ruth's breathing becomes heavier. ''You do whatever is good for you. If we bother you, you send us out,'' Agnes remarks. Ruth walks in circles. During her contractions she puts her hands on Paul's shoulders and moves her hips. Agnes continues writing, now and then taking a glimpse of Ruth. ''You are doing well. It is hard to recognize that you are about to give birth; you are all calm and concentrated.'' Then she picks up the Doppler and holds the transducer to Ruth's belly. We hear a slight throbbing. ''It [the child] is all calm as if nothing happened. It is sleeping now.'' Both Ruth and Agnes are familiar with the easy chatting as a technique that marks Agnes's presence. In this situation, the chatting routinizes Ruth's labor and Agnes's attendance, which also consists of surveilling, observing, and valuing Ruth's birthing body and her way of handling it. When listening to the heart sounds of the child-in-the-belly with the help of the Doppler fetal monitor, midwife Agnes multiplies the measuring of the child's heartbeat into connecting Ruth and the child: It is ''all calm as if nothing happened,'' just like Ruth herself who is ''all calm and concentrated.'' Similar to Tina's midwife, Julia, who emphasizes that if Tina feels well, then usually her child feels well too, Agnes guides Ruth in concentrating on herself and on her body, while Agnes checks on the child's well-being, suggesting that this is a good strategy for dealing with birth at that point. Ruth's body is enacted as a body that is able to co-respond to the undisturbed homebirth surrounding, to the techniques mobilized, and to Agnes' body. Exactly because Ruth's body has learned to be co-responsive, it is susceptible to our presence in a way that might affect the birthing process, which is now working well. Giving birth thereby becomes a physical but also emotional and social endeavor. During the course of Ruth's birth, her body-in-labor has to meet other expectations than in the beginning. Ruth's body position shifts: it needs to co-respond by allowing for being directed more actively by midwife Agnes towards giving birth. My field notes continue: Ruth lies down on the bed next to Paul. Midwife Agnes and I sit on the floor and whisper next to them. Then Agnes approaches Ruth again, the Doppler in her hands. Ruth and Agnes take a short look at each other, and we listen to the child's heartbeat, inalterably calm and steady. After a while Agnes asks Ruth if she might do a vaginal examination in order to find out if the cervix has dilated further. Ruth nods and turns to lie on her back. While examining Agnes asks Ruth if she wanted to get a bath or a belly massage with an essential oil to augment labor. Ruth opts for the massage. Then she lies down on the bed, but gets up again after a short while: ''What should I do?'' Ruth continues standing and then sits down on the gymnastic ball. ''You don't want to [give birth] yet, do you?'' Agnes asks Ruth in a surprised tone. Ruth smiles tiredly, her eyes half closed: ''Well…'' Agnes insists: ''You have to allow the child to come!'' After a short silence Ruth asks Agnes to tell her if she has to change position. ''We don't have to hurry, but you could alternately sit and stand.'' Agnes massages Ruth's lower back again and Ruth leans into Agnes's hands. Agnes listens to the child's heart again. ''Can you feel that it is far down?'' Ruth squats down, supported by Paul. Agnes, putting on her gloves, kneels in front of Ruth, looking at Ruth's vulva opening wide during the contraction. ''Yes, you can do this! You can overcome the threshold! Very good!'' While it was important to find a rhythm of moving and breathing as strategies for staying ''calm and concentrated'' at the onset of giving birth, it is now important to give birth within a certain time span. As a result of the vaginal examination,12 in which Agnes feels that Ruth's cervix is fully dilated, Agnes concludes that augmenting labor is necessary in order to help the birthing process to proceed. She offers several guidance techniques to Ruth's body-in-labor and takes a leading role in directing it: Agnes proposes that Ruth alternates body postures in order to facilitate the descent of the child. While applying the oil to Ruth's belly aims at stimulating contractions, massaging Ruth's back is expected to relieve pain. Both procedures also imply feeling as a means to convey and detect intimacy and trust. Agnes continues to enact Ruth's body as a body that has learned to co-respond, that can follow the guidance she procures. When taking the lead, Agnes does not exclude Ruth but invites her to continue to take part while also acknowledging her hesitation and fatigue . However, Ruth's body struggles with coresponding to the techniques aimed at accelerating birth. Ruth tries to collaborate but she is in pain and exhausted and thus not readily available to the attending techniques. In response, Agnes invites Ruth to engage with Agnes's efforts by facilitating Ruth's motivation and through encouraging Ruth's ''wanting'' and actively ''allowing the child to come'' by letting her body-in-labor keep going. Both Ruth and Agnes know that if Ruth's body does not co-respond to the ''gentle'' ways of stimulating labor, as Agnes and her colleagues call it, Ruth will have to be transferred to the hospital. This, however, does not become necessary, as Ruth and Agnes manage to co-respond in this environment: Agnes: ''Ruth, do you want to walk a bit? Was this most effective? What's making you hesitate? You seemed close to falling asleep now.'' Ruth: ''Yes, that is exactly how I feel.'' During the next contraction, Agnes puts two fingers in Ruth's vagina, and presses softly downwards. ''Yes. Push in that direction. Your child wants to come out now! You can do this!'' Squatting in front of Ruth, Agnes and I can see the child's head appearing between Ruth's labia. In between the contractions, Ruth closes her eyes and lets her head fall. Agnes, however, is strained. She does not take her eyes off of Ruth, encouraging her from the very beginning of every contraction. After three more contractions, the child is born. Agnes wraps the child in the towels that Paul has warmed in the oven and grabs her delivery instruments in order to clamp and cut the child's cord. Through Ruth's birthing story, I have demonstrated how, in the process of coresponding, a mutual and embodied engagement in learning to skillfully respond to the midwifery techniques is foregrounded. The intimacy and trust that are established in the course of becoming acquainted with one another, with the techniques mobilized, and with the home environment are crucial for making the midwifery attending techniques work. Similarly, in the attendance of Lisa's birth, techniques of routinizing and multiplying surveilling interventions are employed in order to allow her homebirth body to co-respond and to give birth in these surroundings. Lisa is accompanied by her family when she arrives at the birthing center on this late Friday afternoon. As Anna, her colleague, and I are still busy with taking care of Mira, who had just given birth an hour ago, we ask Lisa and her family to make themselves comfortable in the second birthing room. When Anna and I enter the room twenty minutes later, Lisa's mother is sitting at the table knitting. Next to her sits Lisa's sister, with Lisa's eldest child on her lap reading a story aloud. Lisa and her partner stand next to the chest of drawers which is the perfect height for Lisa to lean on during her contractions. She is breathing deeply and silently, moving her hips from one side to the other. ''Is everything ok?'' asks Anna. ''Yes.'' Lisa looks up once the contraction is over. ''Do you want to take a bath?'' Anna asks. Lisa shrugs her shoulders: ''Now? I don't know…'' Anna: ''Should we take a look at the cervix?'' Lisa: ''Yes!'' During the next contraction Lisa cannot talk anymore. She closes her eyes and continues breathing deeply. Anna: ''That is already in full swing.'' ''Yes, I think so, too.'' Lisa sighs with a smile. Anna asks if Lisa wants her family to leave the room while she makes the vaginal exam. As they are leaving the room, Lisa lies down on the cot. Anna sits next to her and leans between Lisa's legs. ''Please spread your legs a bit more. I have warm fingers and I will be very careful. Now I am taking a second finger.'' Anna hunches forward a bit further. ''The child's head is already far down. The cervix is dilated to seven centimeters'' Anna says contentedly, pulling back her fingers. ''So it is in full swing! I will run the water in the bathtub!'' Anna turns to lie on her side during the next contraction. Lisa waits for the contraction to cease: ''How do you want this? Who is going to stay? I just had the impression that it was a bit difficult with everyone around.'' Lisa looks at Anna. ''Yes, they have nothing to do. I will talk to them.'' Anna: ''Ok. So you can get up or do whatever you want. But now that you are lying down, I will quickly listen to the child's heartbeat.'' Midwife Anna uses several techniques to routinize her surveilling interventions during Lisa's birth. Firstly, she emphasizes the ''impressions'' she gets from being with and observing Lisa. Based on her impression on how Lisa breathes and moves, Anna identifies that the birth is ''in full swing,'' which means that Lisa's contractions are efficient and that the child will probably be born soon. The vaginal examination that Anna proposes, and to which Lisa agrees, only confirms Anna's and also Lisa's opinions on the progress and stage of birth. It is neither the most important nor the only source of information. The examination is proposed only after the bath, suggesting that it is not the most urgent intervention, but one of several ways to proceed. Furthermore, Anna invites Lisa to become acquainted with the vaginal examination that she carries out. She offers suggestions for how Lisa will experience the examination, namely as comforting: Lisa will be touched carefully and with warm fingers. The midwife also shares the obstetric markers with Lisa, apparently in order to concretize what the birth being ''in full swing'' signifies in obstetric terms. Anna thus signalizes what also matters in homebirth environments: that how women and midwives evaluate the birthing body matches with surveilling parameters. Routinizing is done by privileging interventions that aim to keep up with Lisa's birthing body through creating situations that are deemed not to disturb or to interrupt its activities, but to support it in its ''swing.'' Lisa is offered to take a bath in order to create a situation that would allow her to give birth soon. In this situation, the bath does not offer a mere possibility for soothing labor pains, but it allows the creation of undisturbed and intimate surroundings. Lisa's family seemed busy with knitting and reading stories, but as Lisa pointed out, actually ''they have nothing to do'' besides waiting for the child to be born. Anna refers to the pressure that this waiting might exert on Lisa, when she shares her impression ''that it was a bit difficult with everyone around.'' In this situation, water birth is multiplied into a midwifery technique that aims to facilitate birth through not only relieving pain and maintaining contractions within an individual and bounded body, but also by arranging an intimate environment in which Lisa-and certainly also midwife Anna-have fewer 'exterior' expectations to meet and may thus concentrate more easily on Lisa's body giving birth. Ten minutes later, Lisa, her partner, and her sister enter the bathroom that Anna and I have quickly prepared. Sitting in the tub, Lisa breathes quite fast during her contractions, that have obviously become stronger. ''You are doing great! Try to breathe a bit slower'' Anna advises. And when the contraction is over she asks: ''That contraction surprised you, didn't it?'' Lisa: ''These were the first explosive pains, I think. But I have the feeling that it does not fit yet.'' Anna: ''Then you still have one or two contractions to get used to it. Your baby needs to be patient. Make some space! Very good!'' After three more contractions, the baby's head is visible between Anna's labia, also in the pauses between contractions. Anna guides Lisa's hand to the baby's head. ''You can touch it, then it might dare to come.'' In the next contraction, the child's head is born. ''That is great! Perfect! Yes! Let your back fall. Leave some space! Super! Ok, now just press a tiny bit.'' After two more contractions, the child is born. During Lisa's birth, the co-respondance between Lisa and midwife Anna is repeatedly enacted through Anna sharing her impressions of Lisa and her birthing body, as well as Lisa sharing and confirming these impressions. It is obvious, for example, that Lisa is not particularly keen on taking a bath at first, and however she understands and co-supports the aim of the intervention. Lisa shares her worries that the expulsive pains ''do not fit yet,'' that she is not yet ready to give birth. In response, Anna grants Lisa more time and guides her through the last contractions: ''Make some space'' and ''Now just press a tiny bit.'' As Anna co-responds to Lisa's bodily and verbally expressed needs, and as Lisa manages to follow Anna's interventions and to trust that they are helpful, giving birth in this homelike surrounding works well. --- Conclusion: Enacting Homebirth Bodies In this study, I asked what techniques midwives use in homebirth practices, and which versions of bodies emerge from these practices. In midwifery attendance techniques, several knowledge and skill repertoires are combined in order to configure pregnancies and births as physical, emotional, and social becomings. Quite some efforts are invested into rendering coherent the different approaches in homebirth attendance. As important ingredients of homebirth practices, interventions and devices that overlap with those used in obstetric surroundings get backgrounded. They become midwifery techniques through being routinized and multiplied in order to foreground what matters here: learning to co-respond to each other, to the attendance techniques, and to the homebirth environment. This is made possible not only by midwives assuming responsibility for surveilling and acting upon obstetric parameters but also by women collaborating through handing certain responsibilities over to the midwife and engaging with the attendance techniques offered. Midwifery techniques are used to align women's and midwives' bodies. Homebirth bodies are not enacted as stable and bounded, but are granted capacities of incorporating what is offered to them, of allowing themselves to be guided. I showed that during birth at home, coresponding bodies become prominent in order to make birth in that environment work. To conclude my analysis of homebirth practices in Germany, I use two insights I borrow from a 'technological' strand of the anthropological research of reproduction as a ''looking glass'' that helps us to see a more differentiated and situated picture than that provided by the earlier introduced 'sociocultural' strand, in which homebirth has been traditionally located and studied. I showed that obstetrics and midwifery are not respectively monolithic fields, each determined by its own philosophy, its own body of knowledge, and its own way of engaging with pregnancies and births , despite what the German Association for Midwives and the sociocultural strand in anthropology argue. Instead, attending to the specificities of midwifery and obstetric practices reveals that they are deeply entangled. When describing homebirth practices, I argue that they differ and overlap in terms of what apparently similar procedures conducted in midwifery or obstetric environments such as measuring blood pressure or palpating women's bellies become, what concerns they attend to, which knowledge they mobilize, and which versions of bodies they bring to matter. Different knowledge repertoires can be combined synergistically, for example when 'becoming familiar' makes a more thorough physical examination possible. But as obstetric knowledge is frequently backgrounded during prenatal care, it might create ambiguities when it becomes decisive during pregnancy and birth. This analysis avoids predefining and generalizing that female bodies are either 'natural', knowing, and sufficient or 'technological' and fallible. Instead of pinning down the 'natural,' often synonymous with 'biological' or 'physiological,' onto women's bodies 13 , 14 I propose paying attention to the practicalities, the sociomaterial conditions under which particular versions of bodies emerge. Bodies are cultivated : they need to gain specific skills in order to fit in homebirth practices, especially that of allowing to be guided, which includes being objectified occasionally. 15 Different versions of homebirth bodies are sometimes difficult to combine, which creates ambiguities. Those ambiguities become especially prominent if the body has to live up to obstetric standards while also being trained to gain a certain independence from obstetric markers. In material semiotics, classical nature-culture distinctions have been shown to be ''politics by other means'' 16 : they serve to yield and privilege certain realities while driving back others. Discourses that limit homebirth practices to the natural and non-interventional risk narrowing down the possibilities of being pregnant and giving birth but also the diversity of ways of attending to women and children in pregnancy and birth. Instead it is worth to follow them in detail in order to show how and with which aims midwives and women invest in those long-term engagements that are specific to homebirth surroundings but that could also inform clinical practices. ---

Building on insights from science and technology studies-inspired anthropological research on reproduction, this paper uses a praxiographic approach to analyze homebirth midwifery practices in Germany. I show that such practices are syncretic, and that techniques of routinizing and multiplying obstetrical interventions are combined in more or less coherent ways to configure pregnancies and births as physical, emotional, and social becomings. In the process of attending, homebirth bodies learn to co-respond to each other, to the midwifery techniques, and to the homebirth environment. Understanding how and with which aims midwives and women invest in those longterm engagements specific to homebirth surroundings may inform clinical practices.

Introduction Mood disorders during pregnancy and the postpartum period are potentially devastating conditions that affect a significant number of women and their offspring. Research suggests that 80-85% of expectant mothers will develop a mild mood disturbance during or after their gestational period and 8-15% will experience a major depressive disorder . Hallmark symptoms of these disorders include loss of interest in previously pleasurable activities, feelings of guilt, worthlessness, helplessness and hopelessness that are severe and impairing and last for at least 2 weeks. Among adolescents, this rate is even higher, with estimates of antenatal depression ranging between 16 and 44%, almost twice as high as among adult pregnant women and non-pregnant adolescents . Moreover, prospective studies indicate that elevated levels of postnatal depression are preceded by high levels of prenatal depressive symptoms . This research suggests an advantage of intervening during pregnancy to reduce morbidity associated with depression. Elevated depressive symptoms and depressive disorders during pregnancy and postpartum not only affect women but are associated with a host of negative outcomes for their offspring including low birth weight, premature delivery, delivery complications, low fetal growth, and poor performance on neonatal behavioral assessments . The negative impact of depressive symptoms and disorders extends beyond birth outcomes to mother-child bonding and attachment, parenting, and later child functioning . In light of the prevalence and consequences of prenatal and postpartum depression, there is a critical need for both treatment and preventive interventions. Preventive interventions in particular hold the promise of decreasing the likelihood that expectant mothers will experience MDD during pregnancy and postpartum, minimize the need for costly treatment, and ultimately decrease the long-term impact of depression on both the mother and child. Unfortunately, there are few published studies examining the efficacy of preventive interventions during pregnancy using randomized controlled designs; even fewer have been conducted with adolescents, and none have focused on American Indian populations. Although findings have not been uniformly positive , results from several studies suggest that psychosocial interventions, namely interpersonal psychotherapy and cognitive behavioral therapy may reduce depressive symptoms during pregnancy as well as postpartum. For instance, in a pilot study, Zlotnick et al. evaluated the effectiveness of a four session IPT-oriented intervention with pregnant women receiving public assistance. The authors found that at three months postpartum none of the women in the intervention group, compared to 33% in the control group , met criteria for MDD. Similar findings were reported in a larger study . Elliott et al. randomly assigned "vulnerable" pregnant adults to an 11-session group intervention or TAU. Among first time mothers, depression scores on the Edinburgh Postpartum Depression Scale and rates of depressive disorders were significantly lower in the intervention compared to TAU group at a 3 month follow up. While these findings are encouraging, more data are needed to determine the usefulness of preventive interventions for postpartum depression, particularly among vulnerable populations. The current study was designed to extend this literature and to address the needs of one of the most vulnerable populations in the United States, reservation-based expectant AI adolescents and young adults. Reservation-based pregnant White Mountain Apaches were targeted for this study because they are at high risk for the development of depressive symptoms and disorders . AI adolescents in general have the poorest health status among adolescents in the US, as measured by rates of premature mortality and morbidity and high rates of mental illness . AI adolescents also have significant demographic risk factors for depression such as poverty, high substance use and school drop-out rates, and residential instability, that compound normal stressors for pregnant teenagers . These health and demographic disparities are related to both historical and contemporary factors that impact the lives of AI youth and their families. First, a long history of federal policies relating to land subjugation and tribal community relocation has resulted in the loss of tribal territories, degradation of Indian political and economic systems, languages, traditions and cultures; in essence, the destruction of the social fabric of Indian communities, which are now being rebuilt. Second, but not unrelated, day-to-day life for AI families is difficult. AI youth grow up in communities with serious problems of unemployment, poverty, challenged educational systems, and stressful home lives plagued by broken nuclear and extended family networks. Consequently, the needs of this population are significant. The goal of the intervention, entitled Living in Harmony , was to reduce depressive symptoms during pregnancy and prevent the onset of MDD postpartum. The intervention was theoretically-derived, based on a cognitive behavioral therapeutic approach, culturally adapted, and delivered by AI paraprofessionals. The preliminary effectiveness of LIH was evaluated using a randomized controlled trial comparing LIH to an eight lesson education-support comparison condition . Both interventions were comprised of 8 lessons and were delivered during pregnancy . Assessments were conducted at pre and post-intervention and at 4, 12, and 24 weeks postpartum. Our primary hypothesis was that LIH would result in greater reductions in depressive symptoms and fewer women diagnosed with MDD. In addition, because social support has been linked to lower levels of depressive symptoms enhancing social support was targeted in the LIH intervention. Thus, we hypothesized that women in LIH, relative to ES, would report greater improvement in social support. Finally, consistent with previous research examining the impact of treatments for depression , we hypothesized that women in LIH, relative to ES, would report superior functioning as a result of the CBT skills they learned. --- Method Participants Participants were pregnant adolescent and young adult Apache American Indian women. For inclusion in the study, participants needed to be age 15-19 years old, at a gestational age of 28 weeks or less and have a Center for Epidemiological Studies-Depression scale score of 16 or higher, indicating symptoms of depression at baseline. Participants were excluded if they met criteria for current MDD , met criteria for another psychiatric disorder or condition requiring immediate intervention, and/or were receiving psychiatric care for depression or any other mental or behavioral health problem. Sixty-one participants completed an initial assessment. Among them, 14 were not randomized due to either having a CES-D score below 16, needing immediate treatment, or they were unable to be located, leaving 47 randomized participants, of whom 22 were assigned to LIH and 25 to ES. --- Measures Center for Epidemiological Studies-Depression scale-. The CES-D is a widely used self-report 20-item depression-screening instrument assessing symptom frequency on a 4-point Likert scale . This measure has been well validated, and was chosen due to widespread use with adult and adolescent samples. Higher scores represented more depressive symptoms. Cronbach's coefficient alphas were 0.76, 0.78, 0.79, 0.70, 0.72 at pre and post-intervention and at 4, 12, and 24 weeks postpartum, respectively. The alphas were stable overtime. That is, they were not significantly different from each other based on the procedure outlined by Feldt et al. . Cox et al. 1987) is among the most widely used self-report screens for adult postpartum depression. It is a self-report instrument and was included in the current battery to evaluate its utility and acceptability for the study population. In addition, the use of this measure will facilitate comparisons between the current study and other studies on postpartum depression. Cronbach's coefficient alphas were 0.80, 0.79, 0.81, 0.80, 0.86 at pre and post-intervention and at 4, 12, and 24 weeks postpartum, respectively. These alphas were stable overtime. Shaffer et al. 2000) was used to assess for the presence of MDD and is a highly structured diagnostic interview designed for use by lay interviewers in epidemiological studies of children and adolescents. For the present study, only the MDD module was used as an outcome. The presence of a diagnosis was determined via a computer algorithm based on the presence, severity, and duration of symptoms. Studies indicate that this instrument has acceptable psychometric properties ). This instrument was specifically selected because it had been used with AI youth on other reservations and can be administered by paraprofessionals. --- Edinburgh Postnatal Depression Scale-(EPDS; --- The Diagnostic Interview Schedule for Children-Computer Version-(DISC; The Social Support Index-. The SSI assessed the degree to which participants felt supported by other people in their life. This self-report measure contains 20 items which were rated on a 5-point Likert scale ranging from 1 to 5 yielding a range of 20-100. Cronbach's coefficient alphas for this measure were 0.92, 0.94, 0.95, 0.97, 0.97 at pre and post-intervention and at 4, 12, and 24 weeks postpartum, respectively. These alphas were stable overtime. Global Assessment Scale for Children-. The CGAS, completed by an AI interviewer, is a modification of the adult Global Assessment Scale and provided a measure of current global impairment and functioning. The scale ranges from 1 to 100 functioning. The CGAS has been widely used in clinical trials. --- Intervention Descriptions Living in Harmony Overview-LIH adapted CBT curricula from the Coping with Depression and The Treatment for Adolescent Depression Study manuals . Modifications were made based on feedback from study staff, participants, and a local Apache Advisory Board. LIH Format-LIH included eight weekly, 30-60-minute in-home sessions initiated prior to 29 weeks of gestation and included three monthly booster sessions. The focus of each session was on teaching new skills and providing the opportunity to practice these skills both within sessions and at home . To help reinforce and generalize the skills, specific homework tasks were assigned at each lesson and a workbook was given to all participants. LIH Content-The main cognitive and behavioral skills taught included: 1. Psychoeducation: Participants were taught to recognize signs and symptoms of depression . They were also provided with important information about how such symptoms could be alleviated via the cognitive behavioral model. --- 2. Identifying and modifying depressive cognitions and behaviors: Participants learned to identify maladaptive cognitions and were taught strategies for adopting realistic, coping-focused thinking. Behavioral skills included increasing engagement in pleasurable and positive activities to improve mood. --- 3. Problem-solving skills: Problem-solving skills were taught to participants to help them plan effective resolutions to conflicts and other interpersonal/emotional difficulties. A six step model involved brainstorming solutions, generating alternative solutions, and examining outcomes of solutions. --- 4. Enhancing social supports: Participants were taught how to increase social support in order to reduce isolation and provide needed assistance. Participants also explored the role changes associated with becoming a parent and ways to access social support including community services. --- 5. Planning for the future: Modeled from relapse prevention strategies, participants were taught how to anticipate future challenges and practiced how to manage difficult situations using acquired skills. LIH Adaptations for AI Participants-Formative research with community stakeholders pertaining to the use of a CBT intervention with the study population supported the use of CBT as the method was consistent with values and practices that are regularly employed on the Reservation for the treatment of low mood and energy. For instance, the Apache people generally understand the role of past experiences as a determinant of current feelings and behavior and focus on how persons in the "here and now" can overcome past difficulties and move positively into the future. Within the Apache community, the term "depression" is not commonly used in the local vernacular. In fact, the constellation of physical symptoms, moods, thoughts and causes ascribed to "depression" in the western vernacular are not lumped under one word or related concept in the Apache language. Physical symptoms such as fatigue or changes in appetite would not be typically understood by Apaches as a function of mental illness. Emotions related to western definitions of depression, such as sadness or anger, are described somewhat differently in Apache, with more nuanced meaning; for example, there are two different words for "sad" or "really sad," as well as for "mad" or "really mad." However, there is an Apache word, "How'shi," translated as "somehow," which is popular in its English form among adolescents and young adults, that comes closest in its negative form to the western term for depression. If an Apache individual says he or she is "somehow," it means the individual is feeling markedly different than usual-off "center" from his or her usual mental, emotional and physical state. When an Apache tells a peer or parent he or she feels "somehow," it is expected to evoke probing to better understand how the individual is feeling. Cognitive distortions associated with western definitions of depression, and even common sarcastic or exaggerated statements that are accepted in the Western vernacular , are traditionally taboo for Apaches. Apaches are trained by culture and traditions to refrain from negative thoughts about oneself in the present or future, because they put one at risk for realizing one's thoughts. Furthermore, in the Apache tradition, the cause of what might be labeled as "mental health problems" is generally ascribed to external factors, which is inconsistent with some Western thinking about the causes of depression . For example, a person could have been "hexed" by another or could have inadvertently and unintentionally violated a tradition, which would invite bad spirits and related problems to perturb the individual. Some but not all of this understanding was present at the onset of the development of the LIH intervention. While a community-based participatory research process was followed to adapt and create the intervention, the evolving relationships and continuing communication among the JHU-Apache research team shed increasing light on the important differences in perspectives on depression and mental health across cultures. Fortunately, all LIH lessons were delivered by Apache paraprofessional Family Health Educators , who naturally navigated preferred cultural styles, accepted taboos, and expected ways one would relate to participants. --- Education-Support Condition -This condition is based on the Family Spirit curriculum used in our previous work, which focuses on pregnancy and parenting education . Study Interventionists and Adherence-The interventionists were AI paraprofessionals with an Associates or Bachelors Degree. FHEs completed extensive training and at least one pilot case prior to enrolling participants in the study. Each FHE delivered both ES and LIH in order to protect against the threat to internal validity . Because of the possibility that the FHEs may use CBT strategies in the ES condition, the use of highly structured manuals, audiotaping of sessions, random checking of intervention adherence, and weekly supervision were used. --- ES Format-ES --- Procedure Participants were referred to the study via the local Indian Health Service prenatal clinic or by word of mouth. Once in contact with study staff, participants and their parents signed a written consent form which described all study procedures. All study procedures were approved by the institutional review board at Johns Hopkins University and tribal councils. Following consent, all women completed the CES-D either in the research office or in their homes. Women who scored lower than 16 on the CES-D were ineligible and offered alternative services. Women who scored 16 or higher on the CES-D completed a baseline assessment. The baseline assessment occurred in a private office in the research trailer. During this assessment participants completed the measures above and were then provided with a computer to complete the C-DISC. Eligible participants, based on the baseline assessment, were randomized 1:1 to either LIH or ES. Any participant who met eligibility criteria, but appeared to have elevated risks was referred to the Apache Behavioral Health Services for an additional interview to confirm that immediate treatment was not necessary. At post-intervention and at 4, 12 and 24 weeks postpartum, participants completed identical evaluations to assess the impact of the interventions. Participants received Walmart gift cards as compensation for each evaluation completed. --- Data Analysis Plan With the exception of presence of MDD, all outcome variables were treated as continuous measures . To examine the differences in program effects on continuous outcomes, two analytical approaches were used. First, differences between the ES and LIH groups at each time point after the intervention were compared using analysis of covariance , adjusting for the baseline scores. We also included the baseline by group interaction in all of the ANCOVA models to test potential differential intervention effects by the baseline measure of the same variable. In the second approach, we examined the differences of the intervention effects over time utilizing all of the repeated measures of the outcomes at baseline, posttest, 4, 12, and 24-week follow-up together within a growth curve modeling framework. All analyses used an intent-totreat approach . Multiple imputation with 20 imputation datasets, using SAS MI Procedure, was performed for handling missing data for examining the intervention effect with ANCOVA and GCM. We also used SAS ® software for the analyses for ANCOVA and GCM. Because of the small sample size for the study, the findings should be interpreted with caution. In general, with categorical outcomes, researchers can test the effects at each time point through logistic regression and examine the differences over time using mixed models. However, because no individual in the LIH group developed MDD and only two in the ES developed major depression, these analytical procedures were neither appropriate nor feasible. Therefore, no analyses were performed for the MDD outcome. --- Results --- Descriptive Statistics Retention-The attrition rate for the entire randomized sample was 28% ; 32% in the LIH condition and 24% in the ES condition. Analyses were conducted to examine differences on demographic and clinical variables between the participants who dropped out of the study and those who completed the study. Participants who dropped out of the study were older but no other significant differences were found. Additional demographics appear in Table 1. Intervention Attendance-Among the LIH and ES participants, the average number of lessons completed was 6 and 7 respectively, with a range of 0-8 and 1-8 respectively. Baseline Group Comparisons-t tests for continuous variables and Chi square analyses for categorical variables were used to examine differences on demographic and clinical variables between the two intervention groups at baseline . No significant differences were found between the two groups on any baseline variable. Outcomes Analyses-Means and standard deviations on the CES-D and EPDS, and the number and percentage of individuals who developed MDD are presented in Table 2. Means and standard deviations on the SSI and CGAS are presented in Table 3. No analysis was performed for MDD; only 2 individuals in the ES group developed MDD across all assessment points. All participants who met diagnostic criteria for a disorder were referred for treatment. Tables 2 and3 show that there were significant improvements on the CES-D, EPDS, and CGAS from baseline to post-intervention and follow-ups for both LIH and ES participants. However, controlling for the baseline differences, the ANCOVAs showed no significant differences between LIH and ES on each outcome measure at each post intervention assessment point. No significant group by baseline interactions were found either. Growth curve models of the outcomes were tested under a multilevel framework using SAS 9.1 PROC MIXED. Two levels of data were included in these models, with repeated measures nested within individuals . The intervention condition was a level-2 predictor. As shown in Fig. 1, three different forms of the intervention effects were considered most plausible . One form of the intervention effect was a change in the growth rate linearly . The time scaling of the linear growth trajectory was set at 0, 2, 3, 5, and 8 to reflect the real time assessment points at pretest, posttest and at 4, 12 and 24 weeks follow-ups after intervention . A second plausible form of the intervention effect was a downward shift or an upward shift of the scores from pretest to posttest and the change of the score maintained over time . The time scaling of the growth trajectory was set at 0, 2, 2, 2, and 2 to reflect the stability of change over the follow-up assessments after intervention. The third plausible form was to simultaneously include both these forms of intervention effects . To show differences of the intervention effects, the rates of changes on slope, the magnitudes of shift, or both need to be significantly different across the two intervention conditions. For contrast, Fig. 1 also included Model 0 to show no change of scores over time for both LIH and ES conditions. We tested the shift and linear growth functions simultaneously to examine their unique and combined effects. The multilevel model equations of the fixed and random effects and the corresponding mixed model equation for testing the intervention effects are shown below: In these model equations, Y ti is the observed outcome score at time t of individual i. "Int" "Shift" and "Time" are the latent growth factors used to capture the growth parameters and intervention effects on these growth parameters. "Int" represents the initial status at the anchor time point, i.e., the pretest. Because it is always coded as "1", it can be omitted from the model equations. "Shift" represents the shift of the mean score at the posttest associated with the intervention . "Time" is associated with the linear growth rate for the intervention effect overtime. Significant ' ' interaction indicates the group difference of the magnitudes of shift at posttest and significant interaction indicates the group difference of the linear growth rate. e ti , r 0 , r 1 , and r 2 are the random effects. Table 4 shows the findings from the growth curve modeling, including the regression coefficients and standard errors of the parameters included in the mixed model equation. The results showed significant intervention effects for both LIH and ES groups on depression symptoms and global assessment of functioning . Both outcomes had growth patterns following Model 2, such that there was a downward shift for CES-D and an upward shift for CGAS from pretest to posttest and the changes were sustained over time. The linear growth rate was not significant on any of the other outcome variables. None of the ' ' and ' ' interactions were significantly different which indicated that the intervention effects were similar across the two groups. We further examined the clinical significance of the improvement on CES-D and CGAS for each of the post-intervention assessments for the LIH and ES groups separately, using the reliable change index and following the procedures outlined by Jacobson and Truax . The index identifies the individual scores that reliably changed from pretest to postintervention. The RC is a function of the standard deviation and the reliability for the variable of interest. We used the pretest Cronbach's coefficient alpha for the reliability measure of CES-D. The reliability measure for CGAS was set conservatively at 0.80. Jacobson and Truax suggested that an RC larger than 1.96 would unlikely occur without actual improvement. Table 5 shows the percentage of the RC scores that exceed the cutoff of 1.96 for CES-D and CGAS for each post-intervention assessment. The rates of improvement were comparable between LIH and ES groups and were all above 50%. Finally, at post-intervention over 80% of participants' scores on the CES-D in each group moved into the "normal" range . --- Discussion Study results suggest that brief interventions designed to prevent postpartum depression for American Indians and delivered by AI paraprofessionals may reduce depressive symptoms and improve functioning over time. While this study was small in scope and requires replication, the findings highlight the potential of prevention efforts to ameliorate the psychological and economic costs associated with postpartum depressive disorders in high risk populations. Specific findings and implications for future research are discussed below. --- Depressive Symptomotology and Diagnoses The current study contributes to a growing literature showing that psychosocial interventions have the potential to reduce depressive symptoms. Contrary to the study hypothesis, however, women in both interventions showed equal rates of reduction in depressive symptoms. Specifically, young women in both groups showed a drop of between 7 and 15 points on the CES-D, and few women met criteria for postpartum depression . The magnitude of change in depressive symptoms was similar to that found in other studies with AIs . There are several potential explanations for the lack of group differences. First, it may be that both interventions were equally effective in reducing depressive symptoms, though they may have reduced depressive symptoms in different ways. For instance, in the ES condition, providing information about pregnancy may have increased teens knowledge, optimism, and sense of control over outcomes which may have in turn lowered their depressive symptoms. In the CBT intervention, psychoeducation about depression, increasing pleasurable activities and cognitive restructuring-all skills that target core symptoms of depression-may have reduced depressive symptoms. Alternatively, non-specific therapeutic factors experienced through the presence of a weekly visit from a caring FHE may have led to reductions in depression. Also possible, and consistent with other published trials comparing two active interventions which have failed to show group differences on self-report measures of depressive symptoms , is that neither intervention had an impact and that the reductions in depressive symptoms over time reflect a natural course. Heron et al. in one of the largest studies to examine the course of depressive symptoms over pre-and postnatal periods found that among women in a community sample, the majority reported moderately stable levels of depressive symptoms over time , though decreases were noted from pre to postpartum. However, among those who scored above the clinical cut-off on at least one postnatal assessment 44% of these cases were "new," that is, did not score in the clinical range during their pregnancy. These patterns raise the possibility that prenatal interventions reduce depressive symptoms or prevent increases in depressive symptoms that would be expected to occur naturally. --- Global Functioning and Social Support Similar to patterns found on measures of depressive symptoms, both groups showed similar and significant improvement on a measure of global functioning. This finding strengthens the credibility of both interventions improving the lives of these young women. In contrast, neither group showed changes on the measure of social support. One potential reason for this unexpected result is that the measure of social support used in this study only assessed the degree to which participants felt supported by other people in their life. It is possible that given the realities of life on the reservation and the isolation that some women experience, neither intervention, which was delivered to each woman individually, could impact the number of supports or degree of satisfaction with social supports in these women's lives. As social support has been found to be a protective factor for many mental and physical health outcomes, alternative interventions to improve social support may be needed. --- Limitations The results of this study should be interpreted in the context of several limitations. The sample size was small, limiting power to detect small intervention effects, which are more common in prevention trials, and restricted the number and type of analyses that could be performed, including the examination of moderators and mediators of intervention response. The absence of independent evaluators to assess primary outcomes and the use of measures that may not have fully captured or been consistent with Apache experiences of depression may have restricted measureable changes in outcomes. Comparisons between the two groups on key baseline demographic and clinical characteristics indicated that randomization was successful, minimizing threats to internal validity and increasing confidence in the main findings. However, the high attrition rates raise a concern that the interventions were not meeting the participants' needs or were too demanding. The FHEs believed that the mothers who dropped out tended to have less support from partners and family members, more exposure to drugs and alcohol, and difficulty maintaining scheduled lessons compared with those who stayed in the study. In addition, given that women in both groups showed improvements, the absence of a no intervention or usual care control group prohibits researchers from examining the natural course of depressive symptoms and functioning over this time period. Finally, any mother who scored in the clinical range for depression were referred to a local mental health provider, although they remained in the trial. Hence, it is not clear whether the interventions themselves or the linkages to community mental health care reduced depressive symptoms in this sample. --- Implementation Challenges of Delivering Preventive Interventions to Expectant Reservation-Based AIs Several challenges were encountered during the implementation of the study. The training was lengthy and there was at times cross-cultural misunderstanding about the terms. The FHEs reported that the LIH content was sometimes unnatural for the participants to relate to and comprehend. For example, terms such "depression" lacked relevancy and familiarity to participants and their families. In addition, the flow of teaching was impeded because the FHEs struggled to put the language and content in terms that would resonate with participants. However, because this area of intervention was new, and the investigators were trying to maintain fidelity to the key components of the intervention, the FHEs were trained to adhere strictly to the LIH curriculum. In future studies, we have tried to strike a better balance in training FHEs on key components but allowing flexibility to build relationships with the participants and adjust concepts, terms and flow of home-visiting lessons to meet participants' cognitive, cultural and linguistic preferences. Perhaps the biggest challenge to implementing this study related to the challenges of the study participants: the adolescent AI mothers had competing priorities , were transient, and were often difficult to track down after initial contact was made. For example, they moved frequently, i.e., from boyfriend's parents' home to their own parents' home, and could be difficult to engage due to a myriad of distractions. It was also common for FHEs to go to a participants' home for a scheduled appointment and have no one answer the door. Direct telephone access was also limited among the study participants, so that appointments generally could not be confirmed in advance. --- Clinical Implications and Conclusions Overall, results suggested interventions delivered during pregnancy to "at risk" adolescents and young women have the potential to lower symptoms of depression. The magnitude of change in depressive symptoms was similar to those found in other studies and appeared to be clinically meaningful in that the majority of women's scores on the CED-D moved into the normal range from pre to post intervention. However, additional research is needed to understand the natural course of depressive symptoms throughout the perinatal period among the White Mountain Apache and other AI populations in order to attribute measured changes to the interventions. Replication with a larger sample, a no intervention control condition, blinded evaluators, and a longer follow-up is also needed. The current study contributes to the literature in that it is the first published randomized controlled trial attempting to reduce the incidence of postpartum depression among American Indians. It also adds to the increasing literature that American Indian paraprofessionals may play an important future role in addressing urgent behavioral and mental health disparities in their communities. .

Background-Postpartum depression is a devastating condition that affects a significant number of women and their offspring. Few preventive interventions have targeted high risk youth, such as American Indians (AIs). Objective-To evaluate the feasibility of a depression prevention program for AI adolescents and young adults.AI women (mean age = 18.15; N = 47) were randomized (1:1) to either the Living in Harmony program (LIH, an 8 lesson cognitive-behaviorally based program) or an Educational-Support program (ES, an 8 lesson education program). Both interventions were

Introduction Why do people stick with their unhealthy habits despite adverse consequences? This is a pressing question for both public health research and policy-makers. For example, the overweight and obesity prevalence has been steadily growing in all Western societies . Smoking continues to be a major public health problem even though its health risks are widely recognised and many behaviours that are acknowledged being essential for healthy lifestyles have not been Habits and the socioeconomic patterning of health-related… universally adopted, such as getting enough exercise or eating sufficient amounts of vegetables . Since risky behaviours are more prevalent in lower socioeconomic groups, understanding why unhealthy behaviours are so resistant to change is vital to tackle inequalities in health. In this article, we argue that there is a theoretical tradition which has been unexplored in this context even though it is well suited for examining the core questions of health behaviour research. This tradition is pragmatism and its conception of habits, which offers a dynamic and action-oriented understanding of the mechanisms that "recruit" individuals to risky health-related behaviours. Health-related behaviour is often understood as an issue having to do with the individual and guided by motivations, intentions, self-efficacy and expectations, as it is the case with influential and widely used planned behaviour theories and the health belief model . In this line of thinking, individuals behave the way they do because their intentions, knowledge, beliefs or motives lead them to do so . The individualised approach is especially visible in many psychological theories of behaviour change and in interventions and programmes designed on the basis of these theories . Behavioural interventions have become increasingly important in public health promotion despite weak evidence for their overall effectiveness in generating long-lasting changes in behaviour and their potential to reduce inequalities in health . Research on social determinants of health often takes a critical stance towards psychological theories and recognises social structures as key contributors to health and health-related behaviours . In health sociology, concepts and measures related to power, cultural norms, social circumstances, societal hierarchies, and material resources, for instance, are used to refer to structural constraints and modifiers of individual action and related outcomes. A large body of research has shown that education, occupational status, financial resources, living area, gender and ethnicity all affect ill health and life expectancy and the ways in which individuals act upon their health. The better off people are, the more likely they are to lead healthy lives and adopt healthy lifestyles . While social structure undoubtedly constrains people's behaviour, people can also exert agency, as they are able to consider different options and to act in discordance with their structural predispositions and social circumstances . The key question in sociological theory is, thus, how individual behaviour can be simultaneously understood as shaped by social structures and as governed by individual choices. It is not enough to state that both social structures and individual intentions are important in explaining behavioural outcomes. One also needs to understand how and why social structures enable or generate particular kinds of behaviour within the context of people's everyday lives. Sociological theorisation on health inevitably falls short if it fails to confront this issue, thus leading to an insufficient understanding of factors that shape health-related behaviours . In this article, we first take a look at sociological theories of health-related behaviour, to which the concepts of lifestyle and, more recently, social practices have been central. Then we move on to discuss the pragmatist concept of habit. The concept of habit has often been used in research on health-related behaviours and behavioural change, and it has proved to be useful in explaining continuities in behaviour . We argue that previous research has not taken into account the pragmatist understanding of the concept as an important contribution to theorisations of health lifestyles and practices. Pragmatism's dynamic and action-oriented understanding of habits helps in conceptualizing how practices are formed in interaction with material and social conditions and what the mechanisms are by which practices recruit individuals. In pragmatism, habits are understood in terms of problem-solving; they are active and creative solutions to practicalities of everyday life and responsive to change, not mere blind routines. We, therefore, focus on the creative and active nature of habit formation, which can be understood as mechanisms by which behavioural patterns emerge. The pragmatist approach not only opens new perspectives in health research but can also give new tools for preventing non-communicable diseases and reducing inequalities in health. Next, we discuss theories of lifestyles and social practices and go on to show how the pragmatist theory of habits anticipated many of these insights but also developed its own framework for analysing the inherent habituality of action. --- The interplay between structure and agency: lifestyles and social practices Attempts to bridge the gap between social structures and individual action in health sociology often draw from a loose tradition of practice theories. They are all based on the attempt to overcome methodological individualism without leaning too much towards methodological holism . This means that practice theories try to take into account both individual action and the role social structures play in explaining action . From the perspective of health sociology, the fundamental question is how to understand the interplay between individual agency and structural factors in health-related matters, such as smoking, drinking or food consumption. In this respect, two concepts have been central: lifestyles and social practices. Biomedical or social epidemiological approaches, which dominate health inequality research, typically frame lifestyle as a set of individual, volitional behaviours . Lifestyle is thus a sum of individual health-related behaviours, such as ways of consuming alcohol or dietary habits. In sociological literature, lifestyle is seen as a collective attribute: lifestyles are shared understandings and ways of operating in the world that have been generated in similar social circumstances . They develop over the life-course and are shaped by social and material conditions . As such, lifestyles are not merely outcomes of choices or personal motives and preferences, but they reflect an individual's position in a wider social structure and are fundamentally shaped by those structures. Cockerham defines health lifestyles as "collective patterns of health-related behaviour based on choices from options available to people according to their life chances". In his Health Lifestyle Theory, Cockerham draws from Max Weber's concept of lifestyles, in which lifestyle-related choices are seen as voluntary but constrained and enabled by life chances that are essentially structural: similar life chances tend to generate similar patterns of voluntary action, thus generating patterns of behaviour . Cockerham considers life chances as consisting of a variety of structural determinants, such as class circumstances, age and gender, which collectively influence agency and choices. The interaction between choices and chances constitutes dispositions to act and resulting lifestyles may have varying effects on health. Health-related behaviour is shown to be clustered within individuals and by socioeconomic status , yet health lifestyles are rarely uniformly health-promoting or healthcompromising, and there is a considerable amount of variation in health behaviour between individuals with similar socioeconomic characteristics . Cockerham's approach, like many other approaches to health-related behaviours , draws on Pierre Bourdieu's concept of habitus. Habitus is a set of dispositions that generate class-specific ways of operating in the world . Habitus develops during the socialisation process in interaction with social circumstances and social relations, and it generates tastes, choices and practices that are subjectively meaningful in given contexts. Accordingly, people accommodate their desired way of life in accordance with their assessment of their circumstances and available resources . From a Bourdieusian perspective, health lifestyles are a product of life conditions and available resources, as well as people's preferences and tastes, which are formed in class-specific circumstances. People's dietary patterns, leisure activities and ways of consuming alcohol therefore reflect class relations and distinctions. Bourdieu's ideas on habitus and practices highlight how people's day-to-day activities tend to be, to a great extent, routine-like and taken for granted: once established, a habitus governs behaviour, enabling everyday practices to be acted out without conscious deliberation. Thus, Bourdieu's approach explains why lifestyles are not random by underlining the importance of class-specific social conditions internalised in the habitus. Bourdieu's approach has been repeatedly criticised for exaggerating objective social structures at the expense of agency and reflexivity . Critics have claimed that Bourdieu's concept of habitus does not allow for voluntary action and thus assumes that existing social structures are reproduced almost automatically. While Bourdieu acknowledges the importance of agency, he still prioritises structural determinants of action at the expense of individual choices, preferences and subjective understandings . In more recent discussions, however, the notions of reflexivity and flexibility of habitus have been more central and the idea of an over-controlling habitus has been rejected . Silva has noted that Bourdieu's conception of habitus changed over time so that in his later work habitus is more 'elastic' compared to his earlier work. In fact, Bourdieu's later ideas of the role of reflexivity in situations when habitus and field collide are very close to pragmatism . Yet, Bourdieu gives priority to social class in the process of lifestyle formation. This means that socioeconomic status determines to a great extent what people do . The impression that structures determine can be seen as a result of Bourdieu's emphasis on classrelated determinants of action. Regarding the possibility to modify health-related habits, Crammond and Carey have emphasized that Bourdieu's notion of habitus does not give credit to public health initiatives or to changing conditions for influencing habitus and behaviour. More recently, the concept of social practices has been suggested as a general conceptual framework for analysing and understanding health-related behaviour. While there is a variety of so-called practice theories and no integrated theory of practice exist, we concentrate on practice theoretical approaches and applications that have been central to the fields of consumption and health sociology . In these fields, Reckwitz's influential article is commonly cited as the source for defining social practices as routine-like behaviour which consist of several interrelated elements, such as bodily and mental functions, objects and their use, knowledge, understanding and motivation . According to Shove et al. , practices integrate three elements: materials , competences and meanings . Practices can refer to any form of coordinated enactment: preparing breakfast, having a break at work or having after-work drinks. Similar to lifestyles, social practices turn attention away from the individual and their intentions and motives towards the routinised ways people carry out their daily lives . The idea is to look at people as carriers of practices because practices guide human action according to their own intrinsic logic . In other words, practices are relatively stable ways of carrying out a set of elements in an integrated manner. It follows, therefore, that they are both performances enacted more or less consistently in daily life, as well as entities that shape the lives of their carriers . The social practices approach points out how smoking, drinking and eating should not be seen merely as single behaviours, but rather as parts of collectively shared practices, which intersect with other everyday routines . For example, in understanding drinking behaviour, one cannot separate the act of drinking from other aspects of the drinking situation, such as the kind of alcohol being consumed, how, where and with whom it is done, and for what purposes . Drinking, smoking and eating, accordingly, are not single entities but parts of different kinds of practices, performed and coordinated with other activities of daily life . As the main aim of practice theoretical approaches is to explain the stability and continuities of behaviour, the approach has difficulties in grasping the role of individual agency in the enactment of practices. According to critics, in some versions of practice theory, the role of individual carriers and the ways in which they make sense and experience practices seems to be more or less neglected . Consideration of individuals' sense of doing things is particularly important when studying aspects of human behaviour that can have adverse consequences and are unequally distributed within society. Therefore, we argue that the practice theoretical approach would benefit from more theorization on individual agency and the mechanisms by which individuals adopt and become carriers of practices. For health sociology, the question of how practices change and how people are recruited as carriers of practices is particularly relevant: how can healthy practices be adopted or how can practices be modified to become healthier? We argue that these issues were fruitfully conceptualized by the philosophical tradition of pragmatism with its concept of habits, which takes the individual actor as a premise without losing sight of the force of everyday routines. --- Habits as dispositions In recent decades, pragmatism has become an important source of inspiration for many social theorists . For example, Joas has argued that pragmatists had a unique viewpoint on the creativity of action, whereas for Gross pragmatism is a key point of departure when discussing social mechanisms. Pragmatism has been previously introduced to health research, for example, in relation to the epistemological problems of different kinds of health knowledge and health services research . Here, we focus on the aspect of pragmatist thought we find most relevant for health sociology, namely, its concept of habits. Classical pragmatist philosophers were active at the end of the nineteenth and the beginning of the twentieth century. They included the likes of George Herbert Mead, William James, Charles S. Peirce and John Dewey. We mainly draw inspiration from John Dewey for his insights into the notion of habit. However, all classical pragmatists shared a similar understanding of the essential role habits play in explaining action . Thus, even though classical pragmatists may have differed in their point of emphasis, Dewey's notion of habits is in many ways representative of the classical pragmatist understanding of habits. In this conceptualisation, habits are acquired dispositions to act in a certain manner, but they do not preclude conscious reflection. Pragmatism, like the social practices approach, puts emphasis on contextual factors and the environments of action in understanding how habits are formed and maintained. Thus, one can argue that pragmatists were precursors to practice theorists. First and foremost, pragmatists highlighted the interaction between environments, habits and actors, by pointing out that people are constantly in the midst of ongoing action. Pragmatism also has an affinity with behaviourist psychology, which emphasises the role of environmental cues in triggering action. Behaviourists maintain that once an actor is conditioned to a reaction in the presence of a particular stimulus, the reaction automatically manifests itself when the stimulus is repeated. Say, a smoker might decide to give up smoking but the presence of familiar cues automatically triggers a response that results in a relapse. Classical pragmatists also thought that everything we do is in relation to certain environmental stimuli, but they did not think of the relationship in such mechanical, automatic terms . What acts as a stimulus depends on the part the stimulus plays in one's habits rather than on simple conditioning . Thus, people are not simple automata that react to individual stimuli in a piecemeal fashion but rather creatures of habit. This means that individual actions get their meaning by being a part of habits . What may trigger the smoker's relapse is not the presence of isolated cues but the habits that they are a part of; having a morning coffee, passing by or going to the local bars and grocery stores, and taking a break at work. Habits make the associated cues familiar and give them meaning. The term habit, both in sociological literature and in common usage, typically refers to an action that has become routine due to repeated exposure to similar environmental stimuli. In this conception, the behaviour in question may originally have been explicitly goal-directed, but by becoming habituated, it becomes an unconscious, non-reflexive routine. As such, habits interfere with individuals' ability to act consciously. In practice theoretical approaches habit is similarly paralleled with routine-like ways of doing things. According to Southerton , habits can be viewed as "observable performances of stable practices" , which are essential for practices to remain stable . In addition, habits are often understood as routines in popular science. According to Duhig , the habit "loop" consists of the association between routines and positive rewards. Pragmatists tend to see habits somewhat differently-as inner dispositions. This conceptual move means that habits have a "mental" component and habits can exist as tendencies even when not overtly expressed. Habits are thus action dispositions rather than the observable behaviour to which they may give rise to . As tendencies, habits include goals of action and not mere overt expressions of action; in other words, they are projective, dynamic and operative as dispositions even when they are not dominating current activities . Habits make one ready to act in a certain way, but this does not mean that one would always act accordingly . To paraphrase Kilpinen , habits enter ongoing action processes in a putative form and we critically review them by means of self-control. In this way, habits are means of action: habits "project themselves" into action and do not wait for our conscious call to act but neither are they beyond conscious reflection. According to classical pragmatists, habits thus do not dictate our behaviour. Rather, habits constitute the so-called selective environment of our action. They give rise to embodied responses in the environments in which they have developed but, as dispositions, habits are tendencies to act in a certain manner, not overt routines that would always manifest themselves in behaviour. What distinguishes habits from inborn instincts is their nature as acquired dispositions. Moreover, habits guide action and make different lines of conduct possible. This is easy to see in the case of skills that require practice; for example, being skilful in the sense that one habitually knows the basic manoeuvres, say, in tennis, does not restrict action but rather makes continuous improvement of the skill in question possible. Simply reading books on tennis does not make anyone a good player of tennis and therefore actual playing is required for habit formation. Furthermore, once habits are acquired as dispositions, not playing tennis for a while does not mean that the habits and related dispositions would immediately disappear. In the pragmatist understanding, habits are not the opposite of agency but rather the foundation upon which agency and reflexive control of action are built. Purely routine habits do, of course, also exist but they tend to be "unintelligent" in Dewey's conceptualisation because they lack the guidance of reflective thought. Furthermore, Dewey argued that conduct is always more or less shared and thus social. This also goes for habits, since they incorporate the objective conditions in which they are born. Action is thus already "grouped" in the sense that action takes place in settled systems of interaction . This is where Dewey's ideas resemble practice theory most because the grouping of action into settled systems of interaction can be interpreted to indicate the kinds of enactments that practice theory is interested in. While repeated action falls within the purview of habits, Dewey was adamant that habits are dispositions rather than particular actions; the essence of habit is thus an acquired predisposition to particular ways or modes of responding in a given environment. Compared with practice theories, this notion of habits underscores competences and meanings . Because habits are dispositions, they are the basis on which more complicated clusters of habits and, thus, practices, can be built. This means that practices can recruit only those who have the habits that predispose them to the enactments related to a practice. --- Habits as practical solutions In the previous section, we explained that pragmatists did not think of habits as mere routines. To be more precise, Dewey distinguished between different kinds of habits on the grounds of the extent of their reflexivity. Dewey labelled those habits that exhibit reflexivity as intelligent habits. Smoking is an example of what Dewey called "bad habits": they feel like they have a hold on us and sometimes make us do things against our conscious decisions. Bad habits are conservative repetitions of past actions, and this can lead to an enslavement to old "ruts" . Habits hold an intimate power over us because habits make our selfhood-"we are the habit", in Dewey's words. However, habits need not be deprived of thought and reasonableness. So-called intelligent habits, in which conscious reflection and guidance play a part, were Dewey's ideal state of affairs. Dewey thought that what makes habits reasonable is mastering the current conditions of action and not letting old habits blindly dominate. There is thus no inherent opposition between reason and habits per se but between routine-like, unintelligent habits and intelligent habits, which are open to criticism and inquiry . Many forms of health-related behaviour can be characterised in Dewey's terms as unintelligent habits. We stick to many habits and rarely reflect on them in our daily lives. However, that there are intelligent and unintelligent habits does not necessarily imply that all healthy habits would be intelligent in the sense of being open to reflection. Further, the unhealthiness of a habit does not in itself make a habit unintelligent in the sense of being an unconscious routine. Rather, all habits are intelligent in that they have an intrinsic relationship with the action environment. They help the actor to operate in a given environment in a functional and meaningful way. For example, smoking can be seen as meaningful in many hierarchical blue-collar work environments, where the way in which work is organised determines, to a great extent, workers' ability to have control over their working conditions. Smoking can be used as a means to widen the scope of personal autonomy because in many workplaces a cigarette break is considered a legitimate time-out from work . Smoking can thus be seen as a solution to a "problem" emerging in a particular environment of action, the lack of personal autonomy. In this sense, it is an intelligent habit that enables workers' to negotiate the extent of autonomy they have and to modify their working conditions . As shared practices, cigarette breaks motivate workers to continue smoking and recruit new smokers, but when smoking becomes a routine, reinforced by nicotine addiction, it does not need to be consciously motivated . In the context of highly routinized moments of daily smoking, reflection on the habit and its adverse consequences to health is often lacking . This means that the habit in question is not fully intelligent in Dewey's terms. The mechanisms of adopting so-called bad habits can be very similar to adopting any kind of habit if we understand habits as enabling a meaningful relationship with the environments and conditions in which they were formed. This idea also helps us rethink the socioeconomic patterning of health-related lifestyles. We do not have to assume that people in lower socioeconomic positions always passively become vehicles of bad habits due to limited life chances. The pragmatist view on habit presupposes an actor who has an active, meaningful relationship with the environment, that is, an actor with a capacity for agency, as our illustration of habits as a way to increase worker autonomy shows. Unlike practice theory or Bourdieu's concept of habitus, the pragmatist concept of habit explains habitual action as a solution to practical problems in daily life. For pragmatists, action is always ongoing, and those activities that work and yield positive results in a given context have the potential to become habitual. We thus use habits to actively solve problems in our living environments, adapt to the fluctuating conditions we live in, and also modify these conditions with our habits. --- Habits, doubt and change So far, we have discussed habits as a relationship between the actor and the environment of action. We already hinted at the pragmatist idea that habits can be reflexive, and we now move on to discuss in more detail how and why habits change. According to Shove et al. , practices are formed and cease to exist when links between materials, competences and meanings are established and dissolved. Additionally, practice theorists have suggested that practices may change when they are moved to a different environment or when new technologies and tools are introduced . Actors may learn new things and perform practices in varying ways as performances are rarely identical . However, it is insufficient to assert that practice theory assumes an active agent with transformative capacity if the underlying view of agency is passive and practices are the ones with agency to recruit actors. Furthermore, the question remains as to when actors are capable of being transformative and when they are confined to the repetition of practical performances. The pragmatist understanding of how habits change, and when and how actors exercise their agency, originates in Charles Peirce's thought. Peirce argued that we strive to build habits of action and often actively avoid situations that place our habits in doubt because doubt is an uncomfortable feeling. However, habits are nevertheless subject to contingencies and unforeseeable circumstances. Doubt cannot thus be avoided and it manifests itself in the crises of our habits that take place in concrete action situations and processes. How should one then go about changing habits? This is a central question in all health sociological theory and has significant practical implications. Dewey was a forerunner of many modern views in that he saw that habits rarely change directly by, for example, simply telling people what they should do. This presupposition is well acknowledged in critical health research, which has repeatedly pointed out that there is a gap between guidelines of healthy living and people's life worlds . It is usually a better idea to approach habit change indirectly by modifying the conditions in which habits occur. In the case of unwanted habits, conditions "have been formed for producing a bad result, and the bad result will occur as long as those conditions exist" . Dewey's emphasis on the role of conditions is well reflected in modern public health promotion, which rely on population-level measures and interventions. Yet, Dewey's notion of the conditions of habits goes beyond macro-level measures, such as taxation, restrictions and creating health promoting living environments, to cover more detailed aspects of our daily life. According to Dewey, changing the conditions can be done by focusing on "the objects which engage attention and which influence the fulfilment of desires" . Assuming that simply telling someone what they should do will bring about a desired course of action amounts to a superstition because it bypasses the needed means of action, that is, habits . Interestingly, Dewey's ideas of behaviour change have many similarities with the approach known as nudging, as both want to modify environmental cues to enable desired behavioural outcomes . According to both of these approaches, behavioural change is often best achieved by focusing on the preconscious level of habitual processes rather than appealing to the conscious mind by informing people of the potential risks associated with, for example, their dietary habits. Despite these similarities, the pragmatist view of habit change cannot be reduced to the idea of modifying people's "choice architectures". As Pedwell has pointed out, advocates of the nudging approach fail to sufficiently analyse how habits are formed in the first place and how they change once nudged. In the nudge theory, habits are analogous to non-reflexive routines, and the change in habitual behaviour occurs due to a change in the immediate environment of action. As a result, nudge advocates conceptualize the environment through a narrower lens than pragmatists and they are less concerned about how broader social, cultural, and political structures influence and shape everyday behaviour . According to pragmatists, changing habits is something that we do on a daily basis, at least to some extent. This does not mean that we would ever completely overhaul our habits. Dewey thought that character consists of the interpenetration of habits, and therefore a continuous modification of habits by other habits is constantly taking place. In addition, habits incorporate some parts of the environments of action, but they can never incorporate all aspects of the contexts of action. What intelligence-or cognition in modern parlance-in general does is that it observes the consequences of action and adjusts habits accordingly. Because habits never incorporate all aspects of the environment of action, there will always be unexpected potential for change when habits are exercised in a different environment than the one in which they were formed . Different or changed contexts of action imply the potential to block the overt manifestation of habits. For example, if workplace smoking policies are changed so that smokers are not allowed to smoke inside, the habit of smoking needs to be reflected upon and the practice of workplace smoking modified. If the employer simultaneously provides aid for quitting smoking, or even better, creates conditions for work which would support workers' experience of agency and autonomy, some may consider breaking the addiction, at least if colleagues are motivated to do the same thing. Such contextual changes lead to moments of doubt in habit manifestation and thus compel us to reflect on behaviour and, in some cases, to come up with seeds for new habits. The habit of smoking can be seen as a way of dealing with "moments of doubt". It is a solution to certain problems of action in a given environment, as in the previous example of workplace smoking and autonomy. If the original context for which the habit was a "solution" to changes, it becomes easier to change the habit as well. Pragmatist thinking thus suggests that here lies one of the keys to reducing unhealthy behaviours. By modifying the environments of habits, it is possible to create moments of doubt that give ground to the formation of new habits. Contrary to nudge theorists, however, pragmatists are not only concerned with promoting change in individual behaviours and its immediate action environment but also in the sociocultural contexts of habit formation by enabling people to create new meaningful capacities and skills . The pragmatists also considered the consequences of moments of doubt on habits. Dewey argued that habits do not cease to exist in moments of doubt but rather continue to operate as desireful thought. The problem with "bad habits" is that a desire to act in accordance with the habit may lead to solving situations of doubt by changing the environment so as to be able to fulfil the habit rather than changing the bad habit. For example, new smoking regulations intended to decrease smoking may not lead to an actual decrease but rather to a search for ways to circumvent the regulation by smokers. A crisis of a particular habit thus need not always result in changes in behaviour, as the disposition does not change overnight and may lead to looking for ways to actively change the environment of action back to what it used to be. Furthermore, the crisis may simply be left unresolved. This is what often happens when people are exposed to knowledge of the adverse consequences of their behaviour. There might be a nagging sense that one really should not behave the way one does, but as long as the environmental cues are in place, the habit is not modified, especially if one's social surroundings reinforce the old habit . It can also happen that one makes minor changes in behaviour, for example, by cutting down instead of quitting smokingwhich can in time lead to falling back on earlier smoking patterns. New workplace smoking policies, therefore, often mean that the practice of smoking is modified, and the smokers adopt new places and times for smoking. While old habits often die hard, discordances between habits and their environments can nevertheless trigger reflection and thus have a potential for change. --- Discussion We have argued that the pragmatist understanding of habits is an often-overlooked forerunner of many modern theories of health behaviour. While the health lifestyle theory helps to analyse the factors by which health lifestyles are patterned and points out that both contexts of action and individual choices are important in lifestyle formation, it is less helpful in empirical analyses on the mechanisms by which particular patterns of behaviour emerge in the interplay between choices and chances. The social practices approach further elaborates the relationship between choices and life chances by turning attention away from the structure-agency distinction towards enactments of everyday life and on how people go about their lives by carrying social practices. However, the social practices approach runs the risk that individual action becomes a mere enactment of practices. Thus, the practices are the true agents and people become mere carriers of practices. In this context, the pragmatist notion of habits can be useful in grounding practices within the clusters of habits that people have, thereby enabling them to be recruited by specific practices. To conclude the paper, we want to stress some of the key pragmatist insights into the theorization of health lifestyles and practices. First, unlike practice theories, pragmatism takes individual actors and their capacity to meaning making and reflexivity as a premise for understanding how habits are formed and maintained. Thus, from the actor's point of view, habits, even "bad" habits, should be understood as functional and meaningful ways of operating in everyday circumstances. Habits are creative solutions to problems confronted in everyday life and reflect individuals' relationships to the material and social world around them. Action that proves useful and meaningful in a particular context is likely to become habitual. In the context of health inequalities, risky health-related habits can often be seen as a way to strive for agency in circumstances that provide little means for expressing personal autonomy. We suggest that this insight should be at the core of designing any public health or behavioural change interventions tackling health inequalities. Second, pragmatism suggests that habits should be understood as dispositions; people are recruited by practices only when their dispositions enable this to happen. Often a lot of habituation is required before the predispositions are in place that make recruitment possible. Third, pragmatism provides tools to analyse how moments of doubt enter habitual flows of action. Doubting habits is an inherent part of our action process, but habits are called into question especially by changes in the environments of action that make particular habits problematic. This, then, can lead to the development of new or modified habits as a response to the "crisis" of action. If the social and material environment of action, to which the habit is a response, stays more or less the same, the habit will be difficult to change. The pragmatist conception of habits, while emphasizing agency and reflexivity, does not ignore the significance of materiality and routines in daily conduct but is able to incorporate these elements of action in a way that benefits empirical analyses of everyday practices. Pragmatism thus suggests a variety of research settings to investigate the mechanisms by which health-related habits are formed. Here, we provide a few examples. On a macro level, it is important to observe how organisational, technological, or legislative changes are manifested in different contexts and how they modify and enable habitual action in different social groups and settings. Structural measures to promote public health are likely to invoke varying effects depending on the contexts of action of different population groups. Although the physical environment may be the same, the environment of action is not the same for everyone. In pragmatist terms, new policies can be understood as modifications of action environments, which potentially create moments of doubt in habitual action. For example, there is considerable evidence that smoke-free workplace policies reduce workers' smoking , but more research is needed to determine how different socioeconomic groups are affected by these policies. Macro-level policy changes create an excellent opportunity to study how policies give rise to new patterns of health-related behaviour, how policies are implemented in different contexts, and how reactions to policies and their effects vary depending on socioeconomic circumstances. A micro-level analysis of health-related behaviour, on the other hand, could focus on the triggers of the immediate environment of action-material, social or cognitive-to examine how habits are formed as practical and creative solutions to specific problems and what kinds of factors create situations of doubt and thus include the potential for habit change. Research should analyse how moments of doubt regarding health-related habits emerge in differing socioeconomic contexts, as well as why unhealthy habits can and often do become deeply routinized and resistant to change. Furthermore, it is essential to find out the problems in relation to which particular habits of action have been formed. In both micro-and macro-level analytical approaches, people's reflexive capacity and the pursuit of a meaningful and functioning relationship with their environments should be at the core of analysis. Methodologically, we suggest that the pragmatist approach to health behaviour research calls for methods that integrate the observation of action and people's accounts of and reasoning about their conduct. Ethnography is one research method suited to this task. With participant observation, it is possible to access lived experiences in local settings through which larger policies affect health and hard-to-reach population groups . So far, ethnographic studies have been rare in health inequality research . One way to proceed is provided by Tavory and Timmermans , who have suggested pragmatism as a theoretical-methodological basis for constructing causal claims in ethnography. They propose that a useful starting point for observation could be the process of meaning making: how individuals creatively navigate their conduct when confronting moments of doubt and how they make sense and respond to them in more or less habitual ways. However, surveys can also be used in creative ways to investigate people's habits, for example, using mobile apps that ask and/or track what people are doing. Other methods besides ethnography are thus needed to test the causal claims made by ethnographers. Lastly, research is needed on how educational systems predispose people to develop reflective habits. One possible explanation for why knowledge about the adverse consequences of health-related behaviour is correlated with people's socioeconomic status, and especially their level of education, is that a higher level of education makes one more sensitive to knowledge-related cues for behaviour. This is because higher educational levels tend to bring about the habit of reflecting on the basis of new knowledge. Education is intimately related with a habit of thinking of things in more abstract terms-distancing oneself from the specifics of particular situations and moving towards more abstract thinking. A high level of education also means the absorption of new knowledge has become habitual. Unfortunately, there are no shortcuts to developing such capacity. This is one of the reasons for why merely providing information on health-related issues will affect different population groups differently. --- Data availability Not applicable as no data was used in the article. --- --- Conflict of Interest The authors have no conflicts of interest to declare. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/ licenses/by/4.0/. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Anu Katainen is a Senior Lecturer in Sociology at the Faculty of Social Sciences, University of Helsinki, Finland. Her research comprises of projects investigating alcohol policy and drinking cultures, as well as social and health inequalities, with a focus on comparative qualitative sociology. Antti Gronow is a Senior Researcher at the Faculty of Social Sciences, University of Helsinki, Finland. His research interests include climate policy, advocacy coalitions, social network analysis, and pragmatist social theory.

Unhealthy behaviours are more prevalent in lower than in higher socioeconomic groups. Sociological attempts to explain the socioeconomic patterning of healthrelated behaviour typically draw on practice theories, as well as on the concept of lifestyles. When accounting for "sticky" habits and social structures, studies often ignore individuals' capacity for reflection. The opposite is also true: research on individual-level factors has difficulty with the social determinants of behaviour. We argue that the pragmatist concept of habit is not only a precursor to practice theories but also offers a dynamic and action-oriented understanding of the mechanisms that "recruit" individuals to health-related practices. In pragmatism, habits are not merely repetitive behaviours, but creative solutions to problems confronted in everyday life and reflect individuals' relationships to the material and social world around them. Ideally, the pragmatist conception of habits lays the theoretical ground for efficient prevention of and effective support for behaviour change.

Introduction C ancer is one of the most severe health problems of our age; it brings along fear, despair, guilt, abandonment, excruciating pain, and death. Accordingly, it leads to psychological imbalance in an individual's life. [1,2] Like many other countries, cancer is ranked as the second cause of death in Turkey. [3] The global cancer burden is estimated to have risen to 18.1 million new cases and cancer caused 9.6 million Nigerian Journal of Clinical Practice ¦ Volume 22 ¦ Issue 8 ¦ August 2019 levels. [5][6][7][8][9] The social support provided by the family and friends of the patient is of critical importance in the fight against cancer. [8][9][10][11][12] Social support caters for the basic social needs of individuals such as compassion, attachment, self-esteem, and feelings of belonging to a group. It enhances an individual's capacity for stress management. Social support is an emotional and physical protection factor against diseases. [13,14] Social support is generally provided by family members, kith and kin and healthcare professionals . The patients can better adapt themselves to the disease as the received social support increases. [15] Isolation can be observed in case of social support inadequacy and the quality of the care will decrease in such cases. [16] In this respect, it is argued that existence of social support leads to a decrease in physical and psychological symptoms of cancer patients and increases their quality of life significantly, [10,17,18] helps patients tolerate side-effects of treatment [19] and decreases the rate of mortality caused by cancer. [16,20] Applebaum et al. reported thatcancer patients' life quality increased and their anxiety levels decreased when they received high level of social support. [21] A study conducted on cancer patients by Hodges and Winstanley found that social support provided cancer patients with a positive outlook, increased their wellness, and had a positive contribution to their recovery. [8] Social support was found to have significant main effects on physical and mental quality of life and on mental health-related issues among male and female survivors. [22] . High levels of social support were associated with low levels of depression. [23] On the other hand, social support also affects the frequency of admission and referral to the hospital. Patients who experienced insufficient social support asked for referrals, however, patients who experienced sufficient support had fewer requests for referral. [24] Social support enables the enhancement of relationships between groups or individuals. When an effective patient-nurse relationship is established, supportive care is also delivered. A qualitative study revealed that supportive nursing improves relationships in which nurses play a key role in improving these relationships. [25] A study emphasized that social support is extremely important for the procurement of psychological recovery and wellness of teenager cancer patients. [11] A study has revealed that social support given by families is a factor that increases patients' level of hopefulness. [26] Another study showed that social support makes it easier to adapt to trauma, lessens post-trauma reactions and protects the patient from secondary traumas. [27] It is emphasized that patients who perceive that they do not get sufficient emotional and social support have negative feelings such as guilt, solitude, hopelessness and anger and that this situation has adverse effects on stress management. [28] As stated in literature, cancer and its treatment affect an individual physically, psychologically, and socially. Social support protects the patient from negative results of stress, creates a positive effect on emotional and physical health and plays a protective role in stress management. Healthcare professionals must take steps towards maintaining and improving the wellness of cancer patients and develop a guideline. It is very important for cancer patients to develop a positive attitude towards the changes caused by cancer in their lives so that they can cope with the stress crated by cancer and related treatments. The demographic parameters that affect social support are significant in cancer patients in developing countries. Studies conducted with patients with cancer in Iran indicate that female patients, for house workers and patients who live alone have higher support needs. [29] Also, the number of children is significantly related to social support. [23] The results of a study in Pakistan showed age of patients with breast cancer significantly correlated with social support. [30] A study in India found that patients with cancer in the age range of 40-59 tended to require more support compared to other age groups. [31] In studies conducted in Turkey, compared to male patients, female patients were found to perceive less social support from their families. Single patients perceived less social support compared to married patients. [32,33] Also educational level, state of employment and perceived economic status affect social support perceptions. [33] In this context, it is clear that social support is an important factor which affects the treatment process of patients receiving cancer treatment. The perception of social support is also influenced by cultural factors. Although it is stated that patients' demographic characteristics affect social support perceptions, studies with large samples and multicenter outpatient studies are limited. The demographic data which affected social support perceptions of cancer patients living in Turkish society were addressed in multidimensional aspects. --- Patients and Methods The population of the study was made up of 6754 cancer patients admitted to the Outpatient Chemotherapy Unit at two university hospitals and a Ministry of Health The Cancer-Specific Social Support Scale : This scale identifies the perceived social support of cancer patients. It was developed and tested for validity/reliability by Berrin Eylen . [34] The scale contains totally 35 items, 13 of them are negative statements and 22 of them are positive statements. The rating is based on the 5-point Likert scale, in which points 5, 4, 3, 2, and 1 stand for "quite applicable to me," "applicable to me," "partly applicable to me," "not applicable to me" "not applicable to me at all," respectively. The scale has three subscales: confidence support, emotional support, information support. To calculate perceived support score, individual's score in the negative statements is reversed and then added to the total score in the positive statements. The social support provided by families as perceived by cancer patients was measured with a five-point scale in which higher scores on the scale reflect higher levels of perceived social support. The Cancer-Specific Social Support Scale was used in this study because it was first developed in Turkey and therefore it was considered to reflect the values of Turkish society. Cronbach alpha coefficient of original scale was 0.92. In this study Cronbach Alpha coefficient was found to be 0.83. --- Ethical considerations For the implementation of the research, written permission was received from the Ethical Committee of Scientific Research at Ege University Faculty of Nursing . Written permissions were taken from three hospitals for conducting the study. The patients who wanted to take part in study voluntarily were verbally informed about the aims of study and were informed about the conditions in case they wanted to be excluded. In regards to vountary aspects of the study, subjects provided verbal consent. Permission was received from Berrin Eylen, the developer of the scale, by email to use the CPSSS. --- Data analysis The data were analyzed by Ege University Department of Biostatistics and Medical Informatics using the SPSS version 21.0 . Descriptive data on patients were expressed as numbers, percentages, means, and standard deviations. Kruskal-Wallis and and Mann-Whitney-U tests were performed for variables that did not exhibit a normal distribution. One-way analysis of variance test and independent-sample t-test were performed for variables that exhibited a normal distribution. Level of significance was assumed to be. 05. --- Results The mean age of patients was 51.75 . It was found that 40.7% of the participants were primary school graduates, 53.2% were female, 84.6% were married, 85,8% had children, only 20.1% worked, 36,6% were housewives and most had social security . Approximately two-thirds of them considered their income status as "middle level". Nearly 50% were members of nuclear families and most had breast cancer. Most of the patients received chemotherapy at Pamukkale University Hospital . While 24.6% of the participants received chemotherapy, 40.0% had surgery and received chemotherapy. Only 25.8% of the patients had metastases and 34.8% had an accompanying disease. Very few patients reported that they did not receive information about their diseases [Table 1]. Most patients reported that they received information about their illness and the relevant treatment, more than half of the patients reported no problems doing their daily work and they were able to get help during their daily chores. Almost all patients were aware that their friends or relatives know about their diseases and 79.2% weren't worried about sharing information with friends and relatives. Most of the patients reported receiving psychological and physiological support during the course of the disease and therapy and 62.3% of them received support from family members. Additionally, 31.2% suffered from side effects of chemotherapy and 45,9% had problems related to were found to be the most important sources of support for cancer patients. Table 3 presents the average distribution of points according to factors that may affect perceived social support. A significant difference was found between CPSSS mean score and patients' age groups, education level, gender, marital status, whether they had children, number of their children, their work status, income level, diagnosis, whether they had information about their disease, whether they were concerned about others' knowing about their disease and the status of getting help . Patients in the advanced age group were found to have higher perceived social support in confidence support , knowledge support and social support dimensions . Male patients, married patients, patients who did not work outside the home and patients with children were found to have high level of perceived social support only in confidence support and in general, while patients who graduated from primary school and patients with three or more children had high level of perceived social support in all sub-dimensions and in general high. The patients who were informed about the disease and its treatment were found to have higher social support . Patients who were provided with information about their diseases and relevant treatments were found to have higher perceived social support . Social support perceived by patients who were worried about others' knowledge of their diseases was found to be higher in the information support sub-dimension. Social support perceived by patients who received support during daily tasks was found to be higher in the sub-dimension of confidence support. No significant differences were found between the score obtained from the CPSSS and the following: occupation , social security , Whom do you live with? , Diagnosis , Metastasis , treatments that were received , accompanying diseases , whether they were provided with information about their diseases and treatment , whether they had difficulty undertaking daily chores , income level , whether their friends or relatives know of their disease , whether they received psychological or physiological support during their disease and treatment? , who they got the most support from , whether they suffered from any side effects , whether they experienced problems with their sexual lives , whether they participated in social events and whether they had time for their hobbies . --- Discussion It is known that perceived social support changes an individual's values and has significant effects on the emergence, prognosis and recovery of many physical and mental illnesses. [35] Many patients with cancer or similar chronic illnesses resort to their own social support network and use various management methods of their own when they face stressful situations. In his study devoted to the perceived social support levels of cancer patients, Özyurt revealed that patients received social support mostly from their spouses, followed by their children and from their siblings and parents. [36] In this study, when patients were asked whether they received psychological or physiological support during their diseases, 73.3% of the patients responded in affirmative and reported that they got social support from their family members mostly , [Table 1]. Naseri and Taleghani found that cancer patients received maximum social support from their families, friends, and relatives. In this regard, the results of previous studies also indicated that patients received the maximum acceptable social support from family. [23] In the study conducted by Eylen to determine the social support level in cancer patients, the support received from families was found to be high. [34] Similar studies also reported that social support is mostly received from family members. [5,37,38] The fact that cancer patients get the highest social support from family members that provide care is in line with the literature. For this reason, it is essential for nurses to guide family members towards supporting the patient. [9,39] Family-centered social support programs where patients undergoing treatment can express their own thoughts and feelings are essential for gaining individual skills. This study found that patients scored high on perceived social support. It was seen that the social support most perceived by the patients was confidence support while the social support least perceived by the patients was information support. In their study, Çalışkan et al. also found that the type of social support perceived the most by patients was confidence support whereas the least perceived social support type was information support. [40] It was demonstrated that senior patients' confidence and information support sub-dimensions and their general perceived social support levels were higher. Senior patients stated that they felt more secure and had better information and general perceived support. A study found that while the problem-solving strategy use decreased with age, the need for social support increased. [36] In a similar study, however, it was explained that the elderly received less social support compared to the young. [37] In terms of the current study, senior patients' perceived social support level was determined to be high, however, the same applies to and is important for all age groups of cancer patients. In the study conducted with breast cancer patients, it was found that perceived social support increased along with educational level. [34] Likewise, it was ascertained that the higher perceived social support mean scores were obtained by patients who were graduates of middle/ high school or higher level of education. [33,40] The study by Costa-Requena et al. has shown that education level has no effect on the level of social support received. [35] Contrary to the literature, the findings of this study show that perceived social support of the patients who were primary school graduates were statistically significant in all sub dimensions. The study concluded that demographic characteristics of the patients affected perceived social support. [41] The sociodemographic characteristics of cancer patients are important in terms of understanding social support differences and evaluating and supporting the patients. [42] It was found that the male patients that took part in the study had a good level of social support, particularly perceived confidence support. Similarly, it was observed that women received lower social support in the long run following the diagnosis compared to men. [35] The study by Paterson et al. showed that male patients with social support managed stress better, suffered from depression less and had increased quality of life. [43] The married patients in the sample group stated that they felt more secure and received better social support. The type of the support that patients need can vary according to the individual, time, and circumstances. While single patients may benefit from emotional support, married patients with children can benefit from social support provided by neighbors. A study has shown that patients that are married for longer periods of time adapted to their situation better and that they received most of the support from their spouses. [44] Literature points out that patients receive most of the support from their spouses and that spousal support is especially significant in fighting the disease, sharing responsibilities and dealing with the physiological/ psychological problems that may arise. [39,40] A similar study revealed that single patients received social support mostly from family members and friends. [13] Leung et al found that single patients receiving chemotherapy faced with more psychological problems than married patients and that factor that caused the significant difference in Leung's study may be related to the fact that single patients had less social support compared to married patients. [45] It was determined that patients with multiple children had higher levels of general perceived social support. A study found that cancer patients living with their children had a lower risk of mortality. [19] However, a similar study demonstrated that the number of children had no effect on perceived social support. [35] It was observed in the study that whether others had knowledge of subjects' diagnoses did not affect their perceived support and when these patients wanted to get more information about their cases, they did. The findings revealed that the subjects could undertake their daily tasks easily even when they did not receive social support. It should also be kept in mind that family members, who are affected by this illness as much as the patients, are in need of social support as well. [35] The fact that the sample of this study was composed of patients treated at the outpatient chemotherapy unit at three hospitals in two separate provinces in Turkey is the limitation of this study. Patients undergoing different cancer treatments were not included in the study. There is a need for larger-scale and prospective studies for all patients who receive outpatient or long-term inpatient cancer treatment in hospitals to investigate social support status and factors affecting social support. --- Conclusion The cancer patients included in the sample of this study were found to have a high level of perceived social support. Factors such as age, gender, marital status, number of children, and educational status were determined to affect perceived social support. It was determined that perceived social support was high in married individuals, in males, in individuals with low level of education and in seniors. On the other hand, patients' disease and treatment status, working status, and economic factors were found to have no effect on perceived social support. Patients who do not live alone , patients with low level of educational backgrounds who receive sufficient information about their diseases and treatment, and individuals who can get support during their daily tasks were found to have higher perceived social support. In this context, it is recommended to evaluate patient characteristics before providing support since sociodemographic characteristics of cancer patients influence the level of social support receival. It is recommended to raise the awareness of families and organize institutional education programs for them since they play key roles in providing social support. The long-term effects of social support should also be studied. --- Financial support and sponsorship Nil. --- Conflicts of interest There are no conflicts of interest.

Background: Social support, acknowledged as a protective factor against diseases, produces positive results in the prognosis of cancer patients. Objective: There is no relevant studies on perceived social support and factors related to Turkish cancer patients. Hence, the present study aimed to determine the demographic characteristics that affect the perceived social support in cancer patients undergoing chemotherapy in Turkey. Patients and Methods: This study was designed as a descriptive cross-sectional study. The sample included 423 adult cancer patients who admitted to the Outpatient Chemotherapy Unit of three hospitals in Turkey between March 2014 and August 2014. The data were collected through Demographic Characteristics Form and Cancer-Specific Social Support Scale (CPSSS). Results: Patients were between 19 and 85 years of age with a mean 51.75 years. Almost half of them (40.7%) were primary school graduates, 53.2% were female, 84.6% were married and 85.8% had children. It was found that the patients generally received a high score (127.86 ± 17.44) from the CPSSS scale. The highest scores were obtained on the Confidence Support sub dimension. It was revealed that women and married patients needed more confidence support and general social support (P < 0.05). Social support perceived by primary school graduates was statistically significant to a great extent (P < 0.05). In spite of the difference between the patients with metastasis and their perceived social support grade averages (P > 0.05), it was seen that patients with multiple children utilized more confidence support and general social support in social support reception (P < 0.05). Conclusion: The study showed that study participants received a high amount of social support. Age, gender, marital status, number of children, and educational status were determined to affect perceived social support. It is recommended to support the patients who do not receive sufficient social.

INTRODUCTION Online social networks have great benefits and advantages. They allow for the quasiinstantaneous exchange of up-to-date information and give access to persons around the world with different backgrounds, experiences and opinions. They also create communities with sizes well beyond the usual social constraints, and perhaps even beyond cognitive ones 1 . Nevertheless, the constant increase in network size and complexity may introduce more information than we can normally process 2 , as well as promoting passionate debates. Beyond the initial excitement these networks provided, the regular polarization of positions on social media appears worrisome. For example, it promotes severe conflicts between communities expressing opposite beliefs, while also making social networks particularly vulnerable to manipulation or propaganda, for instance, by bots accused of interference with presidential elections 3 . Solutions need to be found, but without sacrificing the advantages of worldwide information access or impoverishing social interactions. In our view, the problem goes far beyond the propagation of fake news, which is a symptom as much as a cause of polarization. More than the content of one's belief, the issue seems to revolve around overconfidence and excessive trust in information confirming these beliefs. Many of the most polarizing issues discussed on the internet may not even have a universally defined, knownable, or absolute answer . For these issues, radicalization can be defined as people reaching unreasonably confident and monolithic beliefs based on multifaceted, biased or untrustworthy data 4 . Additionally, the emergence of two or more radicalized groups with opposite, irreconcilable beliefs results in polarization. To capture these phenomena in a simplified, mathematically grounded but intuitive framework, we treat large-scale opinion sharing in social networks as a form of probabilistic inference. People's beliefs are modeled as the probability of giving an answer to a particular question . Rather than just deciding "yes" or "no" once for all , someone could have a graded confidence level represented with a probability, close to 100% or 0% for high confidence or equivalently strong opinions, but approaching 50% if the person is uncertain. Agents embedded in a social network derive their beliefs both from external or private sources of evidence and from the expressed opinions of people they are connected to or communicate with . Through communication, that is, the propagation of messages within a social network, each person's opinion should ideally become as informed as possible, integrating the knowledge and experience from all the network members. In other words, we work under the hypothesis of normativity, according to which the purpose of communication is to ensure that individual opinions converge to a consensus corresponding to the posterior probability of the correct answer given all the external evidence. This "ideal" situation, well defined mathematically, represents a benchmark against which various message propagation schemes can be compared, while significant deviation could be considered irrational. Unfortunately, the structure of social networks renders simple message passing schemes fatally flawed as an inference mechanism . In particular, every loop in a social graph forms an echo chamber where opinions can reverberate ad infinitum and be artificially amplified 5,6 . We thus confront both the strengths and weaknesses of the massification of social media: social networks could make local information globally available as never before. However, they also tend to aberrantly amplify confidence, leading to radicalization and polarization and, as we will see, severely limiting their true information sharing capability. The goal of this paper is twofold. First, we provide a simple account of echo chambers using a probabilistic inference framework applied to realistic social graphs and systematically study their consequences. Second, we propose a method that limits these detrimental effects by trying to achieve normality, bringing the confidence levels generated in the network closer to informed rationality. We demonstrate the efficiency of this algorithm in both toy graph-models and more realistic graph structures borrowed from popular social networks. answer could be all the information Robert received, including information that came from Anna . This approach is suboptimal since redundant information is exchanged and thus counted several times. To address this overcounting, a second implementation would consist of sending Anna all the information received by Robert except the information coming from her . This message cancellation is indicated by the green arrow. However, the problem becomes much more complex when Anna and Robert have common friends like Susan. In this case, Robert's belief is corrupted by what Susan knows from Anna. Therefore, an extra correction has to be applied to control the flow of circular messages in the social graph. The Circular Belief Propagation algorithm implements such a correction. Without proper control, the highlighted problem becomes much more serious when the social graph is highly cyclic, when adding new friends/followers such as Mary, Meg and Bob . --- RESULTS --- -A BRIEF OVERVIEW OF MESSAGE PASSING SCHEMES In our simplified social network model, each agent is a node of the network, and edges of the network represent the social circle of agents . We assume that each agent i estimates the exact probability distribution of a binary variable with the estimate probability distribution, given diverse sources of external information and the agent's preference. In the following, we will refer to as the true probability and to as the estimate probability. For convenience, we also define the belief of agent i such that ; see Sup.Fig. 1. The sign of describes the agent's opinion about the binary question: if for example , the agent believes that the answer to the corresponding question is more likely to be "yes" than "no". Additionally, the absolute value |Bi| quantifies the confidence of agent . The higher the confidence, the more certain the agent is about the answer, while implies complete uncertainty. To determine the value of his/her belief , the agent has to combine two types of information: -External sources of information received by this agent, or agent's preference, grouped together and quantified as the external message . Such a message is negative if it supports "no", and positive if it supports "yes" ; The amplitude of this external message indicates its assumed reliability. -Information provided by the opinions broadcasted by members of the agent's social circle . denotes the message sent from agent i to agent j. An agent's core belief is the sum of all the internal and external messages it receives : . Meanwhile, the message depends on the belief of agent j, and the amount of trust 7 between the two agents i and j. In the simplest possible message passing scheme, called variational message-passing 8 , the message corresponds to the sender's belief modulated by trust: , where is a sigmoidal function which depends on the amount of trust between the two agents. This naive method of communication assumes that agents systematically broadcast their opinion to their entire social circle, and in turn combine internal and external messages to update their own beliefs . This message-passing algorithm corresponds to what was proposed in previous models of opinion dynamics in social networks 5,9,10 with slight differences in the precise form of the sigmoidal function . However, the above mean-field scheme is highly suboptimal at performing inference in a graph. In particular, it creates a reverberation of messages between each connected pair of nodes: agent j influences i, who influences j, etc. Humans probably never communicate this way; for instance, we only tell our friends things they presumably do not already know. A less naive communication method, which we hypothesize to be our model for communication, ensures that messages do not include the messages sent previously in the opposite direction . Messages are updated iteratively as follows: . The resulting message passing scheme corresponds to a widely used inference algorithm called Belief Propagation 11 . Despite its simplicity, this algorithm is surprisingly powerful as an inference method 12 . In fact, BP is even exact in graphs without cycles, that is, it converges to the true posterior probabilities. However, in the presence of cycles, messages can still be reverberated and artificially amplified, leading to overconfidence, shown schematically in Fig. 1d. Unfortunately, social networks contain a large number of such loops . As a result, we will see that BP, considered as a model of social communication, systematically leads to radicalization and polarization in cyclic social graphs. As a society, we urgently need to find solutions that can preserve the global knowledge sharing capabilities of social networks, while suppressing the detrimental effects of loops or echo chambers. When integrating information from someone, one should in theory consider all the indirect ways the content has been brought to him or her in order to not take into account the same piece of information twice. With this goal in mind, we introduce an adaptation of the Belief Propagation algorithm called Circular Belief Propagation 13 which aims at actively removing redundancies between messages introduced by loops and amplification of messages through cycles. The resulting message passing scheme can be written as follows: where beliefs are defined by: In contrast to BP, CBP contains two types of control parameters: a gain applied to each node, and a loop correction term applied to each link. The idea the first equation is to subtract more than once the opposite message from the belief of agent j. This is based on the fact that agent j is not only influenced directly by i, but also indirectly by any person k . Intuitively speaking, the loop correction term " " subtracts the predictable "redundant" part from incoming messages, which is the result of the reverberation of the outgoing message through all the graph's loops. Similarly, the gain in the second equation prevents the amplification of beliefs due to excess correlations between all incoming messages as introduced by loops . Note that these control parameters need to be adjusted to the specific graph structure of the social network, thus posing an additional challenge. Here we will consider two methods of finding a good set of control parameters : a supervised learning method, that can only be used in extremely small graphs, and a local unsupervised learning rule that is less optimal but applicable to graphs of arbitrary sizes . All control parameters can be trained in an unsupervised manner by ensuring that incoming and outgoing messages remain as decorrelated as possible when they contain no meaningful information. To model opinion formation in a social network, we iterate 100 times the BP/CBP message passing scheme simultaneously in all the nodes, to let the information provided by the external messages propagate in the entire graph . Further details are provided in the Methods section and the pseudo-code is given in the Math Appendix. We first tested the performance of BP and CBP in small graphs where running exact probabilistic inference is still practical, as well as supervised learning . This way, the resulting exact posterior marginal probabilities can be used as a benchmark for comparison with BP or CBP. The graphs were generated to have a Watts-Strogatz small-world structure, as the latter share some features with social networks 14 . Such a graph structure is controlled by two parameters: the mean degree for each node, called , and the probability that a connection is "long range" as opposed to local between neighbors, called . --- -PERFORMANCE OF THE BP AND CBP ALGORITHMS We compared the approximate posterior probability solutions found by different message passing schemes with the exact posteriors, , for given external messages . To ensure the generality of the results, this comparison was performed for several randomly generated graph structures and in response to numerous sets of randomly generated external messages . Each dot in Fig. 2 corresponds to an approximate posterior probability for one node in a given trial. As the density of the network increased, the performance of BP degraded . In particular, all beliefs became too extreme, resulting in a condensation of approximate posteriors close to 100% or 0% even when the external evidence did not justify such confidence . Next, we tested CBP after learning the control parameters using a supervised learning method . The parameters and were chosen to minimize the distance between the approximate and true posterior . After optimization, CBP matched the exact inference very closely, with no sign of overconfidence . Such supervised optimization is only possible in networks with a relatively small number of nodes. In larger networks, and any realistic social graph, exact probabilistic inference is impossible because it scales exponentially with the number of nodes. Fortunately, CBP parameters can also be trained without any knowledge of the true posteriors. Using purely local learning rules, the control parameters can be trained to remove correlations between incoming and outgoing messages and to suppress redundancies between incoming messages . Despite the heuristic nature of these learning rules, the approximate posteriors remain close matches to the true posteriors . This toy example demonstrates that CBP can alleviate the overconfidence problem associated with BP in cyclic graphs, resulting in more rational beliefs. Since we move on in the next section to larger graphs where exact inference is intractable, it is assumed that the parameters of CBP were trained for each graph structure using the proposed local, unsupervised learning rules, rather than with supervised learning. --- -TOWARDS GREATER REALISM: LARGER GRAPHS The next step was to investigate how these effects generalize to more realistic social graph structures. First, we investigated larger Watts-Strogatz graphs with 200 nodes. By systematically varying and , we explored the impact of the number of connections per node and long-range connections. These findings will be useful for explaining more complex behavior in "realistic" social graphs . Fig. 3 shows an example graph with moderate degree ( ) and proportion of long-range connections ( ). We provided unreliable external messages that did not strongly support a "yes" or "no" answer. More specifically, in each "trial", each external message was sampled from a Gaussian distribution with zero mean and standard deviation . To understand how opinions are formed, it can be useful to visualize the belief trajectory during the deliberation process, that is, while messages are still being propagated. Fig. 3a, top row, examines the case of BP. Starting from complete uncertainty , the beliefs in the different nodes evolve over the iterations of the BP message passing scheme until they stabilize at constant levels, representing the opinions generated by BP. Differences in opinions among the nodes are induced by random variations in local graph structures and in the external messages the nodes receive. Each new trial generates a different set of opinions . Note that the beliefs converge to very large values , most agents being at least 99% confident in having a correct answer . While it is not tractable to compute the exact posteriors, we can estimate an upper bound on "rationality" . This corresponds to the belief of a universal observer summing all the external messages directly: 1 . Beliefs larger than are necessarily overconfident, since they go beyond the total external evidence. As we can see, BP results in severe overconfidence for most nodes in the graph, despite the true unreliability of external messages. Note that is an upper bound, not an exact posterior. In fact, if inference were exact, the agents would have significantly lower confidence than the universal observer, for two reasons: the nodes do not trust each other completely , and not all of them are connected to all other nodes. In contrast to BP, the CBP algorithm leads to far more moderate opinions , with no sign of radicalization or polarization. The final beliefs are narrowly distributed around a consensus value, which is itself close to zero . The beliefs always remain below the universal observer, as would be expected from a rational deliberation process and are in agreement with the completely uninformative nature of the external messages chosen for these trials. The BP-generated opinions are represented graphically on the top row of Fig. 3b, illustrating how opinions can be distributed as a function of the proximity between two nodes. Only two possible outcomes were observed in those graphs. In the first scenario, the entire population reaches the same extreme opinion, either for or against . We interpret this phenomenon as a radicalization of the entire population. In the second scenario, two populations with opposite but similarly extreme opinions emerge. These populations are separated into 2 or more local clusters within the graph . We interpret this as polarization. We quantify the level of radicalization R as the mean absolute value of the beliefs and the level of polarization P as their mean standard deviation . These definitions correspond to the ones used in other studies 5,9,15,16 . The left panels in Fig. 3a,b have both high radicalization and high polarization, while right panels have high radicalization but low polarization. Note that the only thing differing between the two panels are the external messages . ), abbreviated as the "Probability of yes". Thin lines are connections. The two trials shown here are the same as in , with the top and bottom rows corresponding to BP/CBP. The relationship between the belief and the "probability of yes" is illustrated in Sup.Fig. 1. The radicalization or polarization due to BP and the suppression of these characteristics by CBP are very general results that are independent of the specific network structure, as illustrated in Fig. 4. In the case of BP , the severity of polarization and radicalization systematically depends on the two structural parameters: radicalization increases quasi-linearly with , while polarization decreases with . Interestingly, polarization is strongest in a sweet spot with a moderate and a small value of . This sweet spot corresponds to a high probability of echo chambers, which corresponds to local clusters of highly interconnected nodes that are relatively isolated from the rest of the graph . Fig. 4b examines in more detail the belief distributions resulting from BP at the level of the population when increasing . Note that the distribution has two distinct modes, whose separation increases with . These features are completely suppressed by CBP. Radicalization and polarization are eliminated , and beliefs are no longer separated into two distinct modes. Instead, the distribution presents a single mode, centered at zero, with a variance increasing with . Preventing radicalization and polarization is not sufficient per se . One must also ensure that the message passing scheme operates properly when external messages are actually informative, that is, when they globally provide more support for one option than the other. That is why we now consider a situation in which there is a true answer supported by evidence . The task of the network is now twofold. First, there should be as many agents as possible whose beliefs point in the direction supported by the evidence . Second, confidence levels should increase in proportion to the true strength of this evidence . We generated informative external messages by sampling them from a biased distribution . This bias was small compared to the variance of the distribution, such that many individual nodes received misleading external messages . Moreover, these external messages were injected into only a small portion of the nodes, while others received no external messages . If the network allows all users to share information optimally, every agent should believe in the answers supported by the highest amount of evidence even if their private external message points in the opposite direction . In investigating inference in the presence of informative messages, we found an interesting dissociation in performance when considering people's choices or their confidence levels. People's choice would correspond to their answer to a survey with only two possible options . Presumably, they would choose the answer they believe the most, that is, answer "yes" if their belief is positive. In contrast, people's confidence would correspond to the absolute value of their beliefs . Let us first consider choices. In a strongly connected network with small mean path length between nodes , the portion of nodes with the "correct choice" after running either BP or CBP increases similarly to the proportion of informed nodes . Moreover, this increase is perfectly predicted by a universal observer summing all the external messages together, whose belief is . In a network containing less long range connections with a longer mean path length , both BP and CBP perform worse than the universal observer, reflecting the limitations introduced by the more indirect communication between nodes. However, CBP now clearly outperforms BP . To intuitively understand why a smaller number of long range connection results in poorer choices, consider an extreme scenario: a network with no long range connections at all , in which case all nodes are organized on a fat ring, with subpopulations at opposite ends having no direct connections. They can only influence each other indirectly by changing the beliefs of intermediate nodes, which is not possible if those nodes are radicalized . By keeping beliefs graded, CBP restores long range communication within the network. Where BP and CBP most strikingly differ is in their confidence levels . In the case of BP, beliefs are always distributed in two extreme modes, leaving no room for uncertainty . As more evidence arrives in support of a positive choice , the proportion of belief in the positive mode increases, but the nodes that are still in the negative mode remain equally overconfident . When a node finally changes its mind, it can only switch between these two extremes, with no intermediate stage of uncertainty. Such phenomena could have potentially deleterious societal consequences: people convinced of their correctness could reject the vaccine at any cost and become impervious to information campaigns and contrary evidence; even if they change their minds, one form of extremism could lead to the opposite one. In contrast, with CBP, the beliefs are far less extreme, and their unimodal distribution gradually shifts toward the positive side as more evidence is provided . In other words, the stronger the evidence, the more confident the correct nodes are of being right. Conversely, the incorrect agents become less confident, as should occur following a rational consensus building process. --- Figure 5. Responses of the 200-node small world graphs to informative external messages. Average of 6 graphs with = 30, = 0.2. Same as but for = 20, = 0.08. Left panels: choice performance of the different message passing schemes. Black: BP, red: CBP, dashed: universal observer, dotted: based on external messages only, without taking into account the internal messages. "% Node informed" is the percentage of nodes receiving non-zero external messages, "% Nodes correct" is the percentage of nodes with a belief whose sign points to the true answer . Middle panels: Belief distribution over all nodes for increasing amounts of external information as a result of CBP. Thin line: 1% nodes informed, normal line: 5% of nodes informed, thick line: 10% of nodes informed. Right panels: Same as the middle panels, but for BP. The more complex structures of these networks made it possible to investigate in more detail the relationships between local graph structures, control parameters , and beliefs . In the case of BP, overconfidence is directly proportional to the degree of the node being considered . Thus, the most connected nodes develop more extreme views. In contrast, CBP results in far more moderate beliefs and globally weakens the relationship between confidence and node degree . CBP achieves this control by learning to decrease the gains and increase the loop corrections in nodes of larger degree . In other words, CBP needs to exert stronger controls on nodes that are most massively connected to the rest of the network and are thus at the largest risk of becoming radicalized. --- -REAL SOCIAL-NETWORK EXAMPLES To investigate the information sharing capabilities of these networks, we tested them with informative messages provided to small subsets of the nodes, as previously done. In these larger and more modular networks , both BP and CBP unsurprisingly perform worse than a universal observer . However, CBP strongly outperforms BP. As before, BP exhibits extreme overconfidence, regardless of whether the nodes are correct or incorrect in their choices . While the distribution of BP-generated beliefs appears unimodal, it is in fact a consequence of the naturally wide distribution of node degrees in social graphs. If only nodes of similar degrees are combined, the distribution of belief once again becomes bimodal , and the separation between the two modes increases with the degree exactly as in Fig. 4b . In contrast, the beliefs generated by CBP remain unimodal at all degrees and moderate but with a marked shift and extension toward larger confidence levels as more external information is provided . In other words, correct nodes become more confident, while incorrect nodes become less so. Finally, for both BP and CBP, nodes are more likely to be correct and confident if their degrees are larger, that is, if they directly collect messages from a larger portion of the network. This is why the CBP belief distribution not only shifts but also extends to the right as evidence increases. We can predict from these results what would be the consequences of willfully spreading fake news on people's choices and confidence. Both BP and CBP are relatively resilient when it comes to choices: they integrate all the external messages. Fake news would have to overwhelm "real news" to cause a global change in people's choices. However, the most detrimental effect by far is the potential creation of a small number of extremely polarized nodes, with contrafactual but unshakable beliefs . This does not take place when reverberation in echo chambers is controlled . In this case, fake news decreases the mean confidence level but without causing the emergence of extremism . --- DISCUSSION Social media networks have always been the theater of repeated questioning and reassessment of ideas that were previously considered unshakable. On the one hand, they have repeatedly demonstrated their invaluable power in bringing together thousands of people to support common important causes of the 21 st century. This was notably illustrated by the MeToo hashtag, which is famous today for denouncing sexual harassment and abuse, allowing the empowerment of survivors and often forcing societal actions against perpetrators. On the other hand, the way social networks shape public debate has also been exacerbated by populist parties and supporters of conspiracy theories 18 . This last phenomenon appears to directly benefit from real-world uncertainty 15,19 as well as from the viral spreading of information that may reinforce a monolithic view 20 . Beyond simply modeling echo-chambers in social networks , Circular Belief Propagation allows for a solution to moderate overconfidence, going against the effects of echo-chambers and current recommendation systems. Numerous other theoretical models have been proposed to describe opinion formation in social networks 5,6,[21][22][23][24][25][26][27][28] . Some, like the famous voter model, are simple enough to allow complete mathematical analysis 25 . Others have used Ising models 24 or investigated the specificity of small world social graphs 26 . While many previous models have used binary opinions, others represented them on a continuum of belief 27 such as our model. However, all these models fundamentally differ from ours, not only in their mathematical details, but more importantly in their starting point and objective. All previously cited models take descriptive approaches: their starting point is the description of how agents locally interact and their goal is to understand the emergence of collective dynamics. On the contrary, our approach is normative: its starting point is a functional hypothesis about the purpose of communication -more precisely, that our opinions are formed optimally when considering the whole external information and the levels of trust between individuals . Finally, the objective of our approach is to find strategies to achieve this function, or come close to it. We propose that the root of the radicalization problem is not disinformation or cognitive biases per se . Rather, online message reverberation leads to systematic overconfidence, as information is unknowingly amplified in echo chambers. By using a normative approach of opinion formation in a social graph and exploring graphs of progressive complexity, we quantified these phenomena and demonstrated its generality. In popular online social networks , the resulting strength of convictions will largely exceed the available evidence and irremediably lead to the emergence of incompatible world views in different communities . Confidence levels may become so extreme that opinions are virtually unshakable, remaining the same regardless of the amount of contradictory evidence. Borrowing from variational methods of approximate inference in graphs, we proposed that the Circular Belief Propagation algorithm can alleviate these detrimental effects. This algorithm learns to suppress messages according to how predictable they truly are. In small graphs, we showed that CBP achieves close to optimal performance: the social network generates confidence levels that go hand-in-hand with the amount of available evidence. In larger graphs and realistic social networks, CBP avoids radicalization and polarization and ensures that beliefs remain rational. We are social beings who have exchanged information for millions of years. It would be surprising if we did not have inbuilt cognitive and social strategies to deal with echo chambers in local communities. What may have changed recently is rather the scale and speed of social communication compared to what was previously possible 1 . The worsening trend appears to be domain-general and applies to many different fields. Beyond the political scene that is often taken as an illustration 30 , the scientific community is not immune to overconfidence, particularly since scientific debates have spread from polite but limited academic circles to social media. As a recent example, the results of a trial on the clinical and brain effects of psychedelics in depressive disorders 31 were vividly discussed online with unusual levels of passion even for a scientific debate . Interestingly, problems that are naturally associated with highly interactive social communication may have their counterpart in the maze of our brain cells, another example of large-scale cyclic graphs. Indeed, the CBP algorithm used here was originally proposed in the context of hierarchical brain structures to investigate reverberations in feedforward/feedback circuits, causing so-called circular inferences 33 . Controlling for reverberations in the brain could involve ubiquitous neural mechanisms such as enforcing the excitatory-to-inhibitory balance 34 and account for puzzling perceptual phenomena such as bistable perception 35 . Our simplified model of how communication changes people's opinions does not incorporate numerous aspects of social media communication. For instance, messages are not systematically broadcasted and connections are not necessarily symmetrical . Incorporating this new element into the model would limit polarization. Besides, while we considered stable states after unlimited message exchanges, temporal aspects were ignored. In real life, a piece of news is only propagated for a limited amount of time before becoming obsolete, and our beliefs are constantly updated as new information arrives. On the other hand, we also did not incorporate phenomena that could amplify the severity of echo chambers, such as biased information access , past individual history and priors, or recommendation systems based on preferences 6 that might be used by social media to reinforce the weight of past online activities. Additionally, the present model considers fixed and positive connections, while individuals tend to communicate only with people having similar convictions 37 , may distrust others 38 , and may even actively distrust people with opposite convictions 39 . This last phenomenon would favor polarization. Lastly, the model only tackles communication over one particular topic, although people form opinions on many questions, and discussion about a subject influences our thoughts on related subjects 9 . Despite these theoretical limitations, going towards an experimental validation of the model would be a giant leap forward. Simple online or offline experiments have been proposed, and could potentially be modeled with either BP or CBP. Future work will have to determine how the change brought by our proposed model could be implemented or promoted in real life, as this proposed solution remains theoretical for now. One way would be to inform people by displaying a measure of local polarization caused by the structure of their local interaction graph. This could make users integrate information differently, possibly in a CBP manner. Another way would be to act on recommendation systems by designing them to promote open-mindedness, which could help break echo chambers. This could mean reordering posts on social feeds to propose content according to their unpredictability for the user. This reordering could be monitored by users, for instance through a novelty scale. All messages were updated simultaneously for a total of 100 iterations, using . The coupling function used in CBP is: where since coupling strengths were taken to be positive. Note that this function closely relates to the one used in other models 5,10 : . is bounded between and and has a sigmoidal shape. --- Parameter optimization Control parameters for CBP were adjusted to the specific graph structure in order to improve inference as compared to BP. We considered two methods, supervised learning or unsupervised learning. In supervised learning optimization , the exact marginals were computed using the junction tree algorithm. The control parameters were optimized by minimizing with supervised learning the mean squared error between the exact marginals and the ones from CBP over a set of 300 training examples : To propose unsupervised learning rules , we noted that when the BP algorithm runs on a non-cyclic graph , messages in opposite directions and come from completely disjoint parts of the graph and are therefore uncorrelated. The same is true for different incoming messages to the same node . When external messages are uninformative -and thus uncorrelated -, these internal messages also remain uncorrelated. In contrast, in a cyclic graph, BP results in undue correlations of these opposite messages, which is a direct signature of information reverberation and overcounting in the graph 40 . We thus used learning rules on control parameters that aim at suppressing these detrimental correlations and ensure that they did not result in spurious belief amplification. We generated 2000 training trials with uninformative external messages. After being initialized as their default BP values , control parameters were updated after each trial as follows: The learning rates were adjusted to ensure that control parameters properly converged within the training window. Because coupling weights are positive, the learning rule for enforces uncorrelated incoming and outgoing messages and . The learning rules for gainmodulates beliefs according to how strongly incoming and external messages are correlated with each-other, and therefore fights against spurious belief amplifications. Importantly, we checked that these learning rules applied to an acyclic graph converge to which corresponds to the BP algorithm . This purely heuristic approach results in suboptimal inference but nevertheless, can suppress polarization while improving the information sharing ability of the model social networks. --- Measures of radicalization of polarization In Fig. 4, radicalization was computed by averaging the mean absolute belief over all nodes, test trials and network structures . Polarization was measured as the standard deviation of the beliefs, computed over nodes within a single trial, and then averaged over trials and network structures. The Supplementary Material is available here: --- METHODS Here we describe how to reproduce the simulation results. For the theoretical foundation of BP and CBP equations, see the Supplementary Material -Math Appendix. --- Social graph models Social graphs were formalized as Ising models with coupling strengths and biases corresponding to the external messages . Watts-Strogatz small-world graphs were generated as follows. First, a ring network was constructed by connecting each node to its neighbors on the right and left. Next, with a probability , this local connection was transformed into a long range connection between two randomly selected nodes. The structure of the realistic social networks were obtained from open source data, https://snap.stanford.edu/data/ego-Facebook.html for data from Facebook© and https://snap.stanford.edu/data/ego-Twitter.html for data from Twitter©. We assumed that coupling strengths were positive . For each graph, coupling strengths were selected from a uniform distribution between and . We chose = 0.6 for 10 node graphs, 0.36 for 200 node graphs, and 0.18 for realistic social graphs. --- Generation of external messages Uninformative messages where sampled independently from a Gaussian distribution with mean equal to 0 and standard deviation or . Informative external messages were sampled independently from a Gaussian distribution with mean +/-0.05 and equal variance . These informative external messages were provided sparsely to only a portion of the nodes , where is the number of nodes in the graph and corresponds to the number of nodes receiving non-zero external messages (the proportion of informed nodes is ). For each value of , we generated 200 sets of informative external messages, each sampled independently from the same Gaussian distribution. Each time, these messages were fed to a different random selection of nodes. After running the BP or CBP algorithm, we measured the final percentage of nodes with , which we called "percentage of correct nodes". This percentage was averaged over the 200 trials. In the case of the 200 node toy models , this was also averaged over 6 different random networks generated with the same structural parameters K and . --- Message passing algorithms After being initialized at , messages were propagated according to a damped version of the update equation provided in the Results section :

It was the best of times, it was the worst of times… (Charles Dickens)TEASER (88 characters): Decorrelating messages in social media feeds could limit radicalization in echo chambers.

Introduction According to the World Tourism Organization [1], tourism receipts reach US$ 919 billion. There was a slowdown in the tourism sector due to the global economic crisis of 2008, which caused some instability in the confidence of agents for future events. However, despite the crisis, to make a comparative analysis of ten years prior to 2010, the foreign exchange earnings from tourism in 2008 was 92.67% higher than in 1999 , demonstrating a significant growth in a decade. Also according to the UNWTO [1], between 1999 and 2010, the international flow of tourism in the world grew by 49%, beating the mark of 935 million trips in 2010, an estimated 1.6 billion in 2020. The Figure 1 illustrates the growth of the sector during the period. The UNWTO researches of 2010 also present important data of the tourism sector in Brazil, indicating that in the same period the foreign exchange earnings more than tripled from US$ 1.81 billion in 1999 to US$ 5.92 billion in 2010. Fig. 1: Growth of the tourism sector Contextualizing the city of Teresina, in this article, with about 850,000 inhabitants, Piauí state capital, Brazil, Northeast, one of the poorest in the country, but whose capital emerged as a respectable health pole with qualified service as a result of the entrepreneurial spirit of its main managers. The Health Pole of Teresina has shown strong growth in recent years, becoming a regional referral center. A first diagnosis made by Piaui Hospitals' Union shows that 30% of clients served come from six other states of the federation, generating jobs and income for the city [2]. The beginning of the development of a health cluster, with the formation of a supply chain and connected services, bringing benefits in terms of cost reduction, quality improvement, training of skilled labor, capital attraction and generation of employment and income. This benefit extends to the Health Private Network, increasing their participation in services to the population through an investment which reached US$ 68 million reais in 2011, according to the Medical Union of Piaui The central objective of this article is to demonstrate the contribution of health tourism as an encouraging instrument for the beginning and the growth of companies as well as a mechanism that contributes to the company to assume more effectively its social function, contributing to improving the quality of life of a particular region and also as immediate growthinducing factor of Teresina. --- Definition of Tourism and its Classifications Tourism can be focused as a phenomenon that refers to the movement of people within their own country or crossing national borders . This movement reveals elements such as interactions and individual or team relationships, human understanding, feelings, perceptions, motivations, pressures, satisfaction, the concept of pleasure, etc [3]. According to Castelli [4], the comprehension of the phenomenon of tourism current must necessarily pass through an analysis of the meaning of the trips in the course of history. These, frequently, were moved by economic political and military interests. Nowadays, travels with these same objectives continue to move people from one region to another. Travelling is an action arising from a context which is inserted into the society at a specific moment in history. It has always been one of the elements of economic and social life and, especially, of the world which is inserted. To each type of civilization or society corresponds a way to travel or accept the traveler. As stated by Wahab [3], the anatomy of the phenomenon of tourism would be basically composed of three elements: the man , the space , and time . These elements are representative of the conditions of this phenomenon existence. However, other factors distinguish tourism from the simple act of traveling. Such factors relate mainly to the goals, the temporary nature of displacement, the use of tourist services and equipment and, what would be the most important among them, the concept of pleasure and recreation as fundamental. Olimpio [5] affirms that a revolution in the traditional concepts of trips was performed by the tourism phenomenon from the middle of the nineteenth century when the Protestant Pastor Thomas Cook inaugurated the first tourist agency in England. Countries such as Italy, France and Spain, old military powers and colonialists, managed to stabilize their balances of payments in the postwar due to tourism, reaching to the point of this activity is to constitute real industry, with its own public body, specific legislation, which aims to regulate the various aspects of social, economic, commercial and cultural tourism, always aiming at better income and greater expansion of the tourist industry. --- Factors that Influence Tourist Decisions Every person is a tourist in potential, being necessary, however, the action of the tourist trip to characterize as so. For this reason there are some constraints as: the desire and the possibility [5]. The desire can originate from different causes such as: propaganda, the status, habit, physical convenience, moral or intellectual, seduction, commercial interest or of profit and health. The second constraint involves factors that are beyond the control of man, as: time vacant, money, means of transport, among others. As the possibility was always easier to exist in the most privileged classes, the concept of tourism is usually bound by luxury. However, with paid vacations, social tourism became easier and popular. In this way, there are various reasons that people become tourist: business, religious reasons, health, culture, education and pleasure are just a few examples. According to Arrillaga [6], the needs that tourism satisfy can be very varied, since the subjective causes that determine the tourist trips are as diverse as the needs which the body or the human soul may experience. The tourist motivations or subjective causes of tourism can be classified into primary and secondary; direct and indirect; proximate and remote; individual and social. However, the tourist deciding to travel generally has more than one cause, as shown below. -Primary and Secondary causes. For example, a person who performs a pilgrimage has, as main cause, the obtaining of a grace of the spiritual order, however, also weighs in its decision the possibility to get to new places, visit the famous monuments, rest, among others. -Direct and indirect causes. The achievement of a certain journey may occur due to an invitation or the desire to know some locality. However, it can involve indirect causes, such as the habit of travel or the socioeconomic level of the traveler. -Proximate and remote causes. As an example of proximate cause of a trip, is the propaganda of a travel agent and as remote cause example, the remembrance of a previous trip. -Individual and social causes. The Individual Causes act in the decision of a person to travel and the Social Causes influence by equal in sectors of the population. As an example of the first, we can mention the practice of a sport and of the second, the fashion or ideological affinities. The particular motivation for health tourism allows to consider it due to primary, direct, proximate and individual causes. --- Tourism Classifications According to a first general classification according to Andrade [7], the city of Teresina case and its health pole may be classified as a regional receptive nucleus of people for health treatment, and internal, because the temporary flow is typically regional in the country itself. The Table 1 exposes in objective way the more common tourism types. Is the set of activities of travel, accommodation, food and leisure activities practiced by travelers relating both to commercial and industrial activities; Active sports Refer to all the activities of travel with an objective to participate in sporting events, in the country or abroad; Health Also known as therapeutic tourism or treatment, refers to the set of tourist activities to acquire good physical and mental health; Religious Set of activities that involve visits to the receptive that express mystics feelings or raise the faith, hope and charity to believers; For purpose of this work we will use the typology of health tourism, because that is what happens in some health poles regional throughout Brazil as, for example, the focus of this paper, the case of Teresina. Table 2 Refers to the hosting and the set of activities in a same core, with duration of at least three weeks. This method excludes the tours and excursions to other receptive; Itinerante Involves a series of lodging in different places, consists of visits to the greatest possible number of receptive cores, on a single trip, with short stay in the visited places. According to the form, you can classify the people who migrate to Teresina for health treatment in individual, because most of the time travelers get to the city with its own resources by regional and interstate bus that cut the city through the various highways in the state. --- Tourism Industry The internal relationship that exists between tourism and the economic science is always expressed in terms of the contribution of tourism to economic development. Wahab [3] writes that: "Tourism is a phenomenon that involves the transfer of capital from one country to another through the movement of tourists who go to certain touristic"product" and consume. They are potential consumers of complex goods and services that are offered with a specific goal. The tourism, through the aspects of consumption and investment, affects different sectors of the economic system of a country, and it is believed that its multiplier effect is higher than that observed in other sectors of the economy such as industry. .. " . Tourism is productive, accurate and determined with specific characteristics, joined to the field of development and commerce. So, depending on the stage of development of a country or a region, it can be considered the first force, ahead of industry and agriculture. According to data from the Brazilian Tourism Company -Embratur [8], the tourist industry in recent decades has shown representative economic value. This fact is evidenced by the volume of transactions carried out as a result of increased demand for travel and tourism. The following part of this paper, will discuss some economic concepts that are directly linked to the tourism industry of the real economy of a region. --- Economic Concepts Some economic concepts that influence the study of tourism as an industry are: economic good, utility, economic agents, tourist product, demand and supply of tourism. As Lage [9] presents, all that is rare and exists in less quantity than needs is an economic good. Thus, because of this lack, something need to be produced, taking the form of goods or services. As the consumers cannot get everything they want, they are forced to make choices. Therefore, they may have preference for a given good service but accept a minor amount of another. However, in any situation, consumers act rationally in order to obtain maximum satisfaction from their spending. --- Economic and Social Implications of Tourism As state by Baptista [10], tourism advantages for a country or a receiving state are: -Increase of revenues in foreign currency from the sale of goods and services; -Creation of new sources of revenue in various economic sectors; -Low investment compared to the revenue stream that it promotes; -The integration between people of regions, languages, habits and different tastes. The economic impacts generated by tourism can be classified according to Gunn [9] as : direct impact, known as the total income generated in the tourist sector as a direct result of the variation in spending on these products; indirect impact, the total income generated by the spending of the tourism sectors in goods and services offered in the economy, and induced impact, a result of the direct and indirect impacts of tourism, income levels increase throughout the economy, and that part of this additional income will be spent on goods and services produced domestically. 5 The implication of health tourism at Teresina city As a result of these impacts related to health tourism in Piaui capital, it is possible to show in an objective way, that an integrated set of these impacts on the local economy, created 15,000 direct jobs and an annual average investment of 80 million reais order of direct and indirectly related to sectors linked with the area in question. The tourism industry has the positive impact on the economy of a city, according to Andrade [7], by: increased income of the visited place by foreign exchange inflows; stimulation of investments and creation of new jobs; the redistribution of wealth between the places of origin and destination. As negative impacts are mentioned: inflationary pressure, being harmful to the population of the tourist regions, because the high prices in general also affect goods and essential services; high dependence on tourism, making the economy of the region vulnerable to seasonal fluctuations in demand for tourism products; and the social and environmental costs in the tourist areas and residents; So, with these definitions and characteristics of tourism, it is clear the importance and complexity of the tourism for sector in a particular country or region, since it can not only brings capital exchange, but also prosperity to a particular community with regard to professional qualifications, better quality of life with a better performance of the services offered to the population. --- Conclusion Tourism is an industry with multivariate interests. Bringing the tourist to a particular location has different motivations. A developing region, like the Brazilian Northeast, needs to use all possible ways to seek an economic and socially sustainable development that adds value to their economy. The pursuit of excellence in health services, as demonstrated by the Teresina case, is an economic and social way of doing this, especially in regions waiting fordevelopment.

In recent years, the global growth of the economy contributed to the significant increase of tourism in the world, spreading its concepts in Brazil. There have also grown the related researches in the academic sector, as well as its concern with long life companies, growth in the Northeast region, globalization and other changes occurring in the short term. Thus, this work aims to present how a state among those with lower income per capita in the Federation, as Piauí, has a health care system in its capital which is a benchmark in Northeastern Brazil, influencing positively health tourism. The work deals with issues involving tourism as a product and service, having as a result the existence of a cluster of health in the city of Teresina with economic expression that justifies its importance as a generator of income, employment, science and technology and a promoter of sustainable development in that city.

Background Globally, an estimated 2 million children have lost their lives over the past decade to armed conflict, 6 million have been left severely injured or disabled, 12 million have been left destitute and 300,000 children have served as child soldiers . The stress and trauma of armed conflict can be particularly impactful to adolescent mental health. The period between childhood and adolescence is a developmental period characterized by, "complex and multiple changes across biological, psychological and interpersonal domains" . This complex developmental period makes adolescents vulnerable to potentially traumatic events that can have immediate and long-term impact on social, affective and cognitive development. The psychological impact of armed conflict on adolescents can increase symptoms of post-traumatic stress disorder , anxiety, depression , cause psychophysiological disturbances such as nightmares and trouble sleeping, fear, grief, behavioral problems , and reduce educational attainment and performance, lack of hope and cause personality changes . However, recent research has also indicated that adolescents exposed to armed conflict do not necessarily develop psychopathology, and, in contrast, many adolescents demonstrate remarkable resilience. Research has increasingly focused on resilience, seeking to identify factors that allow some individuals to cope better than others when faced with adversity . Adolescents coconstruct their social worlds through relationships with family, peers, and communities, and these relationships can be instrumental in promoting resilience . Parents/caretakers provide both material and emotional support to their children, and also model cognitive and behavioral strategies for coping with trauma and other forms of stress. Protective factors at the family level include provision of support, stability and parental monitoring . Peer relationships and feeling supported by peers helps prevent isolation and can provide additional sources of emotional support. At the community level, attachment to community through school or churches, community acceptance, and access to social support have been associated with adolescent resilience . --- Adolescent Coping Strategies Lazarus and Folkman originated the term "coping" to describe responses to stress . Lazarus and Folkman defined stress as a condition or feeling experienced when a person appraises an event as "exceeding his or her resources and as endangering well-being" . More recent research has defined coping as, "the set of cognitive and behavioral strategies used by an individual to manage the demands of stressful situations" . Coping is effective if stress is accurately appraised and specific behavioral and cognitive strategies are used to manage, reduce or tolerate stressful events . Coping strategies are important because employment of coping strategies can influence psychosocial, emotional and behavioral functioning of adolescents . The risk or benefit of coping strategies is context dependent and what is adaptive in one context may be maladaptive under different circumstances . In humanitarian emergencies and conflict settings, problem focused coping may be less appropriate than emotion focused coping because emotion focused coping may be the only practical strategy available when fixing the problem or stressor is not a achievable in that context . The majority of qualitative research on coping strategies among conflict-affected adolescents have included older adolescents, however, the findings from this older age group offer insight into how cultural context can motivate use and acceptance of particular coping strategies among adolescents. Qualitative case-study research with Cambodian refugees ages 14-20, found that avoidant coping strategies were common among traumatized Cambodian refugees and use of these strategies may have been motivated by a cultural climate that sought to avoid thoughts, behaviors and activities that reminded them of "dishonorable events in Cambodian history" and collective shame felt by Cambodians . Qualitative narrative research with 14 Sudanese refugees ages 16-18, found that suppression and distraction were common coping strategies and use of these strategies may be motivated by a sense of communal self and as protection against "feelings that they feel powerless to handle" . Research with Zimbabwean adolescents ages 17-19, found greater use of emotion focused strategies rather than problem solving strategies because cultural norms discouraged problem solving strategies that could involve confrontation or challenging of elders . In addition, gender norms within a culture can shape use of coping strategies. A mixed method study among Palestinian refugee adolescents ages 8-17 living in Gaza found that gender norms influenced coping strategy use so that girls were less likely to use leisure or relaxation coping strategies, including going shopping or exercising, that would require them to be far from home, whereas boys were more likely to use these coping strategies . The Democratic Republic of Congo , the setting of this study, has endured decades of destruction to health and social infrastructure . A history of colonialism, theft of the DRC's mineral wealth and strategic 'pitting' of ethnic groups against one another has resulted in prolonged conflict and persisting insecurity. The effects of conflict have been particularly profound in rural areas of the South Kivu Province in Eastern DRC. High levels of poverty, limited livelihood and employment opportunities, population displacement resulting from conflict and destruction of health and education institutions has weakened local capacity to respond to adolescent mental health needs. Previous research in eastern DRC found that among 477 girls and 569 boys ages 13-21, approximately 95% had experienced at least one traumatic event; and, on average adolescent were exposed to 4.7 traumatic events over their lifetime, with 52% meeting the criteria for PTSD . Within this context, adolescents, families and communities are rebuilding their households and communities to improve the life and livelihoods of future generations. Despite recognition that coping must be understood from a contextually grounded perspective, limited evidence is available to specify how adolescents cope within conflict and post-conflict settings and which strategies are most beneficial to adolescent mental health. Qualitative techniques such as focus groups and individual interviews define and provide greater depth of knowledge as to the cognitive and behavioral coping strategies utilized by conflict-affected adolescents. Previous research in this context utilized a grounded theory perspective to analyze Congolese adolescents in-depth interviews to gain the perspective of adolescents on trauma and coping . The current study sought to extend this theoretical approach to adolescent's parents/caregivers and key informants within the same contextual setting to elicit understanding of how adults in the community perceived adolescent exposure to trauma and coping strategies utilized in response to trauma. --- Research Aim The purpose of the study is to report mothers/female caregivers and key informant's perceptions of 10-15 year old adolescent exposures to traumatic events, utilization of coping strategies in response to traumatic events, and insight on opportunities for supporting beneficial and mitigating harmful coping strategies to promote resilience among adolescents in a rural, low-resource, post-conflict setting. --- Methods --- Study Setting This study took place in rural villages in the Walungu territory in South Kivu, Eastern DRC. The Walungu territory is 50km south of Bukavu, the capital of South Kivu and has been afflicted by conflict since 1999. The study was conducted in partnership with Programme d'Appui aux Initiatives Economiques , a non-profit, non-governmental Congolese microfinance organization and was nested within two microfinance projects, Rabbits for Resilience and Pigs for Peace . RFR and PFP are pragmatic community trials to test the effectiveness of an adolescent-led rabbit animal husbandry microfinance program combined with an adult pig animal husbandry microfinance program, with the aim of improving health, economic stability and relationships between families and communities through loans and repayment of pigs and rabbits . Male or female adolescents ages 10-15 years were eligible for participation with RFR. Participants who were ages 16 years and older were eligible for participation in PFP. The current analysis includes qualitative data collected in July 2014 with mothers/female caregivers and key informants within four villages participating in the RFR/PFP programs. --- Sample A purposive sampling strategy was used to identify parents/caregivers of adolescents and key informants enrolled in the parent study. Parents/caregivers and key informants were selected from four of the 10 study villages, Karherwa, Cagombe, Izege and Cize. These villages were selected because of the reported high exposure to conflict-related trauma. Within these villages, baseline data from the parent study was used for purposive sampling of adolescents based on age, gender and exposure to traumatic events. Traumatic exposures were represented by a wide range of experiences including murder of family/friends, having ill health without care, lacking food and water, being seriously injured, being close to death, separation from family, experiences in combat and brainwashing. Adolescents were selected to represent a range of exposure to trauma: low exposure to trauma , medium exposure and high exposure . Within each trauma exposure level, sampling involved achieving a balanced distribution of participants by age and gender. Parents/caregivers of each of the 48 adolescents identified were invited to participate in focus groups. Four focus groups were conducted with mothers/female caregivers, as no fathers or male caregivers were successfully recruited. Individual in-depth interviews were planned with adults having roles and responsibilities in the villages that resulted in relationships with adolescents and that would be able to provide insight into trauma exposure and coping strategies. Ten key informants from the four villages were identified and invited to participate in in-depth interviews. --- Study Procedures Congolese team members actively participated in the design, development, piloting and revision of the key informant interview and focus group guides. The team reviewed and revised interview questions to ensure cultural relevance. The interview guides were translated into local languages, Swahili and Mashi, to be used for interviews and focus groups. The final guides consisted of broad open-ended questions related to the following topics, 1) identification of types of trauma-related experiences, 2) methods for coping utilized by adolescents, 3) perception of gender differences in coping, 4) sources of psychosocial support, and, 5) risks or benefits of coping strategies. Probes were developed with Congolese team members to capture greater depth in participant responses during the focus groups and individual interviews. The focus group guide was constructed of three sections. The first section was a free list exercise, a method which asks participants to list answers in response to a single primary question . Participants in each focus group were asked the primary question, "What are all of the ways that youth think or behave in response to traumatic/stressful life events?" The second section asked participants to consider how the aforementioned cognitive and behavioral strategies improved or harmed different aspects of mental health and well being among adolescents. The third section of the focus group guide was constructed of open-ended questions aimed at probing on gender differences in coping behaviors. After final revisions to the interview and focus group guides, the Congolese research team participated in a two day team training in administration of consent , human subjects research ethics, qualitative interview skills and focus group methodology. The training provided discussion on how to support participants that became distressed or upset during an interview or group discussion. --- Data Collection Mothers/caregivers of eligible adolescents and key informants were provided with information pertaining to the purpose of the study, risks and benefits of participation in the study and were asked to provide verbal informed consent. Unique identification numbers were assigned to all participants, all interviews were conducted in private and no information was shared outside the research team. After mothers/caregivers and key informants provided informed consent the interviewer selected a location for the interview or focus group that would allow for privacy and disclosure during the interview or focus group. Trained male and female Congolese interviewers fluent in French, Swahili and Mashi conducted focus groups and interviews. Initial questions allowed the interviewer/facilitator to develop a level of rapport with community members so that participants would feel comfortable answering the questions. The key informant interview lasted 30-60 minutes in length. Each focus group was conducted by a facilitator and a note-taker and lasted 45-60 minutes. Participants were provided with compensation for their time equal to 2USD, an amount considered appropriate after consultation with village leaders and research team members. --- Ethics Statement The Johns Hopkins School of Medicine Institutional Review Board approved this study on June 23, 2014. A committee of respected Congolese educators at the Universite Catholique at Bukavu reviewed and approved this study, as there is no formal institutional review board in South Kivu. The research team received approval to conduct the research with local partners PAIDEK and by village traditional and administrative leaders. --- Analysis To achieve the study purpose, a grounded theory approach, rooted in a participatory transformative paradigm was utilized . The grounded theory methodological approach is based on the constructivist epistemology, that meaning is cocreated in the discourse between people and grounded in a particular cultural context After completion of interviews and focus groups, a Congolese translator completed translation of all transcripts from French or the local language to English. The translations were crosschecked by researchers based in the US and in consultation with the Congolese research team. The free list data provided an overview of ways adolescents coped with stress and trauma from the mother/caregiver's perspective. The free list data were consolidated into a summary list and ranked in decreasing order of how many respondents reported a coping strategy. After completion of translation and review of focus group questions , and key informant in-depth interviews, the analysis commenced with the grounded theory approach . The steps taken to complete the grounded theory methodological approach included: initial codes developed through line-by-line coding of all key informant interviews and focus group transcripts to identify implicit meanings; initial codes to identify common responses and similarities and differences in responses; focused codes that were developed by using the most significant and/or frequent initial codes; axial coding to represent the content of focused codes and to relate common codes, categories and concepts to each other; and, applying a final coding structure to each transcript using Atlas.ti 6 software . During the application of focused and axial coding, memos were written by the researchers to help identify emergent themes, and iterative inductive content analysis was used to identify connections between themes . Throughout the analysis and interpretation phases, team members ensured the quality of findings through, 1) grounding findings in the data and engaging in debriefing and peer review during the analysis process, 2) documenting the process and ensuring transparency, 3) confirming fairness and balance in selecting representative quotes and exemplars to illustrate themes. --- Results A total of 26 mothers/female caregivers participated in one of four focus groups . The majority of female caregivers were 25 years or older, married, had received no schooling and had an average of four children in the home . The number of people in each focus group ranged from five to eight; 5 from Karherwa, 7 from Cagombe, 8 from Izege, and 6 from Cize. Mothers/caregivers represent both male and female adolescents, ages 10-15, with a range of trauma exposure. Demographics of the adolescents represented by mother/caregivers are presented in Table 1. Fourteen caregivers represented adolescents who reported high trauma exposure , 8 caregivers represented adolescents who had experienced medium trauma exposure and 4 caregivers represented adolescents who had low trauma exposure . Nine key informants were interviewed including six teachers and three traditional village chiefs. Key informant interviews included four females and five males . As noted above, the focus group started with a free list exercise in which participants were asked, "What are all of the ways that youth think or behave in response to traumatic/stressful life events?" Results from the free list analysis are presented in Table 2. Twelve common coping strategies were mentioned and similar concepts were grouped together. For example, "the child isolates themselves" was grouped with "the child stays alone". The free list responses served as a reference throughout the focus group to probe further how the aforementioned coping strategies were related to positive or negative mental health and wellbeing outcomes and gender differences in use and effectiveness of these coping strategies. Therefore, the iterative inductive content analysis sought to connect free list responses to descriptions of trauma experiences that respondents believed motivated use of particular coping strategies and nuanced descriptions of how these coping strategies were perceived as positive or negative. In identification of the emergent themes, the analysis also sought to illuminate any gender differences in coping strategy use or effectiveness. Analysis resulted in three salient themes, 1) Adolescent exposures to trauma promote maladaptive risk taking and contribute to a cycle of violence 2) Gender differences in behavioral and cognitive avoidance, and, 3) Positive effects of social support seeking in response to exposure to trauma. The themes were organized from least helpful/harmful coping strategies , mixed perceptions of the benefit or harm of cognitive and behavioral avoidance, and, finally the most helpful coping strategies . --- Emergent theme: Exposure to trauma, adolescent risk-taking and the cycle of violence Mothers/caregivers and key informants reported past and present trauma and violence experienced by adolescents within the home and in the community. The legacy of the conflict in the DRC has perpetuated insecurity. Militia groups have continued to destabilize the Kivu regions and interviews revealed a wide range of adolescent exposures to conflict related violence. Participants explained that some traumatic experiences resulting from the conflict had been "inherited" by adolescents. For example, rape of women was a common practice of militia during the multiple periods of conflict in the region. Some of these rapes resulted in pregnancy, and survivors gave birth to their attacker's children. A teacher recalled, "For instance, during the war, some women were abducted into the forest by Interahamwe [armed group] elements. They were raped, and some of them got pregnant and eventually delivered. When these women were freed and came home, they brought their children along with them. A lucky woman could be welcome by her husband, but not the rape-born child." Some of these children are rejected from their families and forced to find alternative means of survival. A teacher described mental health effects of children born from militia rape. "The child or adolescent must be affected and afraid, and think that those events will come back. That's why, when you send the child to fetch water in the evening, he can refuse out of fear. Some Interahamwe's violence-born children fight with their friends when these tell them that they are Interahamwe's sons. And they ask their mothers who a Interahamwe is. I have three such children in my religion class. When I asked them what it was like to be born from unknown fathers, they started to weep and said they would kill their fathers if they could see them. After that, they didn't participate in any class activity. These children finished the year without paying anything, because it is difficult to dismiss children in such a situation. Sometimes a child leaves the classroom very angry and determined to join the military for vengeance when he comes of age." A mother/caregiver echoed how trauma resulted in risk taking, "He can react to such a traumatic event by taking alcohol and using drugs, becoming a thief, an armed robber, selling his property in order to get money for hemp, etc. In the end, he can go mad. By becoming an armed robber or a street kid, he can start killing people in the same way as his father was killed." Participants provided examples of how trauma can lead to a cycle of violence that not only impacts adolescents, but also destabilizes communities. A key informant explained, "These events cause children or adults to be broody. Because of their immaturity, children can end up committing suicide or becoming mineral-diggers. Traumatizing a child is killing him with no gun or poison. It unsettles the child and leads to the stagnation of society." Both mothers/caregivers and community members shared concerns that direct experiences of trauma by adolescents and traumatic experiences of family and community members "inherited" by adolescents can lead to adolescent risk-taking and perpetuate violence and instability. Furthermore, cultural belief systems could be potentially harmful to adolescent mental health. Belief that sorcerers exist in communities and community discourse speculating who may be a sorcerer or victim of sorcery could destabilize relationships. Participants reported that a continuing threat of violence was from "poisoning" or other forms of harm from sorcerers. Sorcerers were described as being any age and are believed to have the power to "poison" others or cause ill health. Participants explained that community members often accused others of being sorcerers, and that these accusations had the potential to escalate community violence. A key informant described this impact, "They have little meetings, and fight. For instance, yesterday, an angry adolescent attacked a woman, accusing her of sorcery. They started to beat her up." Another key informant echoed this statement and explained that adolescents themselves can use "sorcery" to survive. "As they've already gone bad, some will stick to alcoholism, magic and sorcery as income-generating activities. If they don't steal, they can't survive. They don't accept advice and help from a wise person in the village. They don't want to see anybody, and just keep doing what pleases them. Abandoning them will result in their total decline." Labeling an adolescent as a sorcerer has potential risks such as causing stigma and isolation by families and communities. Furthermore, the existence of belief systems around sorcery could perpetuate community instability and be an obstacle to the benefit of building community trust and support networks that help to stabilize families and communities. --- Emergent Theme: Gender differences and the consequences of cognitive avoidance and behavioral isolation A common response from participants was that adolescents cope by avoiding the memory of their trauma. There were a variety of ways in which respondents described this coping strategy ranging from cognitive avoidance such as "forgetting" the trauma to behavioral avoidance such as "isolating" from others. The perception of whether this coping strategy was beneficial or not varied. For example, many participants described adolescents "forgetting" the trauma as a positive adaptation and the ultimate goal of recovery from trauma. Many different activities were described as helping children to "forget it all" such as play, work and prayer. Some respondents described difference by gender in the ability to "forget it all". For example, a teacher reported, "Girls are affected differently from boys. Boys forget faster than girls. When something bad happens in the life of a girl, she'll keep alluding to it every time. A boy, however, forgets with time." Another distancing technique described was adolescent's behavioral avoidance by isolation, either self-imposed or isolation resulting from rejection of peers or family. A key informant explained the different ways isolation could occur for a traumatized adolescent. "When such a adolescent is made ill at ease by his peer, he can sometimes decide to follow his father in the mountains, because when his mates want to tease him, they show him the mountains from which his Interahamwe father came. The child can also commit suicide, or grievously wound, in anger, his mates, and that's a serious problem. He can drop out of school, refuse to eat and put the whole family in trouble. He can disappear from the house and become a street kid. Such an adolescent isolates himself from his friends. He can become naughty, smoke hemp, take alcohol, turn into a thief, sleep outside, etc." While cognitive avoidance of the memory of trauma was perceived as a positive adaptation to trauma, behavioral avoidance and isolation from peer, family and community relationships were perceived as particularly harmful to adolescent mental health. Being rejected from family may have gender specific consequences for adolescents. For girls, early marriage and sex work were identified as resulting from the need to fulfill basic needs and a lack of other viable livelihood and income generating opportunities. A teacher explained, "In Kaniola, for example, 8, 9 and 10-year-old girls go to mining zones for prostitution with diggers, when they are abandoned by the family. They behave like mature women." Another caregiver warned of girls engaging in prostitution after having experienced traumatic events, "Girls start seducing boys and become proud. They dress up immorally and don't listen to their parents. They wear trousers, imitate western ladies' behavior, become prostitutes, and hang out with married men." Participants reported that girls might engage in prostitution or early marriage as a coping strategy to gain resources, however other motivations for entering early marriage were suggested. For example, participants explained that marriage might provide emotional support. For example, one teacher explained, "When they're sad, some girls become prostitutes for survival, and get pregnant. When they suffer, girls feel the need to get married quickly in order to end their suffering." --- Emergent Theme: Positive social support can benefit adolescents A key coping strategy thematic in focus groups and interviews was the benefit of seeking social support. Support seeking was described at multiple levels in an adolescent's social ecology. Peer relationships were described as particularly important to adolescent boys. For example a mother/caregiver explained, "Bad friends and neighbors can also have a negative influence on the behavior of a child. If it's a good friendship, the child will be stable and behave in a responsible manner." However, other participants cautioned that these relationships could motivate potentially harmful behavior. A mother/caregiver explained that peers can be, "bad company, my boy was not smoking before he started to hang out with smoking friends." Many interviewees echoed the perception that boys were particularly vulnerable to participating in deviant behavior with peers. Family was also described as a critical source of support for adolescents, particularly the role of the family in providing basic needs . Family was also perceived as an important source of counsel or advice. A caregiver explained the importance of how caregivers should support adolescents, "The child can change if his family helps him, encouraging him to get back on the right path". Respondents also explained that family relationships have the potential to model positive social interactions and would benefit adolescents. For example, a caregiver explained, "What can make a child stable is, first and foremost, happiness in his/her own family. Mutual respect between parents can also help children. They must learn respect from their own parents. Parents must satisfy children's daily needs so that they can be stable." However, respondents also described that lack of family support was a threat to adolescent mental health that could be mitigated through greater community ties. Adolescents without family or who have been rejected from family face additional stressors such as lack of shelter, resources to meet basic needs, and a lack of security. When family support systems are unavailable, participants suggested adolescents rely more heavily on community support systems to supply the resources typically provided by families. Furthermore, participants explained that community members could help support parents/caregivers, helping to guide parenting decisions and investments in adolescents. A teacher explained, "If the family has given him advice, but he doesn't listen, they can turn to the neighbors and his friends for help. In our community, once we've identified such a child, we first visit his family and ask them questions about the child." By providing support to caregivers, community members work with families as a unified team to guide adolescents towards positive coping strategies. A community leader recounted the importance of families working with the community to support adolescents, "The community must get closer to the adolescent, because it can save the life of a slandered child on the brink of suicide. For instance, the adolescent who was rescued by his mother started to sing in the choir afterwards. The community must take care of the adolescent and provide them with vocational training. If an adolescent rejects the community, a friend must be sent to sensitize him. Once he gets back under control, he must receive advice from the community, little by little." Community support systems are integrated into adolescent lives in a variety of forms. In DRC, churches and schools are the most prominent community resources accessible to adolescents and work synergistically with family to provide support adolescents. --- Discussion This study sought to understand how mothers/caregivers and key informants perceived the effects of exposure to trauma on adolescents and coping strategies used in response to trauma. Findings highlight that adolescent exposure to trauma impacts not only the individual but also harms family and communities and can lead to a cycle of violence. In the free list interviews, mothers/caregivers listed mostly maladaptive strategies in response to trauma. These maladaptive strategies often involved risk-taking behavior such as drinking, stealing, and seeking revenge. Importantly, participants described cognitive avoidance or distancing oneself from memories of trauma as beneficial, however behavioral avoidance such as self-isolation or family/community rejection was perceived as particularly harmful. This distinction is important because it parallels findings from other collectivistic contexts that show that cognitive avoidance is often preferred by adolescents in these contexts, and, at the same time, behavioral isolation or lack of engagement in social relationships is harmful to adolescents . Reasons provided for why cognitive avoidance and distancing may be preferred include that adolescents are respecting social norms that encourage accommodation and acceptance in collectivistic societies , that this may be the only strategy available to adolescents , and suggestions that avoidance may be an adaptive strategy for conflict-affected adolescents in the short-term . Importantly, in this study, rather than perceiving "seeking social support" as a coping strategy, caregivers and key informants described social support in the family and community as a solution to remedy maladaptive coping strategies. Respondents explained that while support within the family was important to adolescent mental health, community support systems could also be a resource for youth outside of the family, and could simultaneously support families to better parent their children. --- Implications for Intervention Programming to Support Positive Coping Strategies Findings from this study support intervention planning that utilizes a social ecological approach. Bronfenbrenner fist described social ecological theory in the 1970s to explain human behavior from multiple levels including the individual level , microenvironment and mesoenvironment . Utilizing a social ecological approach, intervention programming should be developed to include peers, parents/caregivers and community members such as teachers, traditional and religious leaders. This approach is useful because it takes into account the complex array of interrelationships, bi-directional feedback loops and opportunities to target multiple influences adolescent developmental trajectories. For example, social support and feeling connected to neighbors and schools has been found to result in better mental health outcomes in children . Developmental science on adolescence indicates that during adolescence there is a strong increase in young people's desire to understand their social world and adolescents are increasingly sensitive to social acceptance . Peer relationships provide opportunities to express feelings, shared experiences, and provide opportunities engage in healthy activities such as civic engagement that can contribute to social cohesion and benefit health and well-being . Reaching out to youth who isolate themselves or are rejected from others is of particular importance because in this study, isolated youth were described as more likely to engage in risk taking activities such as stealing or prostitution. By engaging adolescents in community activities such as sports, church groups, choir, and play, adolescents may be less likely to be stigmatized or isolated and feel closer to peers, family and community members . Investment in community support systems can also help to protect against stigmatization of youth as sorcerers or witches, a belief that was often attributed to "troubled" youth in this study and could result in further rejection from the family and community and motivate risktaking behaviors. Collaboration with local leaders to develop intervention approaches that minimize isolation of youth and emphasize the importance of adolescents as an asset to communities and their future could be effective in breaking a cycle of violence experienced in communities. Future research and programing should also consider how gender could inform intervention programming. Prior research in humanitarian settings has indicated separation from families, limited economic opportunities and weakened community protection mechanisms can result in adolescent girls being at risk for early marriage, sexual exploitation and sexual violence . Teaching girls skills that empower their ability to seek support outside of early marriage is critical not only to protect girls from coercion, but also because girls who delay marriage are more likely to complete more years of school and have better economic opportunities in the future . For example, providing more opportunities for girls to gain social support through peer and community relationships could help create meaningful roles in the community that lead to economic autonomy and empower girl's agency and deter girls from entering early marriage or engaging in prostitution . Educating families on alternatives to early marriage for girls are an important component to intervention programming. For example, research from Jordan found that some families viewed child marriage as a form of "protection", a way for families to keep the 'honour' of their daughters, as well as a means to reduce the number of people in the household and associated economic burden . Girl's sexuality is also often perceived as their sole commodity to be exchanged to gain resources through transactional sex . Providing alternative paths to gain assets for girls could protect against early marriage and transactional sex. Community support of girl's engagement in livelihood activities can provide an alternative solution to meeting basic needs and shift cultural norms to help prevent early marriage or prostitution . Furthermore, provision of opportunities to engage in positive activities outside of the home could have other benefits to mental health and well-being of girls. Having greater mobility and programs outside the home can also be leveraged as potential areas to disseminate health information and access to protection services . In this study, boys were perceived as being particularly vulnerable to negative peer pressure in this context. Research shows that a history of peer rejection can make adolescents more vulnerable to peer influence and result in self organization into peer groups that support problem behavior and lead to more serious consequences such as increases in violence . Furthermore, research indicates that as youth engage in more deviant peer groups, a reciprocal disengagement from parental influence can occur as youth avoid supervision . In this study, mothers/caregivers described how boys were resistant to parental advice. Involving parents in intervention activities could help to facilitate stronger relationships and respect for parent's supervision during adolescence. Participants in this study also suggested that community members could help to guide parents and support positive interactions with boys and girls. Providing activities where boys have the opportunity for peer reinforcement within a structured environment could help deter youth from seeking other types of rewarding peer interactions that can include deviant talk, behavior and attitudes . In this context, participation in sports, youth groups and faith-based groups would be culturally appropriate activities for boys and provide positive environments for boys to interact with peer groups. Previous prevention studies have found that provision of safe spaces for adolescents to participate in activities promoted overall well-being and resilience in both boys and girls . These safe spaces can be designed to target gender specific vulnerabilities. For example, there are promising advances in intervention programming addressing risks to adolescent girls. The Creating Opportunities through Mentorship, Parental involvement and Safe Spaces program is an intervention currently being evaluated in DRC and seeks to offer gender specific interventions for girls through the implementation of structured activities that focused on topics such as self-confidence, building friendships, gender based violence and creating healthy relationships to improve girl's confidence in negotiating sex and understanding their self worth . Future research should continue to implement gender specific interventions aimed at promoting girl's agency, access to activities that support empowerment, and mobility outside the home and reduce vulnerability to coercion, sexual exploitation and early marriage. --- Limitations. Results from this study may not be generalizable to other post-conflict settings as the utility of coping strategies were defined within the cultural context of the Walungu Territory in rural Eastern DRC. However, we provided 'thick description' in order to support transferability in which readers can consider if other contexts are sufficiently similar to assess if findings may be applicable to another context . For example, in more urban technologically advanced contexts, adolescents may engage more with social media to interact with peers and the potential for interventions that include social media could be important. Coping strategies may also differ in societies that are more individualistic and less collectivistic. In DRC, community relationships are an important part of an adolescents' social ecology and are a critical resource to be leveraged in intervention designs . This research is limited to reports from mothers/caregivers and key informants on young adolescents ages 10-15. Future research should investigate perspectives from caregivers and community members on developmental changes in coping strategy use throughout adolescence and compare and contrast those perceptions with reports from adolescents. Results from studies that examine coping strategies used in response to traumatic stress along adolescent developmental trajectories can be used to tailor programs to be developmentally appropriate and enhance the effectiveness of programs seeking to support positive coping in adolescence. --- Conclusion This qualitative research provides a culturally specific understanding of adolescent coping strategies from the perspective of mothers/caregivers and key informants within rural communities that have experienced prolonged conflict. Findings suggest that interventions which support positive coping strategies such as provision of safe spaces for adolescents to engage with peers may promote adolescent mental health resilience in response to trauma. Furthermore, structured and mentored activities facilitated by trusted and respected adults in the community can provide new ways for adolescents to receive support and prevent risky behaviors that can result in further trauma and violence in rural communities in the Eastern DRC. Of importance is the opportunity to include gender specific intervention programming that seeks to reduce risk-taking behaviors that are more common among girls and boys . Interventions should also aim to promote positive aspects of mental health and well being for both boys and girls including promotion of self-esteem, hope for the future, empowerment and empathy. Engagement of positive support structures between adolescents and peers, family and community members in intervention planning can take advantage of the complex, bidirectional and mutually reinforcing relationships within an adolescent's social ecology to ultimately improve the mental health and well-being of adolescents and promote a positive future for communities in post-conflict settings. --- --- Conflict of Interest Statement The authors declare they have no other disclosures. Free list responses from mothers/caregivers in focus groups . --- List of abbreviations What are coping strategies of children who are exposed to traumatic/stressful life events? The child does not talk/remains silent The child cries/cries for no reason The child refuses to work The child steals from others The child isolates themselves from others/stays alone The child is very wicked/naughty/becomes rebellious The child is fearful The child drinks alcohol/smoke The child doesn't play with others The child becomes a street kid/starts begging The child gets involved with sorcery/witchcraft The child seeks to get married Note. Coping Strategies are listed in decreasing order of frequency mentioned

The aim of this qualitative study was to provide a context specific understanding of adolescent (10-15 years of age) exposure and coping with trauma from the perspective of mothers/female caregivers and key informants in rural conflict-affected villages. Focus groups were conducted with mothers/female caregivers (4 groups with total participants n=26) and interviews were conducted with key informants (n=9) in four villages in the Walungu Territory, Eastern Democratic Republic of Congo. Focus group and interview guides were designed to elicit responses by participants on adolescent exposures to trauma and coping strategies. A grounded theory approach was used to identify emergent themes. Mothers/female caregivers and key informants reported use of both cognitive and behavioral coping strategies by adolescents. Cognitive coping strategies included avoidance and trying to forget. Behavioral strategies included self-imposed isolation, risk taking and social support seeking behaviors. Findings indicated community social support was a critical resource for adolescents, particularly for adolescents that did not have adequate support from their family. Supporting positive peer, family and community relationships can help prevent risk taking behaviors and teach positive coping strategies to break a cycle of violence reported in these communities. Recommendations for gender specific intervention programing are highlighted.

Introduction The attack on the U. S. Capitol on January 6, 2021, led to the killing of 5 people, turning it into an unprecedented attack on democracy [1]. Recent evidence documented that the attack was partially coordinated through free-speech social media platforms [1]. Alt-techs such as Parler, Gab, and Telegram constitute new information, communication, and socialization ecosystems, which lack content moderation, and thus provide a parallel online space for ideas that are outside the boundaries of speech permitted on mainstream platforms. As a result, alt-techs facilitate the circulation of biased, inaccurate, misleading, and conspiratorial content at unprecedented levels and, furthermore, have attracted large numbers of partisan users and extremists. As seen in the example of the attack on the U. S. Capitol and evidence from other violent incidents, the recent emergence of alt-techs might even have dangerous effects on the offline world, including democracies as a whole. --- What are alt-techs? Alt-techs have emerged as a new social media phenomenon [2]. Prominent examples include Parler , Gab , 4chan , 8chan , Voat , Gettr , BitChute , Telegram , Discord , and Mastodon . These are part of the broader alt-tech ecosystem which provides alternative websites, platforms, and services specifically targeted toward certain, often partisan or fringe communities. The features of alt-techs are typically not innovative in themselves but rather offer services largely similar in functionality to those found on mainstream platforms . However, different from mainstream social media, alt-techs self-proclaim as "free-speech" platforms and, because of that, eschew content moderation so that users can post content that would not be permitted on mainstream social media. The reasons for the emergence of alt-techs are many-faceted. In recent years, partisan communities have developed a narrative claiming that speech on mainstream platforms is being "censored" for failure to be "politically correct" [3]. Such views can be partially attributed to the increasing content moderation efforts from mainstream platforms to limit misinformation, conspiracy theories, and hate speech. For instance, when Twitter and other platforms increased efforts to ban malicious accounts or flag misinformation, many conservative users migrated to Gab and Parler [4,1]. Conservative thought leaders also jumped on the bandwagon by endorsing, in particular, Parler as an alternative to mainstream social media [5]. Similarly, many liberals recently endorsed Mastodon as a Twitter alternative being dissatisfied with Elon Musk's new content moderation policy [6]. As such, migrating to alt-techs can be seen as a political statement. Overall, a mixture of content restrictions, deplatforming, and bans imposed by "big tech" has contributed to the popularity of alt-techs. A particular characteristic of alt-techs is their user base. Especially during the 2020 U. S. presidential election, many alt-techs have witnessed stark growth; e. g., Parler counted around one million users in June 2020 but more than 13 million only six months later in January 2021 [5]. The user base of alt-techs tends to be rather homogeneous: a large share of users even self-identify as partisans, extremists, or conspiracy theorists [5,7]. Mirroring the user base, the content on alt-techs covers-to a large extent-extreme viewpoints or conspiracy theories. For example, Gab features high levels of hate speech, toxic, and antisemitic content [7,2,4]. Furthermore, on Parler, hashtags such as #qanon are widespread and posts frequently link to websites known for spreading misinformation [5]. As such, alt-techs form ideologically-driven ecosystems where opposing viewpoints are largely absent [8]. This can be problematic: Even in mainstream social media, where users are regularly exposed to diverse ideological content, echo chambers of users sharing similar worldviews have led to increased political polarization [9]. In the absence of opposing views, alt-techs are likely to accelerate the growth of echo chambers and further benefit polarization and radicalization [8]. --- Why alt-techs are a threat to society Alt-techs pose direct concerns for society, particularly because they have been frequently associated with violent incidents in the offline world. For instance, the offender in the Buffalo shooting on May 14, 2022, allegedly shot 10 people in a racist attack inspired by content on 4chan [10]. Furthermore, Gab was the main communication channel during the Pittsburgh synagogue shooting on October 27, 2018, a white supremacist terrorist attack that caused 11 people to lose their lives [2]. A similar role for alt-techs has been observed for the attack on the U. S. Capitol on January 6, 2021. Rioters communicated via Parler on how to evade police forces or smuggle weapons into the capitol [1]. As such, Parler is nowadays regarded as the primary communication channel of the rioters before and during the Capitol attack [1,11]. Alt-techs further provide fertile grounds for misinformation and even disinformation. This includes deliberately deceptive content, with which users seek to spread conspiracy theories, false rumors, hoaxes, and inflammatory opinions to promote their own ideological viewpoints, decrease trust in mainstream institutions, and win others as followers [8]. The proliferation of such content on alt-techs may be partially attributed to their promise of "uncensored" speech. Specifically, the lack of content moderation allows spreaders of misinformation to fill the curiosity gap by sensationalizing "censored" content . Another driver that may promote the spread of misinformation is the homogeneity of users. Due to the segregation, confrontation and exchange with other views are rare and thus could remove skepticism that spreaders of misinformation may face on mainstream social media. From a societal perspective, the proliferation of misinformation on alt-techs is alarming as it may undermine the concepts of truth and reality among users, which might directly affect the offline world. For instance, the alleged shooter of the Buffalo shooting was suspected to be motivated by the "replacement theory," a conspiracy theory claiming that a cabal attempts to replace white Americans with non-white people that circulated on 4chan [10]. --- What we know about alt-techs Despite imminent concerns, research focusing on a better understanding of alt-techs has remained scarce. One literature stream seeks to generate a better understanding of who uses alt-techs. For example, many users migrated to Parler from Twitter during the 2020 U. S. presidential election [1], and, on top of that, the average user on Parler has strong partisan views [5]. Similarly, many users who previously violated Twitter's community guidelines and then migrated to Gab contributed to a more toxic and radical ecosystem [4]. Besides that, users of alt-techs commonly advocate conspiracy theories [5,7]. For example, prior research has shown that users frequently discuss QAnon on Parler, 4chan, and Voat [5,12]. Another literature stream studies what is shared on alt-techs and, for that purpose, compares alt-techs against mainstream social media. For example, content shared on Parler is significantly different from that shared on Twitter, especially before and after the attack on the U. S. Capitol: many Parler users expressed a less negative sentiment and lower levels of guilt compared to users on Twitter [11]. Furthermore, content on alt-techs such as 4chan, BitChute, and Gab contains more hate speech than mainstream social media [8]. For example, antisemitic hate speech was widely shared on Gab in response to the deadly Pittsburgh synagogue shooting [2]. Moreover, websites known for spreading misinformation and partisan content, such as www.thegatewaypundit.com or www.breitbart.com, are among the most frequently shared links on Parler and Gab [5,7]. Overall, this presents increased risks that make users of alt-techs vulnerable to polarization and radicalization compared to mainstream social media. --- The way forward There is an urge for policy-relevant research to better understand alt-techs as newly emerging phenomena. The knowledge available to policymakers often builds upon anecdotal evidence rather than empirical evidence, which may limit assessments that are rigorous, representative, and comprehensive. Due to differences in the user base and content dynamics, earlier findings from mainstream social media might no longer apply. Interdisciplinary research combining both computational and social science is important to unravel the underlying mechanisms of alt-techs and inform policies to counter emerging threats to society. On the one hand, computational models help social scientists to develop theories that describe the mechanism of information diffusion in social networks. On the other hand, theories from social science help computer scientists by informing the design of computational methods, so that these are effective in capturing the underlying data-generating process . Theories from social science can further explain why and where countermeasures such as de-bunking are effective . Finally, an interdisciplinary research agenda is essential to establish ethical guidelines for scientists studying sensitive personal information or even individual users of alt-techs. To address risks emerging alt-techs, we advocate impactful research along three dimensions, i. e., users, content, and society . At the user level, more effort is needed to characterize the different groups and identify behavioral factors that promote migra-tion tendencies. At the content level, advanced computational methods, e. g., from natural language processing, combined with cross-platform analyses, can provide tools to systematically analyze the proliferation of hate speech, conspiracy theories, and misinformation on alt-techs, as well as their role during specific offline events. At the societal level, there is a demand to evaluate the effect of alt-techs on the segregation, polarization, and radicalization in the offline world, as well as voter behavior and disinformation in war and conflict . There is further a need for evidence-based policies that mitigate the risks alt-techs pose to the functioning of modern societies. However, alt-techs have only little incentive to adopt stricter content moderation policies as the lack of such policies is the main reason for their popularity. Here, research such as that discussed above can inform new regulations that enforce alt-techs to establish a level of content moderation that complies with democratic law . For instance, the Network Enforcement Act in Germany and the Cybersecurity Law in China force platforms to counter misinformation and hate speech by imposing severe penalties for misconduct. In addition, public pressure can have a critical impact on the platforms' businesses. As an example, Parler's alleged role in the storming of the U.S. Capitol in January 2021 led to the removal of the Parler app from both the Google Play Store and Apple's App Store, as well as Amazon stopping to host the website, which seized Parler from operating for multiple months. Regulatory policies and public pressure may help in incentivizing platforms to foster academic research. For example, in light of the EU Digital Services Act and pressure from U. S. politics, Twitter and Meta launched programs such as the Twitter Transparency Center or the Meta Ad Library to enhance transparency and accessibility of internal data for researchers. We expect that similar efforts will also be effective for alt-techs . Along these lines, policymakers can enforce transparency to enable enhanced data access or even mandate platform providers to implement measures for content moderation and have their effectiveness evaluated through third-party bodies . Overall, this might contribute to a less radical ecosystem and mitigate the safety risk originating from alt-techs. To sum up, alt-techs have emerged from niche existence to social media ecosystems that host a large base of partisan users, extremists, and conspiracy theorists. A better understanding of alt-techs is critical as user counts reach new highs and new platforms such as Truth Social by Donald Trump emerge. Hence, more effort is needed to inform evidence-based policies and other regulations that could mitigate the risks alt-techs pose to society. --- Area Research

In recent years, several free-speech social media platforms (so-called "alt-techs") have emerged, such as Parler, Gab, and Telegram. These platforms market themselves as alternatives to mainstream social media and proclaim "free-speech" due to the absence of content moderation, which has been attracting a large base of partisan users, extremists, and supporters of conspiracy theories. In this comment, we discuss some of the threats that emerge from such social media platforms and call for more policy efforts directed at understanding and countering the risks for society.

Introduction Cannabis made its appearance in the United States Pharmacopeia in 1850, and late in the nineteenth century, cannabis regularly appeared on pharmacy shelves as an active ingredient in many medications [1]. After the first federal restrictions were placed on cannabis in 1937 with the Marihuana Tax Act, cannabis was removed from the United States Pharmacopeia in 1942. In 1951 and 1956, legal penalties for cannabis possession increased, and by 1970, cannabis was federally outlawed, with the implementation of the Controlled Substances Act. This legislation officially relegated cannabis as a Schedule I substance, classifying it as having no medicinal uses and having a high likelihood of addiction after use [2]. This scheduling of cannabis is still in effect, dramatically limiting research efforts and contributing to a stigma surrounding its use [3,4]. Despite the federal restrictions on cannabis, several states have enacted legislation to improve patient access to the plant. California began this trend in 1996 with the Compassionate Use Act, allowing physicians to recommend cannabis to suffering patients and ensuring that patients had the right to grow or consume cannabis [2]. Since then, several states and districts of the United States have followed suit, with 37 states, the District of Columbia, Guam, Puerto Rico, and the U.S. Virgin Islands having passed legislation that legalizes the use of medical cannabis [5]. The battle for cannabis legalization, though, is multifaceted. In addition to MC, some states have legalized recreational cannabis for adult use. As of September 2021, when data collection for this research concluded, 11 states and the District of Columbia have implemented legislation that effectively legalizes RC use for adults in addition to legalized MC [5]. After the November 2022 elections, 10 additional statutes have increased recreational access [6]. Arguments in favor of legalizing RC include economic and sociocultural benefits, as well as the knowledge that it reduces prescription opioid overdoses [7,8]. Conversely, those who oppose the legalization of RC are often concerned about increased motor vehicle accidents due to impaired driving, increased inadvertent access to youth, and increased cannabis use disorder rates [8]. Previous research has shown that race, political party affiliation, political views, religion, state legal status, and cannabis use strongly influence attitudes toward MC [9]. Additionally, previous qualitative research has shown differences in justification for cannabis legalization efforts based on gender [10]. Females described it as less harmful than other substances that are legal, such as alcohol and tobacco, while males reference personal freedom of choice [10]. There is also evidence to suggest that those who have used cannabis at any point in their lives are more likely to have favorable views toward it [9,11]. Much of what is known about the current climate surrounding cannabis use is specific to MC use or cannabis use in general. The present research sought to explore the demographic profiles that influence attitudes toward RC, including gender, age, ethnicity, race, level of education, marital status, number of children, the legal status of cannabis in the state of residency, employment status, political party affiliation, political views, and religion. Given the totality of this previous research, it was predicted these same factors relevant to MC attitudes in addition to gender would be pertinent to contributing toward attitudes regarding RC. --- Materials And Methods --- Scale selection This study used the Recreational and Medical Cannabis Scale, which is composed of two separate components, medical and recreational [12]. This research only utilized the Recreational Cannabis Attitudes Scale to measure participants' attitudes toward RC. RCAS consists of four Likert-scale questions with a possible composite score ranging from 4 to 20. The reliability coefficient of this scale was reported to be 0.91 [12]. Each respondent was also asked a set of demographic and lifestyle questions. In measuring the impact of state legal status on RC attitudes, respondents only provided their state of residence; the researchers coded the state's legal status manually based on current regulations to avoid potentially inaccurate data being reported by the respondents. --- Survey administration The Qualtrics-moderated survey was made available to United States respondents from February 2021 to September 2021. Respondents were recruited through social media . The primary author's learning management system was used for his undergraduate courses and distribution to the university's lifelong learning community. Any survey that was not completed in its entirety was excluded from the analysis. --- Data analysis A one-way analysis of variance was used to determine variations in RCAS scores between different demographic groups. If the assumption of homogeneity was not met, a one-way Welch ANOVA was run. Post hoc analyses were performed to determine which group within a given independent variable significantly impacted the mean RCAS score differences. For the one-way ANOVA, the post hoc analysis was completed using the Tukey-Kramer post hoc analysis; the Games-Howell post hoc analysis was used for the one-way Welch ANOVA [13,14]. Statistical significance was set at P < 0.05. --- Ethical approval The protocols of this study were reviewed and approved by the Institutional Review Board at California State University Channel Islands . Respondents acknowledged informed consent electronically before completing the survey. Due to the sensitive nature of some of the questions, raw data is not being made publicly available. --- Results After the survey administration period, 673 participants completed some portion of the survey. There were incomplete responses from 28 participants. There were also nine outliers observed from the visual inspection of boxplots, but they were included in the final analysis because they were determined to be valid responses and had a minimal impact on the final means. Therefore, the final analysis included 645 respondents . RCAS scores were not normally distributed, as assessed by the Shapiro-Wilk's test and visual inspection of the Q-Q plots. The one-way ANOVA was deemed to be appropriate, nonetheless, due to its robustness to deviations from normality [15,16]. Statistically significant variations in RCAS scores were observed between groups based on gender, state legal status, employment status, political party, political views, religion, and current/past cannabis use . n No significant variations in the RCAS scores were observed between groups based on age, ethnicity, race, education, marital status, and number of children. --- Significant results --- Gender The assumption of homogeneity of variances was met for gender, as assessed by Levene's test for equality of variances . The RCAS score was significantly higher for men than that for women, F = 3.273, and P = 0.039 . Men approved of RC 13.7-15.0) more than women . --- FIGURE 1: Mean RCAS for gender. --- RCAS, Recreational Cannabis Attitudes Scale --- State Legal Status The assumption of homogeneity of variances was met for state legal status, as assessed by Levene's test for equality of variances . The RCAS score was significantly different based on the legal status of cannabis in the participants' states of residence, F = 5.765, and P = 0.003 . The Tukey-Kramer post hoc analysis uncovered a statistically significant lower RCAS for respondents who lived in a state with MC and RC access compared to those with only MC access . Residents of a state with only MC access averaged a significantly higher RCAS compared to residents of a state with legal access to MC and RC . No other state legal status differences were statistically significant. --- FIGURE 2: Mean RCAS for legal status of state of residence. --- RCAS, Recreational Cannabis Attitudes Scale --- Employment Status The assumption of homogeneity of variances was met for the employment status data, as assessed by Levene's test for equality of variances . The RCAS score varied significantly based on employment status, F = 2.622, and P = 0.016 . However, the Tukey-Kramer post hoc analysis did not show any significance between the direct employment group comparisons. --- RCAS, Recreational Cannabis Attitudes Scale --- Political Party The assumption of homogeneity of variances was not met for political party affiliation, as assessed by Levene's test for equality of variances . The RCAS score was significantly different along party affiliation, Welch's F = 4.189, and P = 0.002 . The Games-Howell post hoc analysis revealed Republicans had statistically lower RCAS than two other groups: • Democrats' RCAS scores were higher than Republicans' by 2.16 . • Independent/no party affiliation voters' RCAS scores were higher than Republicans' by 2.29 . No other political affiliation comparisons were significant. --- FIGURE 4: Mean RCAS for political party affiliation. --- RCAS, Recreational Cannabis Attitudes Scale --- Political Views The assumption of homogeneity of variances was violated for the data on political views, as assessed by Levene's test for equality of variances . There were statistically significant differences in RCAS scores between groups, Welch's F = 17.134, and P < 0.0005 . The Games-Howell post hoc analysis revealed those who were identified as very liberal had a statistically significant higher RCAS than every other view: • 1.89 higher than slightly liberal • 2.26 higher than moderate • 3.92 higher than slightly conservative • 4.58 higher than very conservative Those identifying as slightly liberal had a mean RCAS score of 2.03 significantly higher than those identifying as slightly conservative and 2.69 significantly higher than those identifying as very conservative . Individuals who reported being moderate had a mean RCAS score of 1.66 higher than slightly conservative individuals. --- FIGURE 5: Mean RCAS for political views. --- RCAS, Recreational Cannabis Attitudes Scale --- Religion Levene's test for equality of variance with the data on religion showed the assumption of homogeneity was met . The differences in mean RCAS scores were significant, F = 7.09, and P < 0.0005 . The Tukey-Kramer post hoc analysis revealed a higher RCAS in those reporting no religious affiliation compared to those reporting being Catholic/Christian , a statistically significant difference . No other religious group differences were statistically significant. However, the one individual who identified as Hindu was not included in the post hoc analysis because the test of equality means cannot be performed with a group of less than two. --- RCAS, Recreational Cannabis Attitudes Scale --- Cannabis Use With the cannabis use data, the assumption of homogeneity of variances was met, as assessed by Levene's test for equality of variances . The mean RCAS was significantly higher for those who have used cannabis than for those who have never used it , F = 122.713, and P < 0.0005 . --- FIGURE 7: Mean RCAS for current/past users of cannabis versus those who have never used it. --- RCAS, Recreational Cannabis Attitudes Scale --- Nonsignificant results --- Age The assumption of homogeneity of variances was violated, as assessed by Levene's test for equality of variances . There were no statistically significant differences in the RCAS score between different age groups, Welch's F = 1.453, and P = 0.218. --- Ethnicity Levene's test for equality of variance showed the assumption of homogeneity was violated . The differences in mean RCAS scores between those identifying as Hispanic and non-Hispanic did not vary significantly, Welch's F = 1.097, and P = 0.318. --- Race Levene's test for equality of variance showed the assumption of homogeneity was not met for race . The mean RCAS scores did not vary significantly between self-identified racial groups, Welch's F = 0.969, and P = 0.456. --- Education Levene's test for equality of variances revealed the data met the assumption of homogeneity . The mean RCAS between different education levels did not vary significantly, F = 0.134, and P = 0.984. --- Marital Status Regarding marital status, Levene's test for equality of variance showed the assumption of homogeneity was not met . The mean RCAS scores did not vary significantly regardless of marital status, Welch's F = 1.34, and P = 0.282. --- Number of Children Levene's test for equality of variances showed the data for the number of children violated the assumption of homogeneity . The mean RCAS score did not vary significantly regardless of the number of children reported by the respondent, Welch's F = 0.839, and P = 0.474. --- Discussion Cannabis has had a volatile status in American society for nearly 100 years [1]. Once considered a valuable medical treatment, it was later vilified by government bureaucrats and politicians [2]. Although there has been a resurgence in its medical applications, there are still lingering stigmas surrounding its use [3,4]. This research helped identify demographic factors that shape attitudes toward cannabis. This current analysis revealed gender, state legal status, employment status, political party, political views, religion, and cannabis use as contributing to attitudes toward RC. A previous analysis found that state legal status, political party, political views, religion, cannabis use, and race were variables shaping attitudes toward MC [9]. While there is considerable overlap between these two lists from the previous and current research, unique factors shaped views separately toward MC and RC . Gender and employment status impacted attitudes toward RC but not MC; race contributed to attitudes toward MC but not RC. --- Demographic factor Recreational Medical Gender Here, × means factors significant for influencing attitudes toward medical cannabis and recreational cannabis. There was a significant difference in attitudes toward RC noted between genders. Men had a more favorable view of the plant than women. This is similar to previous findings, which determined males and females as having different justifications for supporting cannabis legalization [10]. Residents of states with only legal access to MC were more likely to have positive views toward RC than those who live in a state with legal access to MC and RC. This is similar to previous findings of demographic factors that shape attitudes toward MC [9]. This is thought to potentially be a result of the undesirable elements that have been associated with the legalization of RC, most notably complaints about odors, higher referral rates for cannabis use at public schools, increased hospitalizations, and more frequent traffic accidents with the driver being under the influence of cannabis [17][18][19]. While employment status was determined to have a significant variation in attitudes from the ANOVA, the post hoc analysis did not show any differences. This disagreement between the one-way ANOVA and Tukey-Kramer post hoc can occur because of the distribution of the means [13]. This demographic factor, though, is one determined to contribute to attitudes toward RC but not MC [9]. It has also been noted that cannabis use is associated with the termination of employment, lower income, and problems at work [20]. Political party affiliation was shown to be a significant factor in shaping attitudes toward RC. Republican voters had a significantly lower RCAS score than both Democratic and Independent/no party affiliation voters. Republican voters have been shown to have less favorable views toward MC in prior research, though they also moved toward a more favorable view of MC after being educated on its history, benefits, risks, and medical applications of it [21]. Political views and attitudes toward RC had an inverse relationship. The more conservative one viewed themselves, the less likely they were to have a favorable perception of RC . Similar results were observed with attitudes toward MC [9]. This is also consistent with those who historically supported the war on drugs and efforts to counteract the decriminalization of cannabis [22,23]. Identifying as Christian/Catholic was associated with a less favorable view toward RC than those who held no religious affiliation. There were other religious groups that were associated with less and more favorable views, Islam and Hinduism, respectively, but did not reach statistical significance due to the small number of respondents identifying with those religions. This result of Christians/Catholics having less favorable views of the plant is consistent with other research that found religious college students use cannabis less often, spend less time with cannabis users, and are less likely to support cannabis legalization [24]. Those who had experience with cannabis, medical or recreational, were more likely to have a favorable view of it than those who had never used cannabis. The same findings were found regarding attitudes toward MC [9]. Literature suggests that the favorable attitudes toward cannabis by those who have used it are a result of positive experiences [11]. Those who have personal familiarity with the effects of cannabis are less likely to find it harmful than other substances, such as alcohol. It is important to understand what influences attitudes toward cannabis for cannabis advocates and patients to work for reform of the current health policy. Given that the legal status of cannabis in a given geography is linked to shifting views of both RC and MC, understanding the specific factors shaping views of RC can help improve attitudes toward MC and the lingering stigma associated with the use of cannabis [3][4]9]. Further, education efforts have been shown to improve attitudes toward MC; although the same is yet to be determined for RC, the data from this and previous research will help cannabis advocates target their education efforts [9,21]. There is a notable limitation of this research. Women, those identifying as white, cannabis users, and those residing in a state with access to both MC and RC were overrepresented in the sample. The high number of full-access state residents and women is due, in part, to the authors residing in California and being affiliated with a university with a high number of female students. This research could be expanded to collect data more uniformly with the population distribution across the country for a more accurate picture of how demographic factors influence attitudes toward RC. Future research would also benefit from larger sample sizes to accurately determine the relationship between Buddhism or Hinduism and cannabis. --- Conclusions Various factors influence attitudes toward RC. The attitudes toward RC varied between genders, residents of different states, employment status, political party, political view, religion, and experience with cannabis. Understanding what factors contribute to support or opposition of the plant can help cannabis advocates to improve patient access. --- Additional Information Disclosures Human subjects: Consent was obtained or waived by all participants in this study. California State University Channel Islands issued approval IO5559. The protocols of this study were reviewed and approved by the Institutional Review Board at California State University Channel Islands . Respondents acknowledged the informed consent electronically before completing the survey. Due to the sensitive nature of some of the questions, the raw data is not being made publicly available. Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue. Conflicts of interest: In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.

While cannabis legalization in the United States has become more commonplace, differences in attitudes toward its use persist. Negative attitudes toward cannabis create barriers to care for those seeking its use for therapeutic purposes. Existing research regarding the attitudes surrounding cannabis is specific to medical cannabis (MC) use or cannabis use in general. To address this gap, the present research sought to explore the demographic factors that influence attitudes toward recreational cannabis, including gender, age, ethnicity, race, level of education, marital status, number of children, the legal status of cannabis in the state of residency, employment status, political party affiliation, political view, and religion. The Recreational Cannabis Attitudes Scale (RCAS) was used to measure participants' attitudes toward recreational cannabis. A one-way analysis of variance (ANOVA) or one-way Welch ANOVA was used to determine variations in RCAS scores between different demographic groups. Data from 645 participants indicated that gender (P = 0.039), employment status (P = 0.016), political party affiliation (P = 0.002), political view (P = 0.0005), the legal status of the state of residence (P = 0.003), religion (P = 0.0005), and experience with cannabis (P = 0.0005) had significant variations between groups regarding attitudes toward recreational cannabis. Understanding the factors that inform attitudes is critical to efforts to destigmatize cannabis use. Education about cannabis is an effective measure in reducing stigma, and paired with demographic information, advocacy efforts can be more accurately targeted.

Background There is much yet to learn about the learning health system, a variously defined concept whose hallmark characteristics are poorly articulated. Taking a complex adaptive systems view, we conceptualize the learning health system as an organization whose learning is generated by internal and external interactions and relationships [1,2]. In organizational learning theory, the idea that organizations learn from knowledges in their internal as well as external environments is well established [3]. Theories of social learning further characterize knowledge and learning as context-and relationship-dependent [4][5][6][7]. Yet surprisingly, when it comes to studying learning phenomena in healthcare, the organization's relationships with communities it serves are overlooked. In fact, community relationships represent the least researched dimension of relationship-centred healthcare [8]. In this light, we conducted a scoping review to determine what is known in extant literatures about organizational learning from or with what is arguably the most important stakeholder in healthcare: the community, which we define as one of 'the people'-such as a town or a specific healthcare patient group. Though the rhetoric of community engagement may be strong in healthcare, there is a tendency to refer to organizational learning from these interactions in the abstract [9]. The result is an obscured understanding of the motives, structures, processes and outcomes of learning by a learning health system. This leaves open the possibility to confuse 'learning from' and 'learning with' community, where 'learning from' community entails the organization extracting data about the community and 'learning with' community entails authentic partnership, power-sharing and the co-production of knowledge. Scoping reviews are well-suited to studies that aim to map concepts across interdisciplinary boundaries and review the range of evidence [10]. Following established methodology [11][12][13], our review contributes to an expanded view of organizational learning and a more comprehensive evidence base for building and operating learning health systems, drawing from health services research and other research disciplines including the organization sciences, business, and education. In this way, we shed light on relationships beyond the learning health system's organizational borders, namely with the communities the health system serves. Our findings underscore the importance of meaningful community involvement pursuant to the adage, "nothing about us, without us." --- Methods --- Search strategy Our search included an academic database search, hand searching of relevant journals, search engine queries, targeted website review and reference tracking. We based our search on three concepts: organizational learning, including social learning; the learning organization, including learning health systems; and community. Additional File 1 contains theoretical background on these concepts. Moreover, the search protocol is registered and available with the Open Science Framework . The initial academic search strategy was developed for Ovid MEDLINE® and then translated for other databases. Several preliminary, iterative searches permitted us to refine a search using a combination of subject headings and text words, and tailor to each database. Final searches with no language or date limits were conducted on March 19, 2019 in six databases. The lead author also hand searched all issues of two journals. We searched online, using numerous combinations of key terms in Advanced Google Search and Duck Duck Go. The lead author reviewed search results up to five webpages after the last hit. Websites of four organizations known for their contributions to studying learning health systems were also browsed and searched. When multiple publications were available as part of a series, the most recent or summative publication was chosen. In the case of one evidence synthesis publication found in this way, the lead author found ten additional potentially eligible citations through reference tracking. Table 1 lists our information sources. All potentially eligible citations were loaded into EndNote X9.1.1 for deduplication [14]. --- Eligibility criteria Citations had to fulfill the following criteria to be included in this study: Description of interaction or knowledge transfer between an organization and a community; Reference to organizational learning, a learning organization or learning health system; Description of the relationship between the organization and community where the community should benefit from organizational outputs. Provided that they met these criteria, citations were included based on an inclusive understanding of 'community' . Citations were excluded if they: focused on interorganizational learning or professional communities of practice; described learning at a level other than the organization; were not in English; or offered insufficient evidence, typically because they were book reviews or short commentaries. --- Study selection The lead author and a second reviewer conducted a first phase of screening using a web-based tool [15]. Over four rounds, the two reviewers screened the titles and abstracts of 524 citations with the objective to reach consistent inter-rater agreement above 80 %. Agreement in each successive round was 74 %, 81 %, 83 % and 92 %, respectively, with conflicts resolved through discussion. The lead author screened the remaining 1,423 titles and abstracts independently. Where there was insufficient or unclear information to exclude a citation, the citation was included for full-text review. In total, 98 full texts were reviewed by the two reviewers. Disagreements were resolved by consensus. Another citation was included at this stage after reviewing the reference lists of all documents. Our process is illustrated in Fig. 1. --- Data charting and collation Using an Excel chart that had been pilot tested on five included texts, the lead author extracted data for numerous variables including article characteristics ; theoretical frameworks and definitions; dimensions of learning ; community variables ; and power relations. Results were synthesized using frequencies and thematic analysis [16]. Meta-analysis was not performed. --- Search update We updated our results in August 2020. The lead author performed the same search detailed above, limiting results to items published in 2019 or 2020. The lead author performed study selection and data charting independently, applying the same eligibility criteria as above. --- Results We assessed 1,947 titles and abstracts and 99 full-text documents for eligibility during the initial search in March 2019. Of these, 38 full-text documents fulfilled our criteria. During the study update of August 2020, four texts were added [17][18][19][20]. We therefore analyzed 42 full-text documents from the research disciplines of health services, business, natural resource management, organization and management sciences, education and social services . For more detail, see Additional File 3. --- Detachment from theory Not one healthcare text drew from theory in any research discipline in defining either 'organizational learning', 'social learning', 'learning organization' or 'learning health system'. The learning health system texts were strikingly detached from theory, instead citing practicebased definitions and frameworks encompassed in National Academy of Medicine reports . --- Learning from, and learning with, community As in Fig. 2 and 10 texts focused on learning with community [23][24][25][26][27][28][29][30][31][32]. Some authors associated the quality of organizational learning with the quality of dialogue and engagement with the community [30,33,34]. They implied that although learning with community requires more effortful participatory methods [33], they are preferable to gathering information about the community without its involvement. In comparison, more than half the documents described learning from community without community partnership, typically through the collection and management of information about the community. Healthcare texts tended to focus on learning through patient data collection. Business texts tended to discuss collecting data about customers to learn their needs, preferences and spending patterns. Nine texts suggested that learning from and learning with community was possible concurrently [18,[33][34][35][36][37][38][39][40]. --- Knowledge transfer Thirty-four texts explored the nature of knowledge transfer between the community and organization. Among these, four healthcare texts [41][42][43][44] and three business texts [20,45,46] emphasized administrative or research data as drivers of organizational learning. Alternatively, 17 texts suggested that direct interaction and relationships-such as volunteering or board membership [25,47] or participation on advisory councils [26,27]-drove learning from community knowledges. The authors of six documents spoke about open, crossboundary dialogue to facilitate the flow of ideas [29,32,40,[48][49][50]. Similarly, three articles suggested that there must be an open space of discourse for community evidence to mix with organizational knowledge and help to transform organizational frames of reference [31,38,40]. Five texts incorporated the concepts of tacit and explicit knowledge [19,29,32,49,51]. One suggested that knowledge sharing is easier between different groups when they already share common tacit knowledge [51] and another described how tacit knowledge is externalized through dialogue [19]. --- Learning motives The documents revealed a wide range of motives for organizational learning from or with community. These motives were sometimes implied in the definitions authors used to describe learning. For instance, unlike definitions for 'learning organization', 'organizational learning' or 'social learning', definitions for 'learning health system' were the only ones not to consider transforming underlying organizational norms, assumptions or behaviour [26,27,33]. Other times, the motives for learning from or with community were clearly articulated . Eighteen documents, including all those focused on learning health systems, cited desire to improve the implementation of a service or policy as the purpose of learning. Fourteen documents, nearly all from business research, cited desire to enhance competitive advantage and market performance. By contrast, three of five natural resource management articles cited desire to solve societal problems [38,39,52]. Authors seemingly took for granted that organizations would achieve outcomes that matched their motives . We found insufficient detail regarding the processes linking motives to outcomes. --- Structures and processes All texts except two [49,53] contemplated structures to support organizational learning from or with community. Kass and Faden [42] implied that in healthcare, the structure for organizational learning is, in fact, the learning health system, yet did not describe this structure beyond its obligations and guiding principles. For their part, Reid and Hickman [34] argued that organizational learning depends on participatory structures and Fig. 2 Focus on 'learning from' community, 'learning with' community or both, by number of citing texts relationships. Nonaka et al. [32] described a common space or context shared between individuals and groups where knowledge is jointly created, shared and utilized; this shared context, they suggested, can be a physical or virtual space with fluid boundaries. Across all documents, the distinction between structures and processes for learning was unclear. For the most part, authors described structures like patient advisories or boards in general terms; they seemingly assumed that beneficial learning processes would follow but did not describe them. We noted that definitions given for 'learning health system' prioritized learning through research. Furthermore, informatics and digital data infrastructure were part of the definitions offered Fig. 3 Stated motives for learning from or with community, by number of texts Fig. 4 Strategies to support organizational learning from or with community, by number of citing texts by three documents only, each of which pertained to learning health systems [23,26,41]. --- Strategies The maintenance of relationships and social interaction was the most frequently cited strategy to support organizational learning from or with community, clearly mentioned in 19 texts . Ten texts recommended creating a culture or vision that openly values external knowledge . Nine texts recommended designing internal structures to support access, interpretation and sharing of information . These strategies are depicted in Fig. 4. --- Power relations Among the 10 texts that explored power relations between the organization and community [24,25,30,31,[36][37][38][39][40]56], trust and transparency were prominent themes. Equal partnership was highlighted as a foundation for community trust toward large organizations [30,36]. This sentiment was repeated by others who felt that balanced power relations were necessary for organizational learning outcomes benefitting the community. Whereas trust should be earned, they said, organizations typically enter into community relationships holding significant unearned control and privilege [37]. --- Distinct knowledge systems Four texts paid considerable attention to the position of community knowledge systems relative to organizational, professional or scientific knowledge [31,[38][39][40]. They suggested that local knowledges are habitually devalued compared with explicit, university-based scientific knowledges, thus restricting the ability of communities to participate in learning processes. --- Ethics Four healthcare texts raised ethical concerns that emerged from a focus on research, specifically the tension between patient privacy and disseminating research results [26,36,[41][42][43]. Psek et al. [26] questioned whether patient care and research could operate under joint ethical frameworks. Other ethical debates considered which organizational learning activities require ethical approval [43], how patients engage in research [42,43], and how to translate research results so they benefit patients [41,42]. --- Discussion --- Value in community If the goal is to improve the health of a population-not simply perform within the confines of organizational boundaries and compete with other healthcare organizations-it seems logical to consider learning at multiple system levels, including with community. This should give health services researchers and practitioners pause to reflect on their approaches. What a learning health system learns, how it learns, and to what effect are intimately tied with who contributes to learning. Diversity, not merely in membership but also in discourse, cultivates ideas, drives adaptation and leads to justified outcomes [57,58]. Yet the value of learning with community extends beyond simply diversifying the knowledge base of organizations. Communities extend our conceptualization of what counts as evidence. Striving toward evidence-based medicine and decision making, our reliance on purely explicit knowledge, stripped from context in the name of objectivity, provides only a partial picture. Communities represent knowledge that is discrete from organizational, scientific or other forms, providing insight into healthcare performance at organizational and whole system levels; surely, their knowledge counts as evidence, too. --- Reimagining organization-community relationship Direct relationships and knowledge-intensive interactions between people are needed for organizations to overcome barriers to learning from local knowledge [59], but it is not apparent what the organization-community learning relationship might look like. We noted throughout this review that 'relationship' means many things in the literature, from connection and interactions between different parties, to a one-way flow of data about one party to the other. With diverse meanings and models of relationship in play, it is unclear which kinds of relationships best support organizational learning. Furthermore, there is little guidance regarding how to establish learning relationships with communities. We discovered a tendency to concentrate on collecting information about community -or 'learning from' community-which risks reducing the community's position to that of merely 'being known' rather than enhancing community capacity to 'know.' It follows that 'learning with' community is a defining characteristic of the relationship in which organization and community alike are empowered to exchange their knowledges and ways of knowing, on equal footing. This line of thinking offers conceptual clarity for building and nurturing organizational learning relationships with communities. What remains is to apply these concepts to theoretical and empirical analyses of organizational learning and learning health systems. --- Ways of learning The learning health systems texts included in this review suggested a bias toward quantified, explicit and research-derived knowledge. Such a narrow orientation toward clinical research and health service data neither adequately captures the breadth and depth of relevant theory from other research traditions, nor promotes a model that gives credence to non-clinical or tacit forms of knowledge. The focus on formal research raises a number of ethical questions, including those regarding whose purposes are served ; structures of learning ; and sources of learning . Arguably just as problematic, recent study suggests that a minority of community members feel any responsibility to participate in healthcare research [60]. We speculate that the overwhelming propensity for research and evidence-based medicine could inadvertently, and perversely, steer health system planning away from models of relationship-centred care and toward a penchant for explicit research-based knowledge over tacit knowledge from other sources. Healthcare systems already field criticism for performance frameworks that overemphasize explicit measures, perceived as forsaking service for targets [61,62]. The learning health system may be best supported by a combined approach of emergent organizational learning and deliberate research, with structures in place to support both processes. --- Knowledge and power The extent to which an organization can learn with communities is influenced by an in-built power imbalance [63]. Studies of healthcare organizations have found that co-opting power through the manipulation of supposedly participatory mechanisms erodes community trust and hinders the co-production of knowledge [64][65][66]. In other words, if a partnership is founded in rhetoric more than reality, the extent to which the organization can listen and learn from communities is unclear. Learning with community would require an organization to be comfortable with the possibility that new evidence leads to the transformation of underlying assumptions and values, and the sharing of knowledge and power. --- Learning as transformation Whereas transformation may not always be preferable or lead to the best outcomes for an organization [67], there are nonetheless real reasons to contemplate transformation of the structures, processes, values and assumptions that underlie healthcare organizations, such as the need to improve the accessibility and quality of care for systemically marginalized groups. There is an opportunity for further study and reflection to gauge the value of such transformation through learning, and how this might be supported as a relationship-centred, collaborative process. --- Looking inward to learn with others Looking inward is fundamental for the organization, not only as part of a self-reflexive practice, but also to establish internal structures and processes for the interpretation, dissemination and integration of external knowledge. A single-minded mentality of knowledge acquisition, via data extraction about or from the community, may preclude relationship with that community. Managers must therefore design meaningful interactions that facilitate sharing of explicit and tacit knowledge, within and outside the organization. They must also create favourable conditions for learning that include judgement-free spaces in which to share ideas, as well as a vision that values learning in general and learning with community specifically. To respect and maintain the integrity of community knowledges, the organization would need to partner with the community to set up initial structures, processes and boundaries for collaborative learning and then be open to changes the community may suggest over time. --- Opportunity for theory-building Learning health systems literature is essentially atheoretical. Additional investigation to explore and verify associations between social relationships, structures, processes, types of knowledge and outcomes would contribute to defining characteristics of a learning health system, and possibly clarify which types of learning , driven by which motives, lead to better outcomes. --- Strengths and limitations We endeavoured to make a singular contribution to learning health systems and other organizational learning literatures by drawing from diverse research disciplines. Consistent with scoping review methodology [10,11], we did not critically appraise the reviewed texts. We therefore avoided making inferences in our results beyond what was clearly stated by the authors. Furthermore, given the current dearth of literature, we did not differentiate between community as a collective and community as represented by an individual. This distinction could be made in future research. Finally, this article does not incorporate consultation with community partners, which would be invaluable. --- Conclusions There is substantial work still left to do if we are to progress the concept of a learning health system from the abstract into practice. With this in mind, we looked across research disciplines and shone a spotlight on the communities that an organization serves. Our results encourage a new way of thinking about the learning health system: not solely as an organizational entity, but perhaps as a network of relationships. Future research can explore whether the learning health system may be, by definition, one whose culture, structures and processes afford a meaningful shared context for social interaction and knowledge co-creation with the communities it serves, with borders extending beyond the traditionally defined organization. Such would be a context wherein the organization's relationships are defined by true partnership in place of rhetoric. --- --- --- Additional file 2. Sample Search Strategy --- Additional file 3. Data Charting Authors' contributions CM conceptualized the study and led data collection, analysis and interpretation, and writing of the manuscript. WB supervised the study and contributed to writing the manuscript. All authors read and approved the final manuscript. --- Authors' information Not applicable. --- --- --- Competing interests The authors declare that they have no competing interests. ---

Background: Communities represent a highly relevant source of knowledge with regard to not only healthcare performance but also sociocultural context, yet their role in learning health systems has not been studied. Situating the learning health system as an organization, this paper explores the phenomenon of organizational learning from or with communities (defined as one of 'the people,' such as a town, a specific patient group or another group directly receiving a healthcare service). Methods: We conducted a scoping review to determine what is known about organizational learning from or with communities that the organization serves, and to contribute to a more comprehensive evidence base for building and operating learning health systems. In March 2019, we systematically searched six academic databases and grey literature, applying no date limits, for English language materials that described organizational learning in relation to knowledge transfer between an organization and a community. Numerous variables were charted in Excel and synthesized using frequencies and thematic analysis. We updated this search in August 2020. Results: In total, 42 documents were included in our analysis. We found a disproportionate emphasis on learning explicit knowledge from community rather than on tacit knowledge or learning in equal partnership with community. Our review also revealed inconsistently defined concepts, tenuously linked with their theoretical and empirical foundations. Our findings provide insight to understand the organization-community learning relationship, including motives and power differentials; types of knowledge to be learned; structures and processes for learning; and transformative learning outcomes. Conclusions: Our review makes a singular contribution to organizational learning literatures by drawing from diverse research disciplines such as health services, business and education to map what is known about learning from or with community. Broadly speaking, learning health systems literature would benefit from additional research and theory-building within a sociological paradigm so as to establish key concepts and associations to understand the nature of learning with community, as well as the practices that make it happen.

E. K. Armah-Ansah et al. --- Introduction Millions of women give birth annually and post-partum haemorrhage remains the leading cause of maternal death globally. 1 Although the global maternal mortality rate is still alarming, it has decreased by 2.1% annually since the beginning of the 21st century. 2 Sub-Saharan Africa registered > 40 million pregnancies in 2020, with an MMR of 0.63%. 3 That is about 250 000 women and translates to about 700 deaths per day. 4 Studies have revealed that one of the strategies to help achieve target 3.1 of the Sustainable Development Goals , which seeks to reduce maternal mortality to < 70 per 100 000 live births by 2030 in Madagascar, is utilisation of the services provided by skilled birth attendants . 5 , 6 SBA coverage globally has witnessed a significant increase, from 64% in 2001 to 86% in 2022. 2 Nonetheless, SSA is yet to achieve universal coverage of deliveries attended by skilled professionals as compared with other regions such as North America, Europe and Eastern Asia. 2 The utilisation of SBAs has witnessed a significant increase from 45% in 2001 to 68% among West and Central African countries and from 40% in 2001 to 72% among East and South African countries as of 2022. 2 SBAs have been jointly defined by the World Health Organization , the International Confederation of Midwives and the International Federation of Gynecology and Obstetrics as accredited health professionals who have been educated and trained to proficiency in the skills to manage uncomplicated pregnancies, childbirth and the immediate post-natal period, as well as the identification, management and referral of complications in pregnant women and newborns. 7 The presence of these accredited health professionals, including midwives, doctors and nurses, provides a safe and life-sustaining environment for both women and babies. The presence of an SBA has averted approximately 16-33% of maternal deaths. 8 Although the Madagascar healthcare system is considered among the least well-funded, the MMR has witnessed a significant reduction from 658 to 392 deaths per 100 000 live births in the last 2 decades. 9 , 10 Nevertheless, previous research and surveys have identified the primary causes of the high MMR in Madagascar as obstetric problems resulting from deliveries that took place at home and were assisted by traditional birth attendants , low health insurance subscriptions and inadequate equipment. 11 Studies have shown that all the regions in the country do not offer the recommended number of emergency obstetric and neonatal care facilities. 12 This has made maternal mortality a public health concern in SSA, and specifically Madagascar. 13 However, access to SBAs remains a serious barrier in East African countries, including Madagascar, where healthcare resources are limited. 14 However, several factors, including geographic barriers, cultural norms, socio-economic disparities and health system weaknesses, contribute to the low uptake of SBAs in Madagascar. 15 Studies on the utilisation of SBAs in Madagascar have focused on factors influencing maternal healthcare, 16 maternal health practices, 11 the use of healthcare services 17 and geographic barriers to achieving universal health coverage. 18 However, there is a paucity of studies focusing on the social determinants of the utilisation of SBAs among married and cohabiting women of Madagascar. The multilevel factor approach will help to improve our understanding of both the individual-/household-and communitylevel factors that are statistically significant in the utilisation of SBAs among married and cohabiting women of Madagascar. Thus this analysis seeks to fill the gap by assessing the social determinants of the utilisation of SBAs among married and cohabiting women of Madagascar. This current analysis provides up-todate evidence for policymakers and other program managers to design and implement programs that may be appropriate interventions to increase the utilisation of SBAs among married and cohabiting women of Madagascar. --- Methods --- Data source and study population Data from the 2021 Madagascar Demographic and Health Surveys was used in this investigation. The fifth edition of the MDHS was conducted in 2021 by the National Institute of Statistics in cooperation with the Ministry of Public Health. This survey seeks to monitor and evaluate national development strategies and programs as well as the SDGs. 18 The Demographic and Health Surveys is a 5-y nationally representative research projec t conduc ted in numerous low-and middle-income countries in Asia and Africa. It focuses on mother and child health through interviews with women of reproductive age . This information was gathered using a stratified sample approach from March to July 2021. This survey had a total sample size of 18 869 women ages 15-49 y from 20 510 households who were present the night before the survey. The MDHS sample process has been widely described in the literature. 19 The DHS follows standardised procedures such as sampling, questionnaires, data collection, cleaning, coding and analyses, allowing for comparisons across countries. Details of the methodology, instruments, pretesting of the instruments, training and recruitment of enumerators are documented in the final report of the 2021 MDHS. 19 The dataset is available for download at https://dhsprogram.com/data/dataset/ Madagascar _ Standard-DHS _ 2021.cfm?flag=1 . --- Study population and inclusion criteria For this analysis we used a file with a total weighted sample of 6997 eligible women ages 15-49 y from the selected clusters. This included all women who were either married or cohabiting and gave birth in the 5 y preceding the survey. The analysis included only women who had complete information on SBA utilisation. --- Description of variables --- Dependent variables This analysis used assistance during delivery as the main outcome variable. Assistance during delivery was obtained from the question, 'Who assisted [NAME] during delivery?' Responses to the question were categorised into 'traditional birth attendants/others' International Health and 'skilled birth attendants' . This was dichotomized and coded based on the assistance of delivery . --- Independent variables Based on SBA utilisation studies, 5 , 20 -22 the social determinant variables were categorised at the individual/household and community levels. The individual-/household-level variables included age , education , partner's education , marital status , employment status , religion , health decision-making capacity and antenatal care visits . Other individual-or household-level variables are parity , sex of household head , health insurance coverage , mass media and wealth index . The community-level variables are accessing healthcare , place of residence , community literacy level and community socio-economic status . --- Operational definition --- Mass media Listening to radio, watching television and reading newspapers and magazines were classified as media exposure. These three variables had the same response options: 'not at all', 'less than once a week' and 'at least once a week'. Based on the literature, 23 , 24 we grouped the response options into 'no,' which meant no mass media exposure , and 'yes,' which meant mass media exposure . --- Accessing healthcare The MDHS dataset asked eligible women of reproductive age a variety of questions that could pose a significant barrier to accessing healthcare. The questions included getting the money needed for treatment, the distance to a health facility, getting permission to go and not wanting to go alone. These questions had the same response options: 'a big problem' and 'not a big problem'. Based on the literature, 11 we coded these questions as accessing healthcare and coded the response options as '1' and '0' . --- Community literacy level The proportion of women who could either read and write or could not read and write at all was used to calculate the community literacy level. --- Community socio-economic status Community socio-economic status was assessed based on the employment, wealth and education of the women who resided in that community. We utilised principal component analysis to assess the number of women who were unemployed, illiterate and poor. A standardised score was established with a mean score of 0 and a standard deviation. The scores were then split into three tertiles: 1 , 2 and 3 , with tertile 1 representing a better socio-economic position and tertile 3 representing a worse socioeconomic position. 25 --- Statistical analysis The data were cleaned and analysed with Stata version 14.2 for Mac. Descriptive, bivariate and multivariate logistic regression analyses were performed to determine the individual-/household-and community-level factors of SBA among the independent variables. That is, the descriptive analysis was performed to describe the study sample. The bivariate analysis was done using Pearson's χ 2 to assess the associations between the utilisation of SBAs and all the explanatory variables of the study. Variables that showed statistical significance in the bivariate analysis were further moved to the multilevel regression model. The results for the multilevel regression analysis were presented as adjusted odds ratios with 95% confidence intervals . The multicollinearity test showed there is no evidence of collinearity among the explanatory variables . Four models were constructed, as shown in Table 2 . The first model was the empty model, which had no explanatory variable but showed the variance of the outcome variable attributable to the distribution of the primary sampling units. The second model contained only the individual-/householdlevel factors , while model 2 had only the communitylevel factors. The final model was the complete model that had both individual-/household-and community-level variables. The Stata command 'melogit' was used in fitting these models. Model comparison was also done using the log-likelihood ratio and Akaike's information criterion tests. The study sample was weighted and the survey set command 'svy' in Stata was used in the analyses to account for the survey's complex nature and the generalizability of the findings. --- Results --- Prevalence of SBAs From the analysis, about 64.4% of married and cohabiting women in Madagascar utilised SBAs . --- Bivariable analysis of social determinants in utilising SBAs among married and cohabiting women of Madagascar In Table 1 , it was found that age , education level , partner's education level , marital status , employment status , parity , mass media , ANC visits , health insurance coverage , healthcare decision-making capacity E. K. Armah-Ansah et al. and religion were statistically significant. Additionally, wealth index , sex of the household head , place of residence , accessing healthcare , community literacy level and community socio-economic status were statistically associated with SBA utilisation among married and cohabiting women of Madagascar. --- Multilevel regression analysis of social determinants of utilisation of SBAs among married and cohabiting women of Madagascar Table 2 shows the hierarchical regression analysis output on the factors associated with SBA utilisation among married and cohabiting women of Madagascar. In model 3, the results indicate that women 35-39 y of age and women with secondary or higher education levels had the highest likelihood to utilise SBAs compared with those 15-19 y of age and women with no formal education, respectively. Women whose partners had secondary or higher education levels had a higher likelihood of utilising SBAs as compared with those with no formal education. Women who are cohabiting had a higher likelihood of utilising SBAs as compared with those who were married. The odds of utilising SBAs among married and cohabiting women of Madagascar who had four or more ANC visits were higher compared with women who had less than four ANC visits . Muslim women had higher odds of utilising SBAs as compared with Christians. As compared with women with one birth, women who had four or more births had lower odds of utilising SBAs . Regarding healthcare decision-making capacity, women who do not take healthcare decisions alone were less likely to utilise SBAs compared with those who make healthcare decisions alone. The odds of utilising SBAs among married and cohabiting women of Madagascar were higher among those with female household heads compared with those who had male household heads. The odds of SBA utilisation increased with increasing wealth status. Specifically, women in the richest wealth index had a greater likelihood of SBA utilisation compared with those in the poorest wealth status. Compared with women who lived in communities with low literacy levels, those who lived in communities with high literacy levels had higher odds of utilising SBAs. --- Random effects results The result of the random effects model indicates that there was a statistically significant variation in SBA utilisation across the clusters. In the empty model, there were substantial variations in the likelihood of SBA utilisation across the clustering of the primary sampling units . The intraclass correlation value for model 0 shows that 46% of the variation in SBA utilisation was attributed to the between-cluster variations of the characteristics. The variation between clusters then decreased to 28% in model 1, which was the individual-/household-level only model. The ICC then increased to 30% in model 2, which had communitylevel factors only. In the final model , the betweencluster variation further decreased to 27%. This can be attributed to the differences in the clustering of the PSUs, which account for the variations in SBA utilisation. From the model specification analysis, model 3, which is the complete model with individual-/household-and community-level factors, had the lowest AIC compared with the other models, affirming the goodness-of-fit of the model. --- International Health --- Discussion One of the strategies to save the lives of mothers and their newborn babies depends on the use of SBAs. The complications that arise during, immediately after and beyond childbirth continue to kill many women in developing regions, including Madagascar. The best approach to saving the lives of newborns and mothers is the utilisation of SBAs during delivery and prompt diagnosis and care of obstetric problems. 5 The purpose of this analysis was to ascertain the prevalence and social determinants of SBA utilisation among married and cohabiting women in Madagascar using data from the 2021 MDHS. As found in this analysis, the prevalence of SBA utilisation among married and cohabiting women of Madagascar was 64.4%. The prevalence found in this analysis was higher than the reported prevalence of 43.7% in Kenya 1 and 23.5% in northeast Ethiopia. 26 However, the preva-lence of SBA utilisation in this analysis was lower than the 66.2% in Cameroon 5 , 66.7% in Togo 27 and 80.3% in Namibia. 28 The low prevalence of SBA utilisation among married and cohabiting women in Madagascar could be attributed to the fact that most healthcare facilities are not accessible year-round. 16 In addition, the average healthcare facility in Madagascar has less than three health professionals, with a lower concentration in rural settings. 18 In the last decade, healthcare expenditures and gross domestic product have declined from US$23 to US$18 per capita and 4.44% to 3.88%, respectively. 29 Empirical literature has revealed that about 3 in 10 women in Madagascar live > 5 km from the nearest healthcare facility, and these facilities are either understaffed or inaccessible throughout the year. 15 , 30 From the multivariate analysis, a significant association was found between SBA utilisation and social determinant variables, including age, education status, partner's education level, marital E. K. Armah-Ansah et al. status, parity, ANC visits, healthcare decision-making capacity, religion, wealth, sex of household head and community literacy level. Age is an important social determinant of SBA utilisation among married and cohabiting women in Madagascar. Specifically, the finding from this analysis indicates that utilisation of SBAs is highly concentrated among women ≥25 y of age. This finding corroborates other studies in Ghana, 31 , 32 Nepal 33 and Bangladesh. 34 Young women's fear of being stigmatised, devalued, stereotyped and embarrassed in medical settings may help to explain this finding. 21 Another reason could be that older women may have a better understanding of and be able to ascertain the benefits associated with the utilisation of SBAs during delivery. 13 Empirical studies have shown that enforcing formal education among women in SSA is a crucial indirect strategy for lowering the risk of infant delivery in areas overseen by an untrained birth attendant. 35 Consistent with studies in Ghana 36 and southeast Ethiopia, 37 this analysis revealed that education has a significant association with the likelihood of SBA utilisation among married and cohabiting women in Madagascar. This observation is consistent with other studies conducted in Kenya, 1 northeast Ethiopia, 26 Namibia 28 and Ghana. 36 This presupposes that education enhances women's exposure to information, knowledge, financial empowerment and access to media. Education influences the use of maternal health services and healthcare decisions and the seeking of appropriate medical treatment. 26 , 38 In this analysis, the partner's education status was found to predict the odds of SBA utilisation. The finding from this analysis shows that women whose partners had at least a primary level of education had higher odds of SBA utilisation than those with no formal education. This finding supports previous research conducted in Kenya 1 and Ghana. 39 , 40 The possible explanation is that women's partners who have at least a primary education may be well informed about pregnancy risk symptoms and the need for an SBA during childbirth. 39 This implies that educated partners may have good health-seeking behaviour. Also, due to the patriarchal structure of most societies in SSA, males often make health-related decisions and pay for bills, travel and the diets of their partners. 1 , 39 Studies conducted in Ghana, 41 Tanzania 42 and Papua New Guinea 43 have revealed the relationship between marital status and SBA utilisation. In this analysis, we found that women in Madagascar who were cohabiting had higher odds of utilising SBA than those who were married. Women in SSA mostly depend on their partners to decide whether to use an SBA. 44 Conversely, this observation contradicts other findings from Papua New Guinea. 43 The possible reason could be that cohabiting women in Madagascar may have greater autonomy and healthcare decision-making capacity. This discovery necessitates a qualitative analysis to understand more about how marriage affects the utilisation of SBA among Madagascar's women. Another critical finding in this analysis was the relationship between parity and the utilisation of SBAs. In the present analysis, women of Madagascar with four or more births had lower odds of utilising SBAs. This analysis revealed that SBA utilisation decreased with increasing parity. This finding is supported by studies conducted in Cameroon, 5 Ghana, 31 Papua New Guinea, 43 Nigeria 45 , 46 and Ethiopia. 47 , 48 This presumably reflects the notion that women with greater parity are less likely to give birth in a medical facility due to their prior maternity experiences or because having a big family implies having less time and money to invest in maternal healthcare. 45 Another possible reason could be that women who have one child may choose SBAs because they are more concerned about delivery complications. 46 Another social determinant variable in this analysis that was significantly associated with married and cohabiting women of Madagascar utilising SBAs was ANC visits. This observation is consistent with postulations in other studies. 1 , 26 , 49 , 50 Worldwide, ANC visits are intended to provide preparation for childbirth for all women, and current WHO recommendations require women to undergo a minimum of four ANC visits because of the benefits they have been shown to provide in preserving both the mother's and baby's health. 38 Specifically, high utilisation of SBAs was found among women who had four or more ANC visits compared with those who had fewer than four ANC visits. This is consistent with earlier research conducted in Kenya, 1 Sierra Leone, 38 Ghana 39 , 51 and Ethiopia. 37 , 47 , 52 The possible reason could be that as the number of ANC visits increases, women are exposed to health education and information on the advantages of having an SBA present during labour. 13 , 37 , 47 Studies have indicated that women and religious leaders incorporate their spiritual beliefs and practices into the process of childbirth. 5 These customs, observed before and during childbirth, aim to promote a safe delivery. 53 These religious leaders encompass various elements such as anointing oil, blessed water, a blessed white cloth, blessed sand, the Bible and the rosary. 53 , 54 This illustrates the impact of religion on women's use of SBAs. Consistent with previous studies in Cameroon 5 and Ghana, 55 the present analysis shows that the religion of married and cohabiting women in Madagascar is strongly associated with the utilisation of SBAs. We found that married and cohabiting women in Madagascar who are Muslims were more likely to use SBAs. In contrast, some studies conducted in Cameroon 5 and Ghana 55 have revealed that Muslim women have lower odds of SBA utilisation. Strong family and community support could be a possible explanation for the higher odds of SBA utilisation in Madagascar. This means that partners of married and cohabiting women in Madagascar who are Muslims may advocate for and ensure that their wives receive appropriate healthcare during pregnancy and childbirth. Several studies in Nigeria, 45 Ethiopia, 56 Ghana 57 and Bangladesh 49 , 58 have further revealed that wealth is significantly associated with SBA utilisation. In this analysis, the utilisation of SBA increased with increasing wealth status. Specifically, married and cohabiting women of Madagascar in the richest wealth index were more likely to use SBA than women in the poorest wealth index. This finding is supported by other studies in Cameroon, 5 Ethiopia, 56 Ghana 57 and Bangladesh. 49 , 58 This could mean that women in the richest wealth index may be better able to afford care-seeking expenses like transportation, drugs and any other fees involved with care, and they may also be more able to access information about the advantages of having an SBA present during delivery. 13 , 28 , 49 In this analysis, we found that female household heads had higher odds of utilising SBA compared with male household heads. Consistent findings have been reported in previous studies in Liberia, 59 Ethiopia, 60 Guinea 61 and Indonesia. 62 This could International Health be attributable to the fact that female household heads have control over the finances of the home and have the means to utilise SBA. Another plausible reason could be that female heads of households understand and respect the need to use maternal healthcare in an appropriate manner to safeguard both their own lives and the lives of their unborn children. 63 With regards to healthcare decision-making capacity, women who do not have full capacity to make healthcare decisions had lower odds of utilising SBAs. The finding of the present analysis is consistent with studies in western Ethiopia 47 and Guinea. 61 However, the finding contradicts the findings of other studies in southeast Ethiopia, 37 southwest Ethiopia 56 and northern Nigeria, 64 which found that women who decide jointly with their partners on healthcare decisions are more likely to utilise SBAs. The conclusion may be a result of the high cost of accessing prenatal care and/or women's limited ability to influence family decisions coupled with low financial status. 36 , 65 Finally, the analysis further found that community literacy level was significantly associated with married and cohabiting women in Madagascar utilising SBAs. The odds of utilising SBAs increased with an increase in the level of community literacy among married and cohabiting women in Madagascar. Married and cohabiting women in Madagascar who lived in communities with high literacy levels were more likely to use SBAs. This finding is consistent with a similar study in Namibia. 28 The possible conclusion could be that women who live in communities with high community literacy levels may subscribe to and disseminate accurate information about maternal healthcare services that influences positive utilisation of SBAs. 21 --- Strengths and limitations of the analysis The strength of this analysis is the rigorous analytical and statistical approach used to increase the dependability of our findings by estimating the cluster effect on SBA utilisation among married and cohabiting women in Madagascar. Furthermore, the use of current nationally representative data from the MDHS and the representativeness of the sampling approach improve the generalizability of the analysis, and the large sample size contributes to the rigor of the analysis. There are a few drawbacks to this analysis. First, the research sample was confined to women of reproductive age who are either married or cohabiting. Moreover, the cross-sectional character of the MDHS and the cause-effect relationship could not be determined. Furthermore, recollection bias may affect survey participants' self-reported data, which may have recall bias and could lead to over-or underreporting. --- Conclusions The current analysis revealed low SBA utilisation among married and cohabiting women in Madagascar. The analysis points to the fact that understaffing and inaccessibility of health facilities remain major contributors to the low utilisation of SBAs. The analysis revealed that women's age, education status, partner's education level, marital status, ANC visits, parity, healthcare decision-making capacity, religion, wealth, sex of household head and community literacy level were the social determinants significantly associated with SBA utilisation among married and co-habiting women in Madagascar. The findings of this analysis call on the government and stakeholders in Madagascar to consider and implement programs and strategies that will focus on disadvantaged groups and empower women. These programs would make it easier for people to utilise SBAs, which will eventually lead to the achievement of SDG 3 by 2030 and universal healthcare. These programs include providing free maternal healthcare services to all pregnant women. This strategy will not only improve the prevalence of SBA utilisation, but will also improve the health of women and their unborn children. Another strategy that stakeholders in Madagascar can implement to improve the prevalence and utilisation of SBAs is intensifying health education programs that target women and their partners with no formal education. --- Authors' contributions: EKA-A and BB conceived the study and conducted the analysis. JD and EKA-A conducted the literature review. EKA-A and BB drafted the manuscript and reviewed it to ensure that it was appropriate for submission. JD, EB, BOA and EKA provided technical support and critically reviewed the manuscript for intellectual content. BB had the final responsibility to submit the manuscript for publication. All the authors read and approved the final manuscript. --- Competing interests: The authors declare that there is no conflict of interests. Ethical approval: Since the data for this analysis is secondary and in the public domain, ethics approval was not necessary. Further information about the Madagascar DHS data usage and ethical standards is available at https://dhsprogram.com/methodology/Protectingthe-Privacy-of-DHS-Survey-Respondents.cfm . Data availability: Data are available from https://dhsprogram.com/ data/dataset/Madagascar _ Standard-DHS _ 2021.cfm?flag=1 .

Background: Maternal mortality ratio (MMR) in Madagascar is 392 deaths per 100 000 live births, and this is a major public health concern. One of the strategies for reducing MMR and achieving target 3.1 of the Sustainable Development Goals (i.e. reducing the global MMR below 70 per 100 000 live births) is the utilisation of skilled birth attendants (SBAs). This analysis examined the prevalence and social determinants of SBA utilisation among married and cohabiting women of Madagascar. Methods: Data from the 2021 Madagascar Demographic and Health Surveys was analysed on a weighted sample of 6997 married and cohabiting women. A multilevel regression was carried out to determine the social determinants of utilising SBAs. The results are presented as odds ratios (ORs) associated with 95% confidence intervals (CIs) and a p-value < 0.05 to determine the significant associations.The prevalence of SBAs among married and cohabiting women of Madagascar was 64.4% (95% CI 0.62 to 0.68). In model 3 of the multilevel regression, women 35-39 y of age (adjusted OR [aOR] 1.86 [95% CI 1.30 to 2.60]), women with secondary/higher education (aOR 1.67 [95% CI 1.32 to 2.10]), women whose partners had secondary/higher education (aOR 1.58 [95% CI 1.25 to 1.99]), cohabiting women (aOR 1.33 [95% CI 1.07 to 1.65]), women who had four or more antenatal care visits (aOR 2.05 [95% CI 1.79 to 2.35]), female household head (aOR 1.44 [95% CI 1.06 to 1.95]), Muslims (aOR 1.58 [95% CI 0.71 to 3.53]), those of the richest wealth index (aOR 4.32 [95% CI 2.93 to 6.36]) and women who lived in communities with high literacy levels (aOR 2.17 [95% CI 1.57 to 3.00]) had higher odds of utilisation of SBA.This current analysis revealed low SBA utilisation among married and cohabiting women in Madagascar. The analysis points to the fact that understaffing and inaccessibility of health facilities remain major contributors to the low utilisation of SBAs. The findings call on the government and stakeholders in Madagascar to consider implementing programs that will empower women and focus on disadvantaged groups. These programs could include providing free maternal healthcare services to all pregnant women and intensifying health education programs that target women and their partners with no formal education.

Introduction Sexuality education is an age-appropriate and culturally relevant approach to teaching about sex and relationships by providing scientifically accurate, realistic, nonjudgmental information . Comprehensive Sexuality Education is a curriculum-based process of teaching and learning about the cognitive, emotional, physical, and social aspects of sexuality . It contributes to 'a fair and compassionate society by empowering the individuals and communities, promoting critical thinking skills, and strengthening young people's citizenship' . CSE is the curriculum-based learning process of age-appropriate, culturally sensitive, and scientific information on physical, emotional, and social aspects of sexuality which can be provided in-school and out-ofschool Nepal government has also shown its commitment to include CSE in its programs in different sectors, more specifically in the education and health. Nepal is a signatory to and actively supports the International Conference on Population and Development . It supports the ICPD beyond the 2014 review document which emphasized the need for a CSE. The programs of CSE are started from 2007 in Nepal and included in various levels of curriculum. National education and health-related policies include the agenda and several organizations have come ahead in supporting the government in the implementation of the agenda, however, teachers' and students' hesitation to teach and learn the contents of CSE are still a barrier in its effective implementation. Moreover, the stigma associated to sexuality contents also add further problem in the process of learning and receiving the services. The sexual and reproductive health and rights of young people in Nepal are particularly vulnerable to a variety of issues. This article sought to highlight the critical role that CSE plays in addressing the significant SRHR situation of young people in Nepal. --- Methods The study was a desk review in which the literature on the SRHR status of Nepalese young people had been found and compiled using narrative review methodologies. This was achieved by going over past arguments, analyzing earlier surveys on the state of young people in Nepal , and connecting those to CSE. Information for this study was gathered by conducting manual literature searches as well as using internet tools like Google Scholar, Pubmed, and Nepjol. The search was conducted using the keywords specified in the abstract. --- Result Youths and adolescents are seen as being at risk for sexual and reproductive health. Adolescents should openly address their sexual and reproductive health concerns or challenges with older family members, teachers, or relatives with whom they feel comfortable in order to ensure a good sexual and reproductive health status. However, because of social stigma against discussing such topics, this doesn't typically occur in Nepal. The Government of Nepal has identified adolescents and youth as the most vulnerable, underserved groups who are at risk for a wide range of sexual and reproductive health problems, from early and unintended pregnancies to sexually transmitted infections including HIV. In 2021, 4% of all people who developed HIV infection were 14 years old and 3.86% were 15-24 years old . --- / 99 Early --- Discussion The results of many surveys indicate that Nepal's youth urgently require educational intervention. The government of Nepal appears to be paying less attention to the SRHR status of young people, which has put them at risk. The main problems among them include early marriage, unmet family planning needs, and early pregnancies. Strategic investments in adolescent health and well-being yield economic and social benefits 10 times their costs, save 12.5 million lives, prevent more than 30 million unwanted pregnancies, and prevent widespread disability . Therefore, there is a need for continued collaborative efforts for common advocacy for CSE to address young people's SRHR. The most effective effort seems to be offering CSE in schools. Many countries worldwide have increasingly recognized the importance of providing young people with the knowledge and skills to make responsible choices in their lives, especially when they are more exposed to sexual content via the Internet and other media. The 2030 Agenda and its global Sustainable Development Goals call for action to leave no one behind, and to achieve human rights and gender equality for all. Mobilizing political commitment to achieve the goals of education, gender equality, health and wellbeing provides an important opportunity to bring CSE to children and youth everywhere. According to UNESCO most countries have prioritized the concept of CSE at the national level and are working to strengthen its implementation. As the world community has included many programs with the aim of aligning adolescents and youth with the Sustainable Development Goals, the Government of Nepal has also tried to improve the CSE knowledge of adolescents and youth but not enough. Due to their numerous psycho-social and physical changes, adolescents and young people need to receive sexuality education. However, a majority of families in Nepal are hesitant to offer this instruction at home. Discussing sex and sexuality is frowned upon here. As a result, a systematic approach to sexuality education is needed to be delivered in a school setting . Information is critical in guiding and altering young people's behavior, much as school instructors are crucial in imparting sexuality and reproductive health information . In a classroom context, CSE can be taught throughout time at the right age and in a developmental order, building new knowledge on previous knowledge . --- Ministry of Health and Population (MoHP Due to sexual transition, peer pressure and lack of life skills, many adolescents and young people in Nepal are at various risks. They need to be prepared for a safe, productive and fulfilling life. CSE empowers adolescents and young adults to understand themselves and others, build healthy relationships, make informed decisions about sexual and reproductive health and sexuality. It is wise for adolescents and young people to acquire this education at the appropriate age and time. Most young people lack the knowledge necessary to make responsible decisions, which puts them at risk of sexually transmitted infections and unwanted pregnancies. Adolescents are particularly confused by their sexual and reproductive status and changes. Due to the lack of appropriate knowledge and skills, there is a possibility of falling into various difficulties if one cannot accept the sexual change positively. There is also a need for competence in the social issues of sexuality. In order to build a successful future by solving problems of this nature and making responsible decisions, adolescents need competence in various areas of sexual education. Adolescents typically deal with one-of-a-kind issues. Their perspective of something is the root of these issues. If they receive prompt and efficient instruction from parents, teachers, or elders, these issues can be resolved. They can talk with their parents about these difficulties more effectively when there is a positive home environment. Having supportive friends around might also help them manage these issues. Every adolescent should talk to their parents or other trustworthy adults about their difficulties. This can only happen after creation of an enabling environment through sexuality education. By covering up natural and normal bodily processes, teenagers become suspicious of life itself. Some teenagers have failed to / 101 fulfill their responsibilities in relationships. They sometime feel failure to plan for major life events. They feel the lack of skills to connect sex with life and society. Therefore, adolescents need sexuality education. --- Conclusion Due to a shortage of sexuality education that emphasizes life skills, adolescents in Nepal face a variety of difficulties. Many young people, especially adolescents, are given confused and contradictory information, which leads to a low SRHR status. Hence, there is a growing demand for reliable information that prepares them for a safe, productive and fulfilling life. CSE addresses this demand by empowering adolescents and youths to make informed decisions about relationships and sexuality. It helps them make life-changing decisions about their sexual and reproductive health. CSE is not only the study of human physiology and sexual behavior, but is a teaching and learning process related to the development of knowledge, perception and skills of various socio-cultural and life skillsrelated topics. The government of Nepal has pledged to implement measures to enhance young people's SRHR status by signing a number of SRHR-related conventions. As a result, Nepal should utilize CSE to solve SRHR-related concerns and enhance the SRHR status of young people. --- Declaration Conflicts of Interest None. --- Funding None. --- Contributors BA, who also engaged in communication with the publication procedures, came up with the idea and wrote the essay. The article was edited by BA, AA and RKP before being published. Each author gives their approval for the publication.

This article reflects the narrative evaluation in the domain of young people's SRHR status and need for CSE. Young people in Nepal are particularly at risk for early marriage and childbearing, as well as other SRHR-related issues. CSE is recommended as a suitable technique to deal with these concerns affecting young people. CSE is a rights-based approach to teaching the cognitive, emotional, physical, and social aspects of sexuality. It helps young adolescents develop the knowledge, ideas, and beliefs they need to choose and embrace their sexuality in intimate relationships on a physical and emotional level. Its promotion is proposed as a solution to assist Nepalese young in enhancing their SRHR status.

Introduction There is increasing evidence that sibling bullying, a form of aggression which is carried out repeatedly, adversely affects the mental health of children. We review how and why sibling relationships may influence child development and mental health, what sibling bullying is and how it relates to rivalry and aggression between siblings, and how prevalent sibling bullying is and what family factors are associated with it. Furthermore, the association between sibling and peer bullying will be reviewed and the impact of sibling bullying on mental health outcomes explored. Some general programmes dealing with sibling relationships in families are identified and future directions of research and clinical interventions are discussed. It is argued that sibling bullying is an important factor to consider in children's development in general, and in particular, by health professionals involved with mental health. --- Siblings Worldwide, most children have siblings. In the UK over 85% of adolescents have at least one sibling 1 ; in other regions it is normal for families to have several siblings 2 , with the largest offspring size found in sub-Saharan Africa 3 . There are exceptions, such as the China "One Child Policy " which restricted the number of children that urban couples could have to one between 1979 and 2013 when it was relaxed 4 . Sibling relationships are usually the most enduring relationships in a lifetime [5][6][7] and by middle childhood children spend more time interacting with siblings than with parents 8 . Siblings play an important role in each other's lives as companions, teachers, and caregivers 9 and can significantly contribute to each other's development and adjustment. Siblings may have either a direct or indirect effect on each other's development 10 . Bullying in the Family: Sibling Bullying Positive sibling ties and interactions can facilitate the acquisition of skills that are important in cognitive development 11,12 , provide emotional support 13 , and buffer siblings from adverse life events 14 , including marital conflicts 15 or poor peer relationships 16 . Furthermore, sibling relationships are independently associated with later adolescent adjustment and well-being, even after controlling for parental and peer influences [17][18][19] . --- Sibling rivalry Sibling relationships however are not always harmonious and supportive. Severe sibling jealousy and rivalry have been documented since ancient times, most notably through the story of Cain and Abel . The term sibling rivalry was coined by child psychiatrist and psychoanalyst D. M. Levy to refer to jealousy of the mother's love 20 . He used an experimental projective approach with children aged 2-13, where clay dolls were used to represent a baby at the mother's breast and an older child standing next to them. Children were asked: "And then the brother/sister sees the new baby at the mother's breast. He/she never saw him before. What does he do?" The young children's responses often involved attacking the baby doll i.e. destroying, biting, tearing or crushing it with his feet 21 . This was often justified as being a prohibitive parent: "The baby is bad" or "because she was bad. She wanted to hit the baby". "We don't need two babies in one house" [p. 361]. Levy's observations of various populations and tribal societies led him to conclude that sibling rivalry is a universal situation among people regardless of their various cultural forms, arising directly out of biologic behaviour. He concluded that sibling rivalry, the aggressive response to the new baby, is so typical that it is a common feature of family life 22 . Sibling rivalry stemming from a new-born baby may be conceptualised within Miller's 23 Frustration-Aggression hypothesis, that postulates that the occurrence of aggression always presupposes the existence of frustration. It is easy to see that a first child may be frustrated by the arrival of a new child in the family. Assuming equal care, the 100% attention that the first child once had will be at least halved, while a second child will only ever have had 50% of attention. For a new arrival this will be further reduced to 33% and so on. In the triangle of sibling rivalry, which comprises the sibling, their beloved parent, and their rival sibling, this is shown in the form of initial frustration and later jealousy, a complex social emotion. Indeed, it has been found that jealousy and related behaviour is linked to persisting poorer sibling relationship quality 24 . Some cultural variations in the nature and dynamics of sibling relationships have been observed, most notably according to individualistic or collectivistic norms. In collectivistic societies, which place greater focus on group rather than individual goals, siblings spend more time together, and have more hierarchical relationships, with older siblings commanding greater respect, but also taking on the responsibility of providing care for younger siblings 25,26 . Accordingly, cross-cultural comparisons have found sibling relationships show greater support, companionship, intimacy, and satisfaction in collectivistic or family orientated societies compared to individualistic ones 25,[27][28][29] . --- Sibling Aggression or Sibling Bullying The lack of an accepted definition, as well as the use of differing terms, such as aggression, violence, abuse, bullying, or rivalry, has been a barrier to research on sibling aggression 1,7,30 . Where possible we focus on sibling bullying, which is a form of aggression between siblings that involves direct or indirect acts, which are performed intentionally, over time, and involve an imbalance of power . It thus excludes infrequent or singular acts of aggression, which may be better described as sibling rivalry, and incidents of extreme violence or sexual abuse, which may be criminal in their nature 7 . --- BOX 1. Definition of sibling bullying Sibling bullying may be defined as "any unwanted aggressive behaviour by a sibling that involves an observed or perceived power imbalance and is repeated multiple times or is highly likely to be repeated; bullying may inflict harm or distress on the targeted sibling including physical, psychological, or social harm. It encompasses two modes of bullying as well as four types of bullying ." Adapted from CDC uniform definition of peer bullying 31 Considering the similarities between children's relationships with siblings and with peers 32,33 , bullying seems to be the most appropriate term to use for several reasons: Firstly, bullying occurs in settings where individuals do not have a say in which group they want to be in. This is the situation for both children in school classrooms and those at home with their siblings. In an analogy to experimental studies on social defeat in animals, this may be considered as siblings being "caged" together in the same space 34 , often in the absence of an adult, which leads to familiarity that can breed contempt 35 . Secondly, siblings differ by sex and age, thus are rarely equal in terms of size or physical or mental strength. Older siblings may use their size advantage to bully a younger sibling, and males may have more physical strength than a female sibling; therefore there is an imbalance of real or perceived power. Thirdly, repeated direct or indirect acts of aggression are a hallmark of bullying definitions 36 . As with peers, repeated acts of aggression may not be seen simply as a reaction to frustration, but as an effort to establish a social hierarchy and access to resources , a characteristic of peer bullying traced across societies and time 37 . It has been shown that conditions that foster higher density and are conducive to hierarchies, i.e. unegalitarian Bullying in the Family: Sibling Bullying conditions that reward getting ahead of others, increase bullying in classrooms 38,39 , at home 33 or even in nations 40 , and are associated with greater stability of bullying over time 41 . Considering the conditions , repeated use of aggression, an imbalance of power, and the effort to seek access to resources, then this aggression between siblings may be considered as sibling bullying . --- BOX 2. Search Strategy To identify research on the nature and correlates of sibling bullying, systematic searches were carried out using the PsycINFO, PubMED, Web of Knowledge, and Google Scholar databases, using the keyword "sibling" in combination with the terms "bullying", "aggression", "rivalry", "abuse", "violence", and conflict". Searches were limited to peerreviewed studies, published since 1990. Additional hand searches were carried out by checking article reference lists. All search results were screened for relevancy to sibling bullying, yielding a total of 19 studies. An additional search focusing on intervention programmes was performed using the PsycINFO, PubMED, ScienceDirect, and Google Scholar databases, with the terms "sibling intervention" or "improving sibling relationship" in conjunction with the words "bullying", "aggression", "rivalry", "abuse", "violence", and conflict". After screening for relevancy, the search returned 7 results. --- Prevalence There is surprising paucity of studies on sibling bullying, or even sibling aggression more generally. Systematic searches of online databases show that in the last 25 years, only a small number of studies have predominately focused on bullying behaviour among siblings . All of these studies fit with the definition provided above, in that they consider physical , verbal , or relational acts of aggression between siblings, that are used repeatedly over time. Most studies rely on child self-reports, and although some also use parent reports, particularly among younger children, these are likely to be underestimates, as parents are probably unaware of every physical fight between siblings, and may know even less about indirect bullying, as this is not always disclosed to the parents, just as with peer bullying 42 . Table 1 here As Table 1 shows, studies vary greatly in their approach to measuring sibling bullying, using differing instruments and cut-off points , yet despite these differences, all agree that sibling bullying is widespread, and experienced by a large proportion of children and adolescents. When considering any form of involvement, prevalence rates vary from around 15-50% for victimisation by siblings, and 10-40% for perpetrating sibling bullying. This prevalence rate is higher than found for peer bullying, where typically between 5-20% of children are victimised, and 2-20% bully others 50 . Direct comparisons show a higher frequency of sibling than peer bullying 33,35,48 . A unique aspect of sibling bullying is the high prevalence of bully-victims, children who are both victimised by, but also bully their siblings 1,33,35 . Most children involved in sibling bullying fall into this category, contrasting distinctly with peer bullying, where children tend to adopt stable victim or bully roles 51,52 , and only a small minority are identified as peer bully-victims 50 . The ability to change between roles in sibling relationships may be evidence of a more fluid power dynamic, whereby siblings use their familiarity to gain an advantage over each other, thus they are less likely to become confined to being only victims or only bullies. The behaviours involved in sibling bullying closely resemble findings on peer bullying, with most children reporting a range of physical, verbal, or relational behaviours 1,35,43,44 . There are also similarities regarding age and sex variations. As with peer bullying 53,54 , incidence of sibling victimisation and bullying perpetration shows some decline with age, particularly in physical aggression 1,6,49 , and males are more likely to perpetrate acts of sibling aggression, although there are no clear gender differences in regards to victimisation 1,35,44,55 . While these characteristics show individual associations, the sex and age composition of the sibling relationship can have a significant impact on overall rates, with more sibling bullying reported in male-male 30 or older male-younger female sibling dyads 55 . Overall, the evidence obtained from both national surveys and opportunistic studies in the USA, UK, Italy, Israel and Australia suggest that sibling bullying is the most frequent form of maltreatment, more so than by parents, adult strangers, or peers 5,45,56,57 . As a result, relationships with siblings are likely to be the most aggressive relationships that children will ever encounter during their childhood. --- Family Factors associated with Sibling Bullying As the primary environment in which sibling's interact, household and family characteristics may have some influence on rates of aggression. A handful of studies have explored these associations. Household or family characteristics can be broadly grouped into three categories: structural factors, including household composition, number, age and sex of siblings; socioeconomic factors, including household income, parental education and occupation; and adult/caretaker behaviour, such as child maltreatment and parenting behaviour. --- Table 2 here The composition of the household has some impact on rates of sibling victimisation and bullying perpetration. Consistent with self-reports, overall rates of sibling bullying are higher in households with male siblings 1,6,30 , and younger children tend to experience more victimisation, often at the hand of an older sibling 47,55 . Supporting Levy's hypothesis 20 , some found sibling bullying occurred more often among siblings who were close in age 30 . Furthermore, as the number of children within a household increases, so does the rate of sibling bullying 1,47 . There is no evidence to show that living in a single-parent or step family increases the risk of sibling bullying with the exception of one study 62 . Few studies examined the association between socioeconomic characteristics and sibling bullying, despite evidence showing greater rates of any intra-familial violence within low socioeconomic families 6,57 . At present, findings are mixed. Although higher rates of aggression have been found in low income families 6 or those experiencing financial stress 1 , higher parental education also predicts more sibling bullying 1,30,62 . Similar to findings on peer victimisation 63 , sibling bullying is frequent in families of all social strata . Cultural variations have also been found through comparisons between native born and immigrant US children. Immigrant siblings spent more time together, were more intimate, and provided greater social support 59,64,65 , while native born US siblings experienced greater conflict and bullying 6,7,30 . It is not yet clear how cultural differences interact with other societal characteristics, such as socioeconomic status or parenting behaviour, to affect sibling relationships 26 . The most consistent associations have been found with adult/caretaker behaviour. Child maltreatment or adult-to-child violence significantly increases the risk of sibling victimisation and bullying perpetration 6,43,45,47 . Similarly, a lack of parental warmth 59 , harsh parenting 1 , and low supervision 59,62 have also been linked with more bullying between siblings. Studies on general sibling conflict 66,67 , as well as on sibling bullying 59 , identified a link with differential parental treatment of siblings, suggesting that sibling bullying may be motivated by inequality and a desire to improve one's status, thus mimicking the motivations that underlie bullying at school 37 . --- Sibling bullyingis there a link to peer bullying? There are reasons to suggest that interactions within the family, including sibling relationships, may generalise to children's interactions with peers in other contexts, such as school 2 . Social learning theory suggests that children learn particular behaviours in relationships with their parents and siblings and that these behaviours generalise to their interactions with peers and friends 68,69 . Attachment theory proposes that children's relationships with peers and siblings are influenced by internal working models of relationships which are carried forward from their earliest relationships with attachment figures 70,71 . Furthermore, there is evidence that children's enduring characteristics, such as temperament, may elicit similar responses from different relationship partners 72 . While each of these theories suggests some transference between children's relationships with siblings, friends and peers; the proposed mechanisms linking relationships differ. Alternatively, there are also reasons to expect few associations between children's familial and extra-familial relationships. Firstly, siblings growing up in the same context only share, on average, 50% of their segregating genes, and they are subject to non-shared experiences in the family and to differential treatment by their parents 73 . Secondly, children in peer relationships are from different families and may have different temperamental characteristics, interests and talents. Their previous experiences in relationships vary and their beliefs and expectations about how to behave in relationships are likely to differ also. Finally, societal norms and culturally held beliefs about how one should behave in particular relationships, such as sibling relationships compared to best friendships, encourage differences and may decrease the likelihood of associations between them 74 . Indeed, the pattern of findings has been inconsistent with no simple 'carry-over' from sibling relationships to friendships 74 . However, is this also true for adverse sibling experiences such as sibling bullying? Studies that have investigated the relationship between sibling and peer bullying are shown in table 3. Table 3 here Two studies just looked at sibling victimisation 2, 48 while the others looked at both bullying perpetration and victimisation among siblings 1,35,55 . Firstly, all findings support a significant carry over from sibling bullying to involvement in peer bullying. Secondly, where studied, the findings indicate a homotypic carry over across contexts 1 and this may apply more so for boys than girls 55 . However, as reported above, most children involved in sibling bullying get victimised and retaliate . The two studies that provided statistics investigated just sibling victimisation and perpetration. Not surprisingly both were highly correlated. Clearly more studies are needed but the evidence supports the theory that adverse sibling relationships transfer to similar experiences in peer relationships. All studies were cross-sectional investigations and do not allow for causal interpretation. Longitudinal studies are necessary to conclude that sibling bullying is a precursor of peer bullying. One innovative study combined direct observation of sibling directed antisocial behaviour in the family's homes at 3 and 6 years and interaction of unfamiliar peers in an experimental setting 75 . Antisocial behaviour between siblings was observed at home when the children were 3 and 6 years old, and at 6 years they were invited to the laboratory where they were paired with two other unfamiliar children for a triadic play situation. Those young children who showed sustained high antisocial behaviour towards their siblings were more likely to bully or refuse to share or interact with unfamiliar peers. Thus, at least in young children, experiences with siblings are predictive of aggressive behaviour towards unfamiliar peers. --- Sibling bullying and emotional and behaviour problems Considering that sibling bullying is widespread, the crucial question is whether it has any adverse emotional or psychiatric outcomes or is just a phenomenon without consequences. As shown in table 4, there are currently 5 cross-sectional studies and only one prospective study that specifically investigated the relationship between sibling bullying and emotional problems. Four of the five studies found highly raised depression and loneliness scores 35 and more behaviour problems within the clinical range using the Strengths and Difficulties Questionnaire 2,44 , a reliable and valid screening questionnaire for psychiatric problems in childhood and adolescence 76,77 or increased mental distress 78 . Three of the studies also reported that a) the associations with behaviour problems were stronger the more severe the sibling victimisation was, i.e. involved both verbal and physical bullying 2 or mental distress was increased the more severe the physical assault 78 ; or b) the odds of behaviour problems were increased up to 14 times if the child was bullied both at home and by peers at school 2,44 . The effects of sibling and peer victimisation were found to be additive rather than interactive in the US survey 78 . Where investigated, it appears that those who were both victims and bullies were at higher risk of behaviour problems than those who were only victimised. No increased risk was found for bullies 2 . In contrast, the UK survey of maltreatment found only increased mental distress in relation to sibling victimisation in children 0-9 years but not in the self-reports of adolescents or young adults 45 . The two national maltreatment surveys in the USA and the UK revealed two other important findings in multivariate regression analyses. Firstly, the US survey found that sibling bullying independently predicted mental distress as much as child maltreatment and more so than sexual victimisation by adults 78 . Secondly, both US and UK surveys found that peer victimisation had stronger associations with mental distress than maltreatment by adults 47 . --- Table 4 here The cross-sectional studies do not allow for the interpretation of the direction of influence. It is possible that children who have emotional or behavioural problems are more likely targets of sibling bullying. A large prospective study of peer and sibling bullying recruited mothers in pregnancy and regular assessments of child and parent mental health and peer bullying were carried out. When the children were 12 years old they completed a detailed questionnaire about sibling bullying and mental health and self-harm experiences were assessed at 18 years of age 47 . This study found that after controlling for a range of family factors, pre-existing behaviour, and emotional problems as well as peer bullying, child maltreatment by adults, and domestic violence in the household, sibling bullying uniquely increased the risk of clinical depression and self-harm about twofold . Furthermore, a dose-response relationship was found, indicated by a linear trend: with increasing exposure to sibling bullying, the odds of mental health problems in young adulthood increased. Together, this suggests a causal relationship between sibling bullying and subsequent mental health problems. --- Balanced sibling relationships Bullying in the Family: Sibling Bullying Just as in relationships between friends, it is normal to have occasional conflict and disagreement between siblings. There is some evidence that small amounts of sibling conflict and their resolution may even have some beneficial effects on child development. Balanced sibling relationships, in which children experience both conflict and support, have been associated with the development of better social-emotional skills, including perspective taking, and the ability to understand and talk about emotions 8,[79][80][81] . Links have also been found with peer relationship quality, whereby having a balanced sibling relationship, comprising equal amounts of conflict and support/warmth, promotes greater social competence and can lead to better quality friendships with peers 82 . Indeed, occasional conflict in non-dominant sibling relationships has been reported to predict less peer victimisation, which may be explained by children acquiring and practicing conflict management skills at home, and transferring these to the school environment 81,83 . While it is clear that sibling bullying increases the risk of behavioural and mental health problems, occasional conflict, as part of a balanced, supportive and involved sibling relationship, provides opportunities for constructive conflict resolutions that can also have some positive benefits on children's emotional and social development 8 . --- Sibling bullying being "near normative" does not mean it is not harmful Considering the accumulating evidence of how widespread sibling bullying is and its adverse effects, it is surprising that there is still so little research on it. Some suggest that this can be traced to the general discounting of the frequency and seriousness by those individuals who most aptly could intervene at an early stage. "The age old adage "Kids will be kids" seems to have led to a pervasive belief that aggression and bullying between brothers and sisters is a rite-of passage and thus likely rarely investigated" 7 . For example, while the act of being hit or shoved off a chair in the office would lead to alarm and possible police involvement if done by a stranger, the same act may just attract a "come on, stop it now" by some parents at home 33 . Why do we assume that if it is done by a stranger it is harmful but if it is perpetrated repeatedly by a sibling it has no ill effects? Why is there a belief that because it is so frequent it does not need any intervention? Let us consider other frequent conditions where nobody doubts that they require prevention and treatment. The lifetime risk of any fracture has been found to be 53.2% by the age of 50 years among women, and 20.7% at the same age among men 84 . It is frequent and nearly normative to fracture a bone but nobody suggests just because it is frequent it can be left untreated and does no harm! The scars of sibling bullying can include physical injury which is often under reported and recorded 7 but many of the scars are also psychological. In reaction to a press release and report of our prospective sibling bullying study 47 , the BBC News received so many emails and letters from the public reporting on their experiences of sibling bullying and how it has affected their life that they ran a Magazine story using reader's experiences of sibling bullying . Christmas was always ruined by it. I was also bullied at school but it wasn't as bad as the Bullying in the Family: Sibling Bullying bullying at home. I suffered from depression for many years and have experienced workplace bullying and domestic violence. I know it is all connected to my childhood. --- Interventions There are no interventions so far that have been tailored to treat or prevent sibling bullying specifically. However, a number of intervention programmes have been developed which focus on improving sibling relationship quality by fostering socio-emotional competencies, emotion regulation abilities, and interpersonal skills, as well as offering parental guidance on how to intervene and mediate disputes or conflicts between their children. --- These general sibling interventions have reported positive effects on both parents and their children and the quality of sibling relationships . They integrate many aspects of behavioural and communication interventions, ranging from reinforcing positive communication and behavioural modelling, to the use of video-clips guiding parents and children to understand and find solutions for conflicts. Parents are often perceived as the "cotherapists" of how to manage and moderate fights and conflicts 85 . Considering the consistent finding of parenting being related to sibling bullying , teaching evidencebased effective mediation strategies for parents, encouraging children to vocalise and seek out conflict resolutions, to become more aware of their siblings perspectives, reflect on their feelings, control emotions and impulses and identify common ground 86 may be promising for developing intervention packages dealing with sibling bullying. Although some of these principles are part of interventions, they have not been sufficiently evaluated as yet. An alternative approach is to improve the wellbeing of the victims of sibling maltreatment and reduce negative emotional consequences by strengthening their family relationships and Bullying in the Family: Sibling Bullying enhancing their self-esteem by assertiveness training and cognitive restructuring rather than the focusing on the sibling relationships per se 87 . Table 5 here Furthermore, clinical treatment approaches using case examples or small evaluation studies have been considered for maltreated children placed in foster homes 92 or children who have experienced sibling abuse 93,94 . One such programme, the Promoting Sibling Bonds 92 used for maltreated children may provide an innovative integrated intervention model for children bullied by siblings in regular families. PSB integrates emotion regulation, social learning, family systems approaches as well as parental mediation to deal with sibling conflict. Considering how widespread the problem of sibling bullying is in all types of families, the use of social media and healthy game approaches that can reach all families may be considered in future prevention or intervention programmes, both for siblings and their parents. --- Future Research The research evidence on sibling bullying, its precursors and consequences, is still weak with just one, as far as we know, prospective study on the consequences of sibling bullying. This needs to be strengthened, and requires funding bodies to hear the voices of those who have been affected by sibling bullying. Firstly, as shown here, there is enough evidence to warrant further investigation of sibling bullying, its precursors and consequences. Secondly, all evidence so far comes from highly developed industrialised countries such as the USA, UK, Italy, Israel or Australia. There is a paucity of research in low and middle income countries and considering cultural differences in sibling relationships, studies among different cultures is required. Thirdly, there is a need to investigate whether reduction of sibling bullying can lead to reduced peer bullying and reduced mental health consequences of children. This requires the development of prevention and intervention programmes specifically against sibling bullying and their evaluation. --- Conclusions Sibling bullying is widespread with up to 40% being targets of sibling aggression every week or several times a week 47 . Sibling bullying, just like peer bullying 63 , is not a problem explained by poverty, poor parent education or single parenting, but is related to parenting quality in all socio-economic strata 1 . Sibling bullying also increases the risk of being involved in peer bullying, with sibling victims more often the target of peer bullying, and sibling bullies more often bullies or bully-victims at school. Sibling bullying is associated with concurrent emotional problems and distress and with diagnoses of depression and increased self-harm in early adulthood. Current evidence suggests that those who are both bullied at home and at school have highly increased emotional problems, likely because they have no safe place and thus no respite from bullying. Where investigated, there is a suggestion that early sibling aggression and bullying is a precursor of adverse relationships with peers 75 . Thus, if one wants to prevent sibling bullying and peer bullying, intervention has to start at home. --- TAKE HOME MESSAGE. While most sibling relationships may involve some rivalry and conflict between siblings, when the conflict results in direct physical or indirect psychological aggression that is repeated with the intent to harm , then it requires firm and fair intervention by parents or caretakers. Indeed, conflict needs to be solved in an amenable way before it becomes sibling bullying and there is a need for general prevention and early intervention trials and their evaluation. Clinicians should ask routinely about sibling bullying. The results indicate a significant linear trend with increased odds of depression disorder and selfharm at 18 years with increasing severity of sibling victimisation at 12 years.

Sibling relationships have a significant and lasting impact on children's development. Many siblings experience some occasional conflict, however, up to 40% are exposed to sibling bullying every week, a repeated and harmful form of intra-familial aggression. Evidence on the precursors, relationship with peer bullying, and mental health consequences of sibling bullying are reviewed. Parenting quality and behaviour are the intra-familial factors most strongly related to bullying between siblings. Sibling bullying increases the risk of being involved in peer bullying, and is independently associated with concurrent and early adult emotional problems, including distress, depression and self-harm. The effects appear to be cumulative, with those bullied by both siblings and peers having highly increased emotional problems, likely because they have no safe place to escape from bullying. The link between sibling and peer bullying suggests interventions should start at home. Health professionals should ask about sibling bullying and interventions are needed for families to prevent and reduce the health burden associated with sibling bullying.

Introduction Since the 1997 handover, the Umbrella Movement can be seen as the seminal politicizing event in Hong Kong politics, shifting the contemporary political context. Spanning from the 2013 publication of an article advocating for civil disobedience, until the People's Republic of China 's final rejection of proposed reforms in mid-2015, the movement critically redefined the relationship between Hong Kongers, their local government, and the prc . In the period between the 1997 handover and the birth of the movement, active dissent against the government was mostly limited to a vocal pro-democratic minority while civil society remained largely dormant , and the majority of Hong Kongers expressed faith in the system, evidenced by over a decade of public opinion polling demonstrating relative satisfaction with the system of government, despite the lack of institutionalized universal suffrage . An absence of democratic representation had been the norm in Hong Kong since its foundation, but the Umbrella Movement acted as catalyst, crystallizing mainstream dissent. By the end of the Umbrella Movement, Hong Kong's civil society and economy were deeply shaken, more so because the protests had disrupted political, economic, and social life, not only through the physical obstruction of normal day-to-day life, but also through external perceptions of political instability. It was whole-society encompassing, with an estimated eighteen per cent of the entire population of Hong Kong having taken part in the Umbrella Movement . This made it impossible to ignore, and it created a new set of expectations on the government from a population that had previously been largely apathetic. The Hong Kong Public Opinion Survey, carried out triannually from 1993, shows that in the second half of 2014, the tide irreparably turned against the Chinese one-country, two-systems framework, reflecting low levels of satisfaction with the Hong Kong government . While dissent expanded as a result of the Umbrella Movement, less clear is where the dissent spread, and, more specifically, amongst which population subgroups these shifts occurred in terms of differential support for the status quo or the governance model within 1C2S. This article examines changes in public opinion, testing whether they were evenly spread throughout the broader population as a result of the social diffusion of support and sympathy, or represented the deeper political polarization of specific subgroups, differentially predisposed towards the contention embodied in this movement. Increasingly, research has conceptualized the movement as a causal mechanism for other phenomena 263-293 2018). Lee, So and Leung analyzed the Umbrella Movement from a social impact perspective and labelled the movement a "critical discourse moment," reforming Hong Kongers' views on civil disobedience. Meanwhile, Lam credited the Umbrella Movement with strengthening Hong Kong's civil society and Hansen Edwards found that the movement changed Hong Kongers' views towards language and identity. Ma claimed that the movement "fostered a new political identity for Hong Kong," increasing disgruntlement and affecting the 2016 Legislative Council elections by making the prc an unpopular symbol . In a later article, Lam asserted that it spurred "decreased political trust, worsened evaluation of government accountability and deteriorated perceptions of social equality and election effectiveness" . While research suggests that the movement eroded support for the political status quo and for the prc, none yet has focused on the intricacies of subgroup changes within the Hong Kong population. This article empirically examines changes in public opinion through differential shifts brought about as a result of this protracted social movement across demographic spectra, some of which were more likely to have either sympathized with or participated in the original movement, in particular across age and gender lines. The Asian Barometer Survey datasets are well established as a tool for analyzing public opinion towards both democracy and regime legitimacy, even tracking changes across waves . This study utilizes data from late 2012 and early 2016 , focusing on views regarding the status quo and existing model of Hong Kong's governance, and expanding to determine which groups turned against the status quo following the Umbrella movement. The aim is to analyze whether the changes in public opinion differed across demographic groups, and whether they were perhaps largely endogenous to the original movement and its participants or universal and permeated equally throughout population subgroups. --- Background of the Umbrella Movement The Umbrella Movement emerged as a culmination of frustration over the stagnant political development in Hong Kong post handover. The yearning for autonomous political agency and representation, institutionalized democracy, and, finally, universal suffrage formed the basis for this discontent. In Hong Kong, the primary political cleavage is over democratization, and has been since the formation of the LegCo 263-293 Chan 2018, 42). Pro-democratic parties and candidates consistently garner the majority of votes from the Hong Kong population, and opposition to perceived democratic deficits drives political mechanisms, including popular dissent and skepticism over the 1C2S system . This is intimately connected with Hong Kong's complex relationship with China, with the cleavage over democratization and China's role in its implementation forming the basis of the identity split . Hong Kong is governed as a Chinese special administrative region under the Hong Kong Basic Law, which serves as a kind of local constitution, formalizing the 1C2S framework. This describes the state of Hong Kong temporarily maintaining an ostensibly distinct system of economic and political control, under an overarching prc superstructure, supervised by the Standing Committee of the National People's Congress of China , and meant to serve as the basis for the prc's administration of Hong Kong for the 50 years between 1997 and 2047. Its structure consists of a single-chamber legislature, the LegCo, and a chief executive. The elections of the chief executive and a large proportion of LegCo seats are carried out by a small set of elite leaders of interest group constituencies, primarily made up of Hong Kongers who support the maintenance of the status quo and "are reluctant to defy Beijing" . The actions of the Chief Executive are overseen by the Legislative Council, whose election process is marginally more representative, but which has no power to introduce bills. The Hong Kong government, under the 1C2S framework, is dependent on, and subordinate to, the npcsc. Thus, Hong Kong governance largely consists of this set of mostly unelected and unrepresentative institutions. Despite the Basic Law specifically stating that its aim is universal suffrage, elite conservatism and npcsc veto power have interminably postponed the promised introduction of full universal suffrage for either branch of the Hong Kong Special Administrative Region government . Despite the suggestion that this would be initiated in 2017, reforms never took place, leading to diminishing support for 1C2S, although Hong Kongers remained largely apathetic pre-2014, according to Hong Kong Public Opinion Research Institute data. These delays came to a head in 2013, when, on January 16, prominent Professor of Law, Benny Tai pushed for a movement of civil disobedience "as a means of democratic reform" . The movement began as a limited protest against delayed democratization, but ended as a broader critique of the status quo, as many previously apolitical citizens were galvanized by the government's harsh crackdown and lack of political responsiveness to what were seen as legitimate calls for change . It constituted a massive social movement, forcing Hong Kongers to take sides not only on Hong Kong's democratization, but also on whether disruptive political dissent was justified in support of that aim. The prc's statement in late 2014 on the "misunderstanding of the meaning of the notion [of 1C2S]," stating that "the high degree of autonomy of hksar […] comes solely from the authorization by the central leadership [… and] is subordinate," made it clear that Hong Kongers had no inherent rights to self-governance and that the Basic Law could be abrogated at the will of the prc . That year also saw the rejection of the liberalization of chief executive elections through the introduction of civic nominations, which would have meant that candidates need not be first nominated by the prc . This was labelled as a "critical turning point," and an "unreasonable" move, leading to deep disillusionment with the status quo and catalyzing opposition mobilization . Sporadic street protests occurred in response to these positions throughout mid-2014, and after the initial rejection of civic nominations, known as the 8/31 decision, a student boycott began to mobilize people for the occupation of central Hong Kong . The protests grew in intensity and in September, after being teargassed, the organizers utilized public outrage to expand mass protests, effectively blocking off and occupying large areas of downtown Hong Kong . The occupation continued for 79 days, until it was broken up by Hong Kong's local police force in what was widely perceived as a brutal response. The intensity of the anger at the perceived interference, with the perception of the prc tampering with Hong Kong "autonomy," led to discussion of "the option of separating Hong Kong from China," highlighting the conceptual linking of the hksar's incompetence with the structural interference caused by 1C2S and their connection with the prc . What differentiated the Umbrella Movement was its length, intensity, and the change in governmental response. Previous administrations had consistently capitulated when protests had grown to a sufficient size. Post 2014, despite clear vociferous opposition, the relationship underwent a permanent shift when the hksar government made it clear that they would no longer concede to protests. Government intransigence and the inability to placate protestors may have exacerbated the mainstreaming of dissent, with the public response largely mediated by snowballing grievances against the state . By mid-2015, the npcsc and LegCo finally rejected Chief Executive election reforms, which established a precedent for substantive questions on the future Innovation in the Social Sciences 1 263-293 of 1C2S and suggested that the hksar government and npcsc would neither tolerate extended unrest, nor respond to popular dissatisfaction with political concessions, including democratization. The Umbrella Movement was largely deinstitutionalized, with participants largely self-mobilizing; that is, it consisted of less strict or formalized modes of political activity, highly driven and coordinated by social media, reacting not only to failed democratization, but also to the state's responses to the movement itself . This deinstitutionalization and lack of organizational culture may explain the differential participation of age and gender subgroups, and the potential subsequent spread of the movement's ideals across demographic boundaries. Lateral mobilization through peer-to-peer social connections may have sociologically limited the action to those connected to novel mobilization networks. The movement's original participants self-defined as largely lower or middle class, harboring negative attitudes towards the government, and were primarily under the age of 35 . This age limit was apparently tied to a rejection of traditional politics because of the apparent impossibilities of advancing democratization, primed by a wave of student-led protests in 2012 . By the end of 2012, this student protest coalition was vindicated as the government conceded to student demands, galvanizing belief in the power of youth social mobilization. The 2012 organizing group, Scholarism, and the Hong Kong Federation of Students, Hong Kong's primary student organization, would go on to play leading roles in the Umbrella Movement, beginning with a student boycott in 2014 . Tracking changes before and after the Umbrella Movement in Hong Kong public opinion polling data reveals significant differences in the concept of national identity for young people, and differences in understanding in the principles of civil disobedience, confirming the relationships between age, mobilization, and polarization, while gender was apparently unrelated to mobilization or to voting preferences prior to the Umbrella Movement , despite suggestions that there may by gendered differences in terms of support for contentious politics . The fact that most of the protestors were young students in the images that blared across Hong Kong televisions triggered outrage when the protests met with a brutal pushback instead of compromise by the hksar government, provoking a critical "mobilizing impact" with "protect Innovation in the Social Sciences 1 263-293 the students," becoming an ubiquitous rallying cry . Survey data from protestors demonstrates that, while democratic values remained of primary importance with a score of 4.78/5, support for the organizational groups was viewed as of low importance compared to protecting the students , with the movement progressively winning sympathy from the older generations . Detailed interviews with Hong Kongers found that although young people were the primary participants, there was cross-generational ideological permeation . While the protest's leaders and early participants were largely politically active youths, reactions to the police response galvanized many others, potentially manifesting the concept of interpersonal solidarities transforming into public politics . All the same, the limited mobilization pathways and social destabilization may have instead exacerbated polarization. --- Theory and Methodology Social movements can be defined not only as a political process focused on achieving concrete ends, but culturally as creating a new political consciousness focused on an oppositional identity and culture, seeking social acceptance . According to this perspective, social movements may expand the base of public and social support for the ideas espoused by the group, influencing public opinion . Public opinion can be conceptualized as a reciprocal flow between the populus, media, and the state, with views regarding democratization, autonomy, and the status quo as much influenced by elite-controlled channels as by social connections . Some suggest that mobilization in a given context is reliant and predicated on existing public views, which then determine public uptake and tolerance of political change, while others alternatively view mobilization, the state response, media coverage, and then the social disruption as themselves dependent on the effects of an autonomous social movement, "reversing the causal arrow" . There is still no consensus on whether society must be sympathetic or ideologically receptive prior to mobilization, or whether movements should be conceived of as independently reshaping society, and, thus, causative. The first view can be seen as requiring a pre-existing "culture of mobilization," that is an inherent oppositional culture as the causal factor driving social movements . The alternative prioritizes cultural changes brought about by a Innovation in the Social Sciences 1 263-293 causative movement, with the latter "cultural creation" conceived of as public opinion-building, distinct from mobilization itself . This research accepts movements as potentially causal, focusing specifically on the degree and range of causal effects. Modern social movements, such as women's and environmental movements, may have been impactful in not only changing discourse but political values, with suggestions that while younger people and the better educated may be the "vanguard of these trends," they then drive society-wide changes . This implies that movements allow subgroups to potentially mainstream their views and aims into public discourse, where they are digested by the population through mass ideational diffusion, across subgroups and demographic boundaries. Hsiao and Radnitz suggest the opposite, that, instead, idea formation and polarization only take place along pre-existing political fault lines. Social movements deepening such splits can form pre-existing heuristic shortcuts influencing opinion formation in later instances . Such a view prioritizes fragmenting polarization within existing cleavages and hardening stances over crossdemographic solidarism and broad societal shifts. The corollary question regards the intrapopulation spread of movement ideals, asking whether the diffusion of these new social values is limited to participants and immediate peer groups, that is, those socially predisposed to dissent and simply mobilized by the movement, or whether the values permeate outside of the movement to those previously not predisposed to those ideas. This is an open theoretical question as to whether the "imagined solidarities" of the Umbrella movement were bridging, or instead sociologically limited to specific subgroups . The existing research lacks detailed quantitative examinations of public opinion shifts, including whether societal coalition forming is primarily endogenous to pre-existing group members and peers, or is instead taken up by the broader population, smoothening ideological differences across demographic lines. With the lack of substantive detail on public opinion shifts in the larger populus as a result of social movements, this article intends to address the gap regarding the vector of public opinion changes. It does this by measuring changes in the relationship between social proximity to the movement, segmented by demographic subgroups based on the likelihood of participation, and their changing views. It intends to test whether the Umbrella Movement simply deepened existing pre-held beliefs within its sympathetic "crowd mind," consisting of the subgroups who actively participated, or instead expanded Innovation in the Social Sciences 1 263-293 to demographic groups less likely to have participated, potentially catalyzing population-wide dissent by eliminating the pre-existing demographic differences that led to differential support prior to the events of the movement. The Umbrella Movement was demographically segmented, being largely propelled by student groups, with participants younger and more educated than the overall population. They largely supported increased democratization, and, thus, were skeptical of hksar and prc governments after the stalling of civic nomination and universal suffrage reforms . According to data collected by Tang , the majority of participants were female and better educated, with 79.3% having tertiary education, despite that statistic being 32.7% for the overall population according to 2016 census data; about 82% were aged 35 and below, while more than 90 % self-identified as lower or middle class. Over its phases, the movement attracted some previously apolitical individuals, including those mobilized in its later, more violent stages, towards political dissention . With certain subgroups more likely to have taken part in the Umbrella Movement-and, thus, holding parallel views prior to the movement-if there was a broad societal diffusion of views over the course of the movement, then as bridging interpersonal solidarities increased, the relationship between demographic subgroups and dissent should have diminished post Umbrella, as non-participants and their demographic cohorts developed sympathy for the group ideals. Instead, a movement in the opposite direction signals that, rather than cross-societal bridging solidarism, a polarization took place between two distinct socio-political poles, which can be demographically discriminated, with the established shifts in public opinion taking place primarily along preexisting demographic fault lines, rather than across them. The primary aim of this article is to test this bridging effect by measuring whether the strength of the relationship between those demographic variables and opinions of the status quo and the idealized model for Hong Kong's future decreased or remained the same over the course of the Umbrella Movement. While the limits of receptivity to opinion shifts may be tied to the effects of social movements themselves, such limits may also interact with demographic differences in receptivity towards contentious political action. While there are ample amounts of case studies and theorization, generalizable or large-sample empirical data on gendered and age effects are less detailed, especially in regard to receptivity to protest-movement aims amongst non-participants. Age has a clear and logical connection with both protest activity and with a dissatisfaction with status quo politics. Young people have lower social, occupational, and economic positions, more freedom, and less of a stake in the status quo. Demographic bulges, especially of capable but underutilized youths, Innovation in the Social Sciences 1 263-293 are frequently associated with unrest and challenging the status quo . Thus, youth participation in the Umbrella Movement is expected, and receptivity towards the movement-a contentious act-and the aims-a reset of the status quo-are expected to be inversely related to age and income, which is intertwined with age. Still, through social connections, it is expected that the relationship between age and status quo support can be challenged by the movement through bridging solidarity, and this notion will be tested. Gender is considerably more complex. There are suggestions that gender effects on both protest participation and support for a non-democratic status quo are not monopolar, but instead multipolar, depending on economic, social, institutional, and cultural factors, including the degree of equality . With disparate causal factors, gender could at once promote and discourage solidarity with such contentious politics. It could be dichotomized between two archetypical visions of a woman's role in society, with advocacy for gender solidarity and an increase in the sociopolitical status of Hong Kong women tied to prodemocratization, and advocacy for limits and stability, the latter culture in Hong Kong described as "fixed, static, monolithic […] hierarchical and patriarchal," tied to conservatism . Gendered political differences, even in terms of differential support for democracy, have been found in certain polities, theorized to come from the same basis of fear of uncertainty, which would predispose against contentious politics . This argument parallels the construction of a culturally embedded "maternal frame," affecting the gendered interpretative lens, but also external expectations of gendered behavior, precipitating a female aversion to political instability . Even within a culture, gendered factors may not be monolithic, but split. These views suggest that, while some might have been more galvanized into political action through discontent with the hksar government and the perceived overreaction of the security forces, for others, the Umbrella movement may have simultaneously promoted the contradictory impulse to reject destabilizing political acts as public endangerment. Parts of the existing social movement literature instead describe increased female support for nontraditional politics as split in terms of explanatory reasoning as either caused by differences in intrinsic "generalized risk perception" and situational analysis, the data for which is summarized well by Xiao and McCright , or by differing socialization processes. In the latter vein, gendered marginalization could promote the search for alternative venues for political representation, predictive of support for both the movement and Innovation in the Social Sciences 1 263-293 the anti-institutional aspects of oppositional culture, deepened through better access to informal social networks . In Hong Kong's case, some suggest that the opposition to hierarchical oppression encouraged female mobilization against the status quo during the Umbrella Movement, and that antihierarchical tendencies in Hong Kong can be gendered by uneven family relationships or resistance to Confucianism or hierarchism , in parallel to the concept of vulnerability theories as essentially solidaristic with anti-marginalization efforts . These traditional views present gender as parallel to social markers like class and age, in that feminist actors seeking empowerment will drive ideological dissent against the status quo, as well as participation in contentious politics. In terms of large-scale quantitative analysis, Dodson in 2015 found gendered differences in cross-national support for contention, with men more likely to participate in confrontational activities. In Taiwan, gendered differences were also found in the form of lower female Umbrella Movement support, but this was found outside of Hong Kong in Taiwanese women who were only exposed to the protests through the news and were not directly influenced by them, leading to the authors' suggestions that such differences relate to a differential knowledge of the movement between genders . Data from China suggest parallel gendered differences when it comes to engaging in political dissidence, with lower overall female support for protests, but suggest a bipolarization of receptivity in females in their finding that the relationship "between critical digital engagement and support for protest is stronger for women" , indicative of a multipolar gendered effect. This article tests whether the complex gendered relationship with confrontational politics indeed tends towards female aversion to contention and potential destabilization, or is multipolar, with data from Hong Kong and a high-visibility contentious social movement. As part of testing the capacity of the movement to bring about bridging changes in public opinion, differences in receptivity for demographic groups, including for age and gender, are measured pre and post the Umbrella Movement. Critical changes in views brought about by the Umbrella Movement are operationalized by examining the responses to two primary questions posed to abs respondents. The first asks whether Hong Kongers believe their "system of government needs change," conceptually a systemic evaluation of the current performance of the government and the 1C2S framework. The second question concerns the choice of an "ideal model for […] future development," and can also be seen as an evaluation of the system of governance, but it dif-Innovation in the Social Sciences 1 263-293 fers specifically in its priming of respondents by suggesting comparison models for answers, as well as in its use of the word "future", reflecting the longer-term prospects of those systems as Hong Kong moves towards 2047. Welsh and Chang examined this second methodologically challenging question, showing that countries with historically fraught relationships with China were less likely to choose it as a model, suggesting that this question is both deeply political and connected to the respondent's perception of their relationship with China . They also found that people who "viewed their own government as non-corrupt/practicing good governance did not choose China as a model or any other country as a model except their own," suggesting that this question similarly acts as an evaluation of the systemic performance of their own system relative to others. Those who viewed their own governments as corrupt, or held pro-authoritarian tendencies, were also more likely to choose the prc, with it serving as a direct alternative model to the US, suggesting a similar polarized dichotomization between democrats and conservatives, as dissatisfaction and democratic values drive model choice. Pan in 2020 analyzed the same question and concurred with this notion of the prc serving as a foil to the US, suggesting that China represents development at the cost of human rights, with those sympathetic to authoritarianism choosing it, while the US choice represents a rebuke to authoritarianism . Thus, drivers of satisfaction with the current system and preference for a future model can include satisfaction with the existing administration, views towards democratization, and views towards the relationship between Hong Kong and China within 1C2S. It is expected that the Umbrella Movement, in broadening a rejection of the prc-driven status quo and the prc's interference in Hong Kong electoral system reforms, would turn people away from the "China" model offered. This article ultimately aims to quantitatively measure whether the expansion in dissent occurred across a broader societal coalition of the general population based on a form of solidarity-building, one that bridged a more diverse set of demographic groups than those who were connected with the original activist group, or rather occurred primarily within the original participant subgroups. --- Research Design The Umbrella Movement, as defined in this study, includes the full context of the initial movement itself, the response elicited from the government, and the societal reaction to that response as mediated by movement expectations. The Innovation in the Social Sciences 1 263-293 prc and hksar's rejection of reforms amounted to, essentially, a continuation of pre-Umbrella status quo politics, so the state response was singularly driven by this movement and is part of the movement's causal package, as was opinion formation propagated through the media, regardless of intent, contained within the broad effects of the Umbrella Movement . hkpori polling suggested that over the course of the movement there was a plunge in support for both the status quo and 1C2S, with net support hovering around zero by the occupation stage of the Umbrella Movement. Knowing that movement participation and, therefore, strong dissent was demographically skewed, if social-bridging solidarism and a diffusion of opposition towards the hksar and 1C2S did indeed take place, then the share of this shift should be spread throughout disparate groups of the population, and, therefore, the correlation between demographics and dissent would be expected to decrease. The alternative is an internal radicalization of views within limited social circles, manifested along demographic lines, with the expectation of a sharper relationship between demographics and views post Umbrella. These two alternative models were tested using this novel means of analysis. Data from waves 3 and 4 of the Asia Barometer Survey were compared as two temporal snapshots, pre and post Umbrella. The dates for wave 3 sampling were from September to November 2012, and wave 4 was from February to April 2016. Sampling for the third wave preceded the January 2013 ideological birth of the Umbrella Movement by more than two months, and the fourth wave followed the Legislative Council's final rejection of the Umbrella Movement's goals in June 2015 by more than eight months. The Occupy Central campaign fell in the center of these two periods from 28 September to 15 December 2014. No potentially confounding major changes in political leadership took place in any related states. Although Xi Jinping's transition took place during polling, he was the presumptive replacement leader since his appointment to the Politburo Standing Committee in 2007. To examine the movement's differential effects across subgroups, demographic cohorts were segmented by gender, age, years of education, and income. Two dependent variables were chosen, a clean measure of support for the status quo and another for support of future movement towards China. Question 87/84, representing general support for the status quo of the system of government in Hong Kong, asked, "Compared with other systems in the world, would you say our system of government works fine as it is, needs minor change, needs major change, or should be replaced?" with answers on an ordinal scale from to , with a higher response indicating more intense dissent against the system of government. Sciences 1 263-293 Question 167/159 asked about an ideal model choice for Hong Kong's future, phrased as, "Which country should be a model for our own country's future development?" Answers included the US, China, India, Japan, Singapore, and Hong Kong as non-ordinal categories. Multinomial logistic regression was used for the selected dependent variable with the China model selected as the baseline. All calculations utilized the original data coded by the Asian Barometer Survey. --- Innovation in the Social Gender was a binary coded variable. Reported income was quantified by quintile, -. Income was separated by quintile. Age was statistically calculated as a raw number, but in data tables age was split at 40 years of age. Most characteristic variables had economic or social connections to the movement, so size of household was introduced as a variable to control against spurious wave effects. The data was cleared for all non-respondents to any pertinent question. The statistical analysis was performed using R software. --- Findings A preliminary examination of the data sets shows that the reported protest participation rates jumped between waves, from 4.44 per cent in 2012 to 8.42 per cent in 2016, suggesting fairly broad population-wide participation. Income was higher in both waves amongst protest participants than nonparticipants, although the margin between the two groups had decreased significantly by 2016. Disillusionment with the status quo was also higher amongst protest participants than non-participants in both waves, and this measure increased across waves, along with the age margin. The mean age of protest participants dropped across waves, and the margins between participants and non-participants increased, indicative of demographic polarization in terms of participation rates. For gender, there were notable differences in participation, and a growing margin post Umbrella suggested reduced female participation. Finally, the average level of education remained higher for participants than non-participants throughout the waves, and the margin again increased significantly, also suggestive of strong polarization. By 2012, there was a clear shift away from support for the status quo. From 2012 to 2016, the proportion of the subgroup members choosing , stating that the system of government "should be replaced," and , that the system of government "works fine" increased in every single subcategory, except for males choosing , and males' selection of significantly outpaced females in both waves, suggestive of a large polarizing split by gender. This is not only a relative difference. The data suggest that females, both pre and post Umbrella, were far more likely to express satisfaction with the status quo, and that, despite higher rates of participation in the Umbrella Movement than males, female proportional shifts against the status quo were almost at parity with female shifts towards the status quo , suggesting an intragender bifurcation. Post Umbrella, lower-income citizens were more likely to have shifted towards , the system "needs major change," than , the system "should be replaced," while higher-income citizens were more likely to select the latter . Age displayed similar trends with similar changes around and , but in the selection of and , younger people moved in large numbers towards replacement , while older people moved in larger numbers towards major change , suggestive of a diffuse effect across generational cohorts, which was less radicalizing amongst older Hong Kongers. A polynomial regression for the question on status quo suggested that age was negatively related to dissent against the status quo in both waves and this relationship persisted post Umbrella. Gender, while not significant beforehand, became strongly associated with status quo support and highly significant, with females expressing far more support for the status quo post Umbrella than males. The income quintile was unrelated, but years of education had high significance in its relationships, with more highly educated Hong Kongers likelier to express dissent against the status quo, but only post Umbrella. These relationships are highly indicative of ideological polarization along demographic lines. Responses to the status quo question on "our system of government" separated by demographic subgroup, with percentages calculated, with data from all respondents . A comparative anova was performed to compare binarized contemporaneous groups , and to evaluate change for same-group results . Scale: "Does not need to change;" "Needs minor change;" "Needs major change;" "Should be replaced" The post-Umbrella aggregate data recorded large increases in support for the China model for future development, mostly at the expense of support for the model of independent Singapore, with increases in every subgroup. The largest increase was for females, low-income respondents, and those over 40, with females far more likely than males to have shifted to China after the Umbrella Movement. Older Hong Kongers' increase in support for China was nearly double that of younger respondents. The intense relationship between age and the choice of the China model, relative to the US model, persisted over the course of the Umbrella Movement, and emerged as significant post Umbrella, relative to the Hong Kong model, with older people less likely to choose it relative to China. Income quintiles displayed no differences in 2012 in terms of the empirically interesting choices of the US and Hong Kong models relative to China, while becoming significantly positive in 2016 towards nearly all models relative to China. Years of Innovation in the Social Sciences 1 263-293 formal education became significant in predicting the choice of the Hong Kong model relative to the China model. Finally gender became highly differentiated as females, post Umbrella, became significantly more likely to select the China model relative to any other choice. The control variable was analytically insignificant in both waves across all statistical analyses. This data is strongly indicative of a large polarization of the choice of the China model relative to the Hong Kong model, with the strength of the relationship between demographic variables and views undergoing large and significant shifts. --- Analysis The clear overall shift in support away from the status quo, with the largest shifts from "it needs a minor change" to "it needs a major change," was consistent with expectations for an aggregate society-wide shift during the period of analysis based on hkpori data. This is suggestive that the Umbrella Movement was the critical moment in politicizing a plurality of Hong Kongers, representing a large-scale shift against the status quo. More significantly though, the bulk of the data suggests that the Umbrella Movement's social bridging was not only insufficient to erase pre-existing relationships between demographics and public opinion, but also that this period exacerbated demographic divides, including along gender and class lines, despite females and non-elites being clearly overrepresented in the movement itself. This was true both in terms of the raw proportions of group shifts and in terms of the relationships between demographic subgroups and public opinion. The implication is that whatever bridging may have occurred was overshadowed by inherent differences in perceptions of and tolerance towards the movement. The increased strength of the relationship between views and demographics for key demographic variables is suggestive of an ideological breakdown between demographic cohorts, as the contentious movement hollowed out the apolitical middle. The increase in the selection of the two extremes and strongly suggests a polarization between movement supporters and detractors, with status quo shifts intimately connected to large movements in support between the US, Singapore, Hong Kong, and China models. Further, subgroup shifts relative to the status quo took place in parallel with subgroup shifts regarding China, implying that the status quo within the hksar and the perception of the China model and, perhaps, prc interference, were conceptually linked. This highlights the complexity of the relationship between Hong Kong and China and reinforces the link between dissatisfaction with the existing hksar government under 1C2S and the basis of both that administration Innovation in the Social Sciences 1 263-293 and the structure, subordinated to the prc, a model that certain groups moved away from in patterns highly similar to their movement away from the status quo. This is reflective of the concept of pre-existing cognitive frames, further bifurcating Hong Kong across its primary cleavage , and suggests that the movement did not create bridges, but instead exacerbated demographic and subcultural differences. Educational status became significantly predictive of dissent against the status quo, as well as support for the Hong Kong model relative to the China model. Because of the role students played in the original Umbrella Movement and parallel political actions, their overwhelming support for democratization, and the differential social status of students, this was expected. However, because of the limited numbers of actual participants out of the broader population, much of the observed effect in this survey was invariably caused by nonparticipants, suggesting that differences along educational attainment lines were also tied to social groupings tied to age and class, in addition to the link between student status and participation. Income itself was not predictive of support for the status quo in either wave but became extremely predictive of support for the US model at the expense of the China model. Considering the economic links with China, the high representation of the economic elite in the functional constituencies, and the fact that original movement participants were less likely to report themselves as high-income earners, this is unexpected. It perhaps represents a covariance with educational status but may also indicate changing views regarding 1C2S and the hksar's capacity to preserve political and, thus, economic stability long-term in Hong Kong, as high-income Hong Kongers living and working in the downtown core were highly affected by the Umbrella Movement. Age was consistently significantly predictive in terms of both support for the status quo and model choice. There seemed to have been little shift based on age, suggesting that ideological diffusion across generational cohorts, which did occur, was more tempered. In examining the raw response numbers, it's clear that older Hong Kongers' positions also largely moved against the status quo, though the extent to which they moved was smaller than for younger Hong Kongers, perhaps linked to differences in their respective stakes in status quo maintenance, consistent with long-standing data linking age with conservatism , perhaps out of an aversion to destabilizing large-scale political reform or aggravating the prc. Gendered differences experienced large shifts, and gender emerged as the most significant demographic variable, with females much more likely post Umbrella to support both the status quo and the China model. Like age, shifts away from were similar for both males and females, with the larger difference lying in where those views shifted to, with females and older Hong Kongers often picking or , while males were more likely to select . A differential movement towards the extremes is suggestive of a gendered polarized radicalization, with nearly as many females increasing their support of the status quo as moving against it. Males moved significantly further from and diverged from females. While the data suggests that females made up the majority of movement participants, both pre and post Umbrella, Hong Kong women appeared to be more, and increasingly, satisfied with the status quo than males, challenging the literature suggesting few gendered political differences in Hong Kong. The significant changes around gender, despite Tang's finding of relative gender parity within the movement itself, reflect significant male shifts in support away from the status quo, but also increasing female support of both the status quo and the China model, with these differences becoming significant post Umbrella. This is indicative of possible gendered distinctions in not only predispositional support for contentious politics, but also the permeability and diffusion of dissent through contention. It supports the notion of a degree of gendered aversion towards destabilizing politics and towards the status quo. However, the source of the socialization of these differences, and whether they are consistent and cross-national or tied to subelements, remains to be explained. The tendency for male but also younger and better-educated Hong Kongers to continue to support the Hong Kong model, in spite of their disillusionment with the 1C2S framework, suggests that, despite sociopolitical turmoil, those groups maintained their aspirations for reform. Persistence and even increases in the relationships between demographic variables and views towards the government, as well as the China and Hong Kong models, suggest that the diffusion of movement ideals was far less prevalent than a sharpening of the demographic cleavages underlying the poles on Hong Kong's ideological spectrum. This suggests that intrapopulation grouping, operationalized by demographic differences, can predicate against wholepopulation change if large groups are not predisposed to receptivity towards both a movement's aims and also its tactics. Thus, instead of shifting population-wide thinking en masse, a movement can alternatively polarize a society, bifurcating social goals. --- Conclusion Based on the predisposition of certain groups to have participated in the Umbrella Movement because of their dissent against the hksar government or the 1C2S framework, the expansion in the relationship between demographic variables and views towards the hksar and prc is strongly indicative of the Umbrella Movement's ideological permeation remaining primarily endogenous to specific predisposed subgroups, while radicalizing outsiders against its views. The directionality and growth of these relationships empirically challenges the notion of interpersonal solidaristic movement building, and questions whether the theorization of a social movement's purpose as a public opinion-shifting exercise can be blind to pre-existing demographic and social splits, highlighting the role of pre-existing identity and subgroup differences in both the movement's mobilization and support, and public opinion changes. The increasing association strongly predicates against any potential bridging effects of the Umbrella Movement, and instead suggests that there was a polarization in many groups against the aims of the movement and in support of the status quo. Polarization around age lines was consistent with expectations and suggested little bridging effect. Counterintuitively, polarization along gender and income lines ran contrary to participation statistics, with the finding that, overall, despite females having been more likely to formally participate than males, they were also more likely than males to support the status quo and the China model both pre and post Umbrella Movement, while also displaying a degree of intra-gender polarization. Such parallels indicate that in Hong Kong, views towards the status quo are inseparably linked with views towards the prc. While the reasons for the relationship between age and educational status are clearer and more tied to political openness versus conservatism, the unexplained gendered demographic differences in terms of support for the status quo and the prc, never before manifesting in electoral data, may instead be explained as latent value differences, which question the very limited literature regarding the gendered effects of social movements on public opinion. Most strikingly, while females represented more than half of the participants in the Umbrella Movement, they were disproportionately far more likely to shift their support in a political direction in opposition to the aims of the movement than males. This apparent paradox in post-movement public opinion shifts deserves a more detailed examination, as does the movement of high-income Hong Kongers away from the China model. These findings have profound implications, not only for understanding Hong Kong opinion changes post the Umbrella Movement, but also for gener- alizing social movements' differential effects on subgroups with distinct predispositions, which may be polarized by protests rather than ideologically captured. With no evidence that bridging occurred, but instead that the relationship between public opinion and demographic variables sharpened, the intersectionality of those demographic traits and subgrouping within the population led to an acute split, not only between supporters and detractors of the movement, but in terms of satisfaction with the status quo and stalled democratization itself, which can be tied to the society-encompassing social movement. With exit polling data, it may be feasible to test whether these polarized changes in views corresponded to actual voting differences. Further, this data was limited to two waves because the questions were only introduced into the Asian Barometer Survey in 2012. Additionally, if data corresponding to the 2019-2020 wave of protests in Hong Kong emerges, one can hope that the opportunity will arise to elaborate on these unexpected findings.

This article explains post-Umbrella Movement shifts in public opinion in Hong Kong by correlating demographic subgroups with changes in support for the status quo, utilizing Asian Barometer Survey data regarding support for the existing system of government and idealized models for the future. The aim is to determine whether public opinion shifts occurred evenly through population subgroups or were limited to participants and peers. The findings show that while Hong Kongers' views underwent a seismic shift, the expansion of dissent was unevenly distributed across demographic cohorts, rather than representing demographically symmetric diffusion reflecting societal bridging and cross-demographic coalition building. This indicates that, while the social movement may have shifted aggregate public opinion away from the status quo, its precepts permeated differently across age, class, and especially gender lines, and that views on the status quo were also strongly linked to idealized models for their future, relative to China.

Introduction The articles in this issue offer six case studies from across Europe, ranging from the 1940s until the present. These represent diverse and distinct contexts, with vastly different systems of culture, morality, governance and welfare. Yet, in all of these case studies, authors look 'from below' to show how individuals have frequently thought of themselves within 'families'. This powerful construct has governed how individuals have organised their reproductive lives, employment choices and leisure time. Particularly from the late twentieth century, amidst new communication technologies, increased education and focus on public participation in political life, ideas of 'family' also came to shape individuals' campaigning and political behaviours. Countless voluntary associations were forged across Europe, with varying aims, but they acted in the name of 'the family', whether looking to defy, ignore or destabilise the expectations of nations and their welfare apparatus. Putting the idea of 'the family' centrally in our historical work then, as an agent of change, can extend our visions of 'the political' and throw new light on our understandings of how change has been enacted across and within Europe. This special issue is fuelled, first, by interest in placing concepts of 'family', 'welfare' and 'democracy' into conversation and, second, by a desire to assess how pressing contemporary concerns have been shaped and their limitations and boundaries negotiated by large-scale European shifts since the 1940s. In this moment, shaped by historical visions of national pride and decline, transnational integration is potentially threatened by movements against the European Union in the United Kingdom and beyond. The COVID-19 crisis also sees governments 'lockdown' their borders, or disincentivise international travel. Following previous periods of warfare and crisis, national populations have responded with sympathy, rage, suspicion and indifference to mass movements of refugees, showing the strictly policed limits of 'universal' welfare systems and the simultaneous porosity and strength of imagined national borders. Echoing previous cyclical and ideological changes, across Europe political parties have imposed periods of austerity amidst debate about the limits of free trade, tariff reform and the 'inherent' productivity of populations. State services and welfare systems have faced increasing demands and decreasing budgets, motivating the revival of historic forms of voluntary action, which challenges and at times replaces government provision and reconstructs boundaries between 'deserving' and 'undeserving' sick and poor. At the same time, highly traditional models of 'the family' are revealed, on the one hand, by media and religion, proffering cultural visions of, for example, royal families, and at the same time by nation-states which provide tax incentives to married couples or those that display preferred models of fertility and breeding. The COVID-19 crisis fuels debate about whether authoritarian states are more able to protect their populations from disease, but at what cost for family and individual freedoms? Amidst concern about 'fake news' and 'distrust of experts', which may feel highly contemporary, this collection traces contingent, dynamic and longstanding interactions between family, civil society and nation-states, from the 1940s until the present, and across Europe. Together, the articles speak to the contemporary challenge of whether historians may think optimistically about the potential, power and future of 'the family', despite the varied baggage and ideological assumptions that have underpinned this term. This special issue also demonstrates that histories of broad-scale state-level change of professions, power and global politicsmust take families seriously and make them central to their analysis, not only as subjects of global change but as active participants. The special issue emerges in part from a collaborative research network, The Quest for Welfare and Democracy: Voluntary Associations, Families, and the State, 1880s to the Present. 1 This network was initially sponsored by the European University Institute and is now supported by a grant from European Co-operation in Science and Technology . The network met for collaborative symposia at the European University Institute in February 2016 and November 2017, forging productive discussions from which many of these articles emerged. Three main ideas developed in these discussions, which will be explored below. The introduction first explores grand narratives of welfare and democracy across Europe in the twentieth century, which tend to position families as recipientsor victimsof state power. It then moves on to offer new working approaches to analyses of 'family' and 'agency', through which we can reassess large-scale narratives of democratic change across this significant period. In a third section, the introduction demonstrates how, by using these broad definitions and taking family seriously, we see new visions of chronologies and geographies of change in mid-to-late twentieth-century Europe, more expansive definitions of 'activism' and the ongoing power of gender hierarchies in structuring daily life. --- Grand Narratives Grand narratives of European welfare and democracy over the twentieth century often place families at the whims and behest of warfare, state politics, and demographic and cultural change. Warfare across and beyond Europe is particularly central to these narratives, as a disruptor of family life, but also as moments in which families appear as a symbol of hope and reconstruction. Vast historiographies surround, for example, the evacuation of children during the Second World War across Europe, perceived to protect future families and generations. This research is very much tied to ideas of welfarein British debates, historians continue to contest and nuance the interpretation of Richard Titmuss that evacuation, revealing the state of working-class children and their health, was a key motivator behind the introduction of the welfare state. 2 These works are significant in connecting historiographies around welfare and warfare but, at the same time, often struggle to centre the experiences of family life or childhoodechoing the critique of Harry Hendrick that the familial issues of separation and emotional adjustment were 'rarely taken seriously by those who made policy' and indeed that 'much more important was the influence of political and economic calculations'. 3 Hendrick argues that the work of 'social investigators, eugenically inclined pronatalists and, most influentially, child psychiatrists/psychoanalysts and social workers' all instead propagated a wartime notion of children as 'social investments', whose futures must be protected because they determined the future of British society. 4 Within this formulation, the family likewise was a unit to be mobilised or utilised by social policy makers and professions, rather than an independent agent of change. Further entrenching the prioritisation of the state as the central actor in existing historiographies, significant literatures also consider ways in which European dictatorships have separated families and subjected them to political violence, coercion and control. In an issue of Contemporary European History, Paul Ginsborg described how the regimes of Adolf Hitler, the communists of the Soviet Union and Francisco Franco persecuted and discriminated against families deemed 'hereditarily unfit', 'hostile', 'worthless', 'foreign' and 'disease-ridden'. 5 Fleeing state violencefrom their own and nearby governmentsrefugees moved across Europe: from the Basque Country to the United Kingdom during the Spanish Civil War; or from Austria, Czechoslovakia, Finland, and Poland to the United Kingdom, Belgium, Sweden, France and Holland before the Second World War. 6 These movements, Peter Anderson has argued, can be perceived as a lens through which to understand 'entangled relationship[s]' between campaigners across borders, as well as 'battle[s] for control' of 'children's minds or souls'. 7 Once more, as in literatures about the recruitment of migrants across Europe for the labour market, the family is presented as an object for government, voluntary or professional visions of recovery, reconstruction and control. 8 Children's and families' 'minds' or 'souls' are represented as malleable and available resources in historiography, as well as in historical policies and representations. States have also mobilised their welfare systemsa central object of interest within this special issue to manipulate and reshape family life. Previous scholarship has looked to consider, for example, political discussions around whether family life should be protected by the state, or shaped by it; a distinction framed by approaching state-family relations 'from above'. 9 State welfare systems could disrupt family livestaking children into careor enrich and support them. This simple dichotomy has been complicated, meanwhile, by the politicisation of welfare systems used to, for example, take children into care if their parents were deemed 'morally unfit' because of their 'religious failings', or if their families opposed current political systems, for example as in 1920s Spain during the Much existing historiography therefore focuses closely on the ideology and ideas of familyon familialismrather than on the agency and actions of families themselves and the power of their everyday, and indeed at times remarkable, lives. Certainly, existing historiography usefully demonstrates ways in which governments have sought to exploit the potential of the family, creating experimental and interventionist 'utopian' schemes. These schemes have tried to reshape society but also, at times, humanity, for example through selective sterilisation, pronatalism and the provisionor absenceof state childcare and education. Families in this vision may be protectors of future hopes, or offer ways to reconcile and recover from violent ideological pasts. 12 The family has not always been central to ideological life and scholars have considered the absence, as well as the presence, of familialism within state apparatus. Elizabeth Waters has argued that the Bolsheviks in Russia did not make the family central but rather expressed interest in liberating women and creating a more productive society, by supporting community rather than family or women's responsibility for childcare, housework and home-making . 13 Families are thus often presented in existing historiography as respondents to, or symbolic within, large-scale change. Other strands of historiography still consider the family as a demographic unit to be analysed and assessed in terms of its quantitative structures. Edward Shorter, John Knodel and Etienne Van de Walle, for example, have argued that between 1880 and 1940 'illegitimate' fertility in Europechildbirth outside of marriagefell 'precipitously', by 50 per cent or more in most countries. Despite exceptions in terms of, for example, the United Kingdom, Switzerland and the Netherlands or Portugal and Sweden , these authors argue that the parallels in timing and pace of decline across Europe were 'close' in a large number of countries and provinces. 14 After 1940, the authors write, some countries experienced an 'illegitimacy boom'. 15 These types of demographic histories provide an analysis of family structure; and yet, this special issue contends, to understand the meanings of these changesand indeed their causes, and the causes of such national variationwe must simultaneously utilise sources that represent the qualitative experiences and lives of those affected. Analysing families as 'agents of change', as in this special issue, enables us to understand the parallels and specificities in national stories, and to see how everyday life interacts dynamically with global change. --- Redefining Family As Alexandra Walsham has argued, 'working definitions' are 'themselves crucial tools for rethinking existing approaches'. 16 This special issue argues that we must redefine 'family' in a broad, dynamic, expansive way which both requires a shift in definition and, in parallel, a rethinking of this unit's role, relevance and relationships to welfare and democracy, whereby family life is placed centrally as an agent of change, not merely as a passive recipient or bystander. In terms of existing definitions of family, a classic formulation by Raymond Williams tells us that the term 'family' only came into 10 Peter Anderson, 'The Struggle over the Evacuation to the United Kingdom', 300. 11 Stokes, 'An Invasion of Guest Worker Children', 372-89. 12 Chappel, 'Nuclear Families in a Nuclear Age', 86. 13 Elizabeth Waters, 'The Bolsheviks and the Family', Contemporary European History, 4, 3 , 275-91. 14 English in the fourteenth and fifteenth centuries, emerging from words for 'household' and 'servant'. The idea of the family as 'the small kin-group, usually living in one house' became powerful by the nineteenth century. 17 This kind of narrow definitionthat focuses on the shape of families, rather than their meaningfails to recognise the ways in which family life has been defined in relation to moral and cultural values. Certainly, over the mid-to-late twentieth century, ideas of 'normality', 'stability' and the 'natural' permeated and shaped how families were recognised and rewarded in welfare states, and how individuals themselves constituted and discussed their roles in household and domestic spaces. Many of the governmental and bureaucratic categories encountered in this collection operated with tight categories of 'family' which often revolved around heteronormative models of masculine leadership in married heterosexual couples. Official inquiries, Lindsey Earner-Bryne argues, 'generalise[d]' about the family 'without due attention to differentials of power and wider structural issues'. Yet, as contributors repeatedly demonstrate, the families that emerged, lived and coped within these systems were not the two-parent and two-child models which the systems assumed. Accordingly, and in order to recognise the significance of daily life, this special issue takes the broadest possible definition of families. It accepts that 'families' have been constructed beyond biological boundaries and reflect a series of personal, community and structural settlements and accommodations. 18 This broad definition enables the issue to discover connections between the shape of families and the moral politics which have been formed around their political actions and daily lives. Indeed, various moral visions of the family are significant throughout this collection: articles explore how cultural and social norms inflected visions of the family as 'functional', 'new' and 'old', 'market-conforming', 'strong', 'civilised', 'patriarchal' and a solution to, or cause of, moral dilemmas. Contributors also analyse a range of family structures, variously brought together by service, generation, duty or payment. Groups have sought out social and economic benefitfrom churches, charities and stateby defining themselves as 'family', and thus have lived their lives in accordance with this idea, reshaping it accordingly. Families themselves have resisted state and professional categories, as well as having adapted to them. We see in this analysis how individuals whose biological families may not have adequately met their needs sought to readopt this concept to reflect the material conditions of their lives. Recognising this broad and expansive definition of family life also means accepting the premise that, for individuals and in culture, families negotiate everyday and political meaning. They are intimately tied toand defined bythe most fundamental human events: birth and death, as well as by complex and powerful social rituals around marriage, separation and friendship. Families are key providers of individual 'fulfilment', social care, and formal and informal education, but also, at times, responsible for psychological trauma, frustrated ambition and unhappiness. The family thus has much potential to provide services and to shape future citizens and their subjectivities as 'docile', 'productive' or 'liberated'. Paul Ginsborg and Jane Humphries have also discussed how the family can be 'subversive' and, as Humphries writes, 'resist impositions which force it and its individual members' into distinct ideological mouldseven those proffered by academic and popular writings. 19 family' may, we see throughout this collection, operate as a unit in responding to social or political change, even while its members live in significant disagreement and conflict. Thus, this special issue not only aims to look at 'the family', but seeks to blend large-scale analysis of changing trendsof political and global changewith recognition of the everyday and the 'small' in history. In doing so, it extends a recent emergent historiographical trend. Significant here is the landmark work of Elaine Tyler May, whose study of post-war families in America located their thoughts, feelings and emotions 'within the larger political culture' of the Cold War, 'not outside it', making a persuasive case that Cold War ideology and domestic revival were 'two sides of the same coin: post-war Americans' intense need to feel liberated from the past and secure in the future'. 20 Paul Ginsborg has likewise drawn on Hegelian analysis to argue that we must consider relationships between the family, civil society and the state, and the 'interlocking and conflictual' relationships between these actors which 'define the boundaries of politics'. 21 For Ginsborg, such analysis will make a 'methodological priority' of connecting family, political and social history. 22 A recent edited collection by Hester Barron and Claudia Siebrecht, Parenting and the State in Britain and Europe, c. 1870-1950 , has also been significant in this trend, showing the benefits of comparing democratic and fascist regimes across Europe, and of centring parents as units of analysis. 23 This analysis aims to subvert and challenge accounts by, for example, Emile Durkheim, Michel Foucault ormore recently -Nikolas Rose, which placed families, citizens and children as objects within national, psychological and educational interventionsabsorbing and reflecting, rather than reshaping, social and cultural norms. In particular, this collection is interested in 'agency' and the family as 'agents of change'. This is a fraught concept. In recent years, attempts to find 'agency' in women's and children's history have been criticised as inadequate. 24 Mona Gleason and Lynn M. Thomas have provocatively asked whether historians have made agency an endpoint of study rather than a beginning. 25 Making a related critique, Chris Millard has explored whether historians valorise 'experience', building on the disciplinary assumptions of psychoanalysis, social history and anthropology, and leave invocations of 'experience' as seemingly irrefutable. 26 This collection takes these lessons seriously, yet argues that agencylike familycan provide a significant lens for analysis. Cutting-edge research, particularly in the history of childhood, has used agency in productive ways. Kathryn Gleadle and Ryan Hanley, for example, have shown that it is productive to read even adult-generated sources through the lens of children's agency, particularly by looking at absences, while Susan A. Miller, Tatek Abebe, Mischa Honeck and Gabriel Rosenberg have unpicked how social contexts shape the boundaries of youthful agency over generations, space and place. 27 The idea of 'agency', in these accounts, enables us to think seriously about the choices that individuals and groups have made, and the paths that they have forged, while also recognising how individual 'choice' has been bounded by structure and relationships. This collectionbecause of its scope across time and spacelooks to further expand the idea of agency as relational, in two ways. First, the case studies in this collection show how the capacity of families to act as 'agents of change' has been contingent on the structural conditions of authoritarian, liberal democratic, or fascistic societies, and on families' identities of gender, race, ethnicity, class, religions and disability. Not all families have been equally able to exercise 'agency' in lobbying for change, despite employing similar mechanisms of campaigning and resistance. Second, the collection shows how the agency of 'the family' has been itself fundamentally shaped by familial discord and conflict. Articles in this collection frequently show moments in particular where women's interests in the family came into conflict with those of men. Yet the collection also discusses when extended family members had different priorities to biological parents and when children were not necessarily best represented by their parents. As mentioned, internally, families have been repressive, patriarchal and abusive as well as moral, supportive and caring. Looking to 'agency' as relational makes this clearer and helps us to pay attention to distinctions within 'the family', while also taking seriously the ways in which individuals have nonetheless used this construct to mobilise for change. The collection thus argues that analysing the family as an 'agent' in social change makes clear the power of this construct in policy, welfare systems and daily life across time and space. The idea of 'agency' also affords space to analyse the multiple relationshipsby and within familiesthat shape action. So, rather than consider how nation-states looked to control families across this period, which has so far been a dominant trend in writing about families in history, this special issue breaks with prevailing interpretations to prioritise families themselves as historical actors, as subjects and agents in historical change. Furthermore, putting familiesnot merely citizens, children or 'members of the public'at the centre of our examination is critical, not least for the political symbolism of this unit. Focused discussion of the family can easily be missed in scholarship, when constructions of family are both highly contested but also mundane: difficult to define and yet also eternally present, in political campaigning, daily life and cultural representation. By making the family central, this special issue looks beyond the individual case studies provided and rather seeks to challenge ongoing scholarship to make the agency of this significant social grouping a central focus. Furthermore, this collection demonstrates that, despite the complexities of the term 'family', placing this unit centrally can lead to dynamic and important scholarship, which interweaves across and between existing literatures on international, national and daily life, and throws new light on ideas of citizenship, solidarity and inclusion. --- New Themes This special issue thus extends and applies a significant, emergent methodological approach. Notably, it offers case studies of families who have not typically been subject to historiographical attention: military families facing reunification; families of 'gifted' or 'autistic' children; mothers in Ireland and Spain; and 'foreign' families settling in West Germany. The articles look to analyse the agency of these families from their own perspectives. For Jennifer Crane and Lauren Stokes, family perspectives are revealed through analysis of voluntary papers, diaries and letters. Articles by Grace Huxford and Elisa Chuliá make use of rich oral histories to uncover and explore family views. Other authors, however, given the sources available, must read 'against the grain' of professional and expert documents to consider the spaces within which families could act, resist or conform. Lindsey Earner-Bryne, for example, carefully pieces together limited newspaper coverage of unmarried Irish mothers, seeking to restore their dignity. Jonathyne Briggs reflects on contemporary criticism by autistic people of historic campaigning, enabling him to critically assess the nature of family dynamics in late twentieth-century France. Furthermore, analysis of the family as an agent of change can disrupt our existing thinking about the overarching chronologies of the mid-to-late twentieth century. A classic chronology of this period in Western Europe has been developed by Paul Ginsborg, who suggests that there was a 'major expansion' of welfare services from 1945, further attention paid to issues of social and gender inequality in the family from 1960 to 1974, and policy fixation on falling birth rates and single-parent families after 1985 as well as 'major battles over efficiency, privatisation and residualisation' for the 1980s and 1990s. 28 Further historiographies place heavy framing on the 1960s as a moment of sexual revolution and the 1970s as a feminist revival. 29 As Ginsborg acknowledges, however, these types of broad chronology require 'heavy qualifications', as distinct regional models have emerged. 30 In particular, Ginsborg highlights: the Scandinavian focus on universalism; Britain and Ireland's focus on lowincome families, a 'corporatist continental model which traces its origins to a Bismarckian tradition'; and the Southern European focus on combining universal health care with 'a deep-rooted clientelist tradition'. 31 This model, furthermore, does not account for changes in Eastern Europe, which must be viewed as entwined with, shaping and shaped by those in 'the West'. By taking the family as the central agent of change, this collection offers new, and more complex, chronologies and boundaries. First, looking through this narrow lens of the family reveals the kinds of complex partnerships that were forged around 'family issues' and 'family life', which crosscut any unitary narratives of power. Lauren Stokes's article, for example, shows how, in West Germany in the 1970s, debate about family migration in the context of welfare opened up the 'possibility for radical alliances between ordoliberals and neoliberals who prefer a limited welfare state and conservative Catholic thinkers'. This kind of uneasy coalition, mobilising around one key issue, breaks down any idea of the 1970s as simply a moment of a liberal 'new' society, or as a regressive 'backlash' against the 1960s revolution. Looking through the lens of the family also breaks down the presumed historiographical separation between 'East' and 'West' Europe in the mid-to-late twentieth century, building on Margaret E. Peacock's comparison of childhood in the Soviet Union and the United States, and a recent study of 'authenticity' collated by Joachim C. Häberlen, Mark Keck-Szajbel and Kate Mahoney. 32 Grace Huxford and Jennifer Crane's articles both show how families were seen in policy as a bastion in building links between East and West, while families themselves often rejected this apparent distinction in 'ways of life'. Lindsey Earner-Bryne argues that the purported liberal values of 'the West' did not extend to 1960s Ireland, where the cultural significance of religion created significant shaming and stigma around birth control and family planning. Despite an apparent 'sexual revolution', women across East and West Europe alike at times continued to have low levels of knowledge about family planning and were not always comfortable discussing it. 33 Approaching the family as an agent, rather than a subject or recipient, of state systems also helps us to think broadly about how families' everyday lives comprise forms of activism. In this special issue, we see a range of voluntary associations forged by families: loose and generational social movements relating to permissiveness; charities and non-governmental organisations providing services neglected by the state, whether due to war or moral and religious politics; and small parents' groups, which took one-off or sustained legal and media action. These types of overt action position the late twentieth century as a moment in which small groupsoften acting as or for familieswere able to attract media and public policy attention to a new degree, effectively utilising their experiential and professional expertise, and mobilising new forms of media and communication technology to demand and create change. These organisations used ideas of 'the family' in ways that could represent a traditional notion of the family that did not recognise childhood voices or experience. This point is central to Jonathyne Briggs' article about autistic families in contemporary France, where he shows that parents came to represent the 'autistic family' instead of autistic young people themselves. This special issue also shows how the inaction of families, or their refusal to comply with state systems, could also drive significant change. For instance, Lauren Stokes's article discusses how 'foreign' couples, invited to West Germany to solve perceived issues from employment to sexual relations, could undermine the guest workers' programme simply by bringing their children with theman action which the programme had failed to account or plan for. 34 Elisa Chuliá's interviews with Spanish women demonstrate that while, for some, education was a 'space for opposition to the dictatorship' of General Franco, other women in hindsight consider themselves 'dumb, blind and deaf' to the political changes of this moment. Women also drove change by covertly gaining hold of contraceptives, the prescription, sale, or advertisement of which were banned in Spain until 1978. Families were therefore agents of change, but that change was at times demanded and led consciously over this period, through longer-standing forms of activism, whereas at other times it was brought about by new moves towards individualism and privacy, driving and reflecting internalist models of family life. 35 Hence, looking at families in this way, we begin to see how they are leaders of change as well as followers, whether driving change through inactivity, demographic shifts, everyday life or conscious efforts at resistance and activism. As Elisa Chuliá argues, the very creation of one's own family can be a 'personal project'. While illuminating changes in national and family lives, this collection positions these as entwined, not separate. Indeed, the themes of gender and nationhood emerge as key throughout this special issue in crosscutting the politics of family and state. In terms of gender, the collection shows the consistency across national and chronological borders, from families, policy makers and voluntary experts alike, with which fathers were seen as key for the 'appropriate' psychological development of children, the maintenance of family life and the priority figure for family reunification. Mothers, meanwhile, were heralded as producing future leaders, but typically only supported if reproducing within moral frameworks such as the two-parent patriarchal family. Mothers were also blamed for conditions such as autism, as Jonathyne Briggs shows, or praised for raising 'gifted children', as Jennifer Crane discusses. The parenting of mothers rather than that of the father or other childcare providers was centrally analysed and criticised in all case studies. 36 Indeed, multiple case studies in this collection demonstrate that, despite the purported rise of feminism, women's reproductive labour was built into, and remained integral within, the various 'welfare states' of Britain, Ireland, and West Germany alike over the late twentieth century. Rigidly hierarchical and patriarchal gendered ideas were also used by states to place new demands on family life, displacing identities of 'mother' or 'father' with illegal migrant or tax-dodging citizen, or, indeed, soldier. The special issue finds cases where families were able to resist gendered assumptions. The oral history studies in the collection, in particular, make clear the variety of strategies of communication and shared work within all family structures. Nonetheless, gendered assumptions remained relatively rigid, powerful and present across Europe and throughout the mid-to-late twentieth century, particularly when emboldened by religious cultural beliefs. Families could also use their agency to further ingrain, as well as to challenge and discredit, existing structural dynamics of power, further adding to our understandings of 'family' as a destructive, as well as supportive, social space. Ideas of the nation also crosscut state and family politics throughout this collection, which chronicles decades in which national conflicts and new transnational alliances were forged, alongside the rise of nationalist-populist movements and a growing sense of national exceptionalism in many countries. 37 Articles by Grace Huxford and Lauren Stokes analyse family reunification, which was premised on, and governed by, state-led expectations of national divides, whereby certain families, deemed 'foreign' or members of the military, were excluded or included in national configurations. Family reunification schemes, seen in this special issue, were intended to solve perceived local issues of, for example, men's sexual aggression or to offer specific exemplars of family life. Yet the invocation of 'Europe' also became a powerful one over this period, with politicians and press alike critically framing their countries as 'the most backward nation in Europe', for example. 38 Families in this collection, and across borders of place and time, were at times affected by, ignorant of, or resistant to such national and transnational visions. Families with a so-called 'gifted' child, for example, as studied in Jennifer Crane's article, were called upon by the conservative press hoping to mobilise these children to reverse economic decline yet, themselves, often worked in co-operation with European voluntary associations looking to share knowledge and resources around children's intellectual resources. Looking to the family, furthermore, shows how ideas of nation were broken down by distinct regional identities and stereotypes operating across, for example, Spain and Italy. Thus, the comparative approach of this collection enables us to consider how nations have been constructed and formed as well as their relationships with shifting family life, across different geographical contexts and in a rapidly changing period of history. Existing historiography has positioned the family as assumed yet also absent within European politics over this period. residency, family stability and entitlement. 39 Nonetheless, Ginsborg has argued that the European Union has rarely 'elaborated a vision of family politics', but rather that policies relating to the family spun out indirectly from other social policy questions. 40 This collection, by taking an expansive view of politics and policy makers, shows how a range of voluntary and professional experts hoped that families would be ambassadors in a new Europe, displaying the 'moral superiority' or 'value' of liberal democratic sentiments as ambassadors and icons, or driving economic prosperity for imagined futures. British military families, policy makers hoped, would display the failures of social policy in Soviet-occupied East Germany and Eastern Europe. Meanwhile, psychologists in Britain planned to draw on lessons from these states to guide how best to identify and mobilise their so-called 'gifted' youth. --- Concluding Remarks In this way, this special issue focuses historians' attention on the European family, in varied forms, but also contends that ideas of family, democracy, welfare, professions and civil society must be seen in constant conversation. By looking at these areas in tandem, with the family as a key lens, this special issue looks at the very nature of power itself in a critical period of modern history, and how it is exchanged, realised, lived and enacted between nation-states, voluntary organisations and family units. The special issue raises key historiographical and contemporary questions. When do professions think that families need 'support' and when do they need 'independence' and 'autonomy'? Who defines family rights, and which social groups are included and excluded from discussion of 'needs', rights and responsibilities? Which organisations provide for families over time, and what are their ethical assumptions, priorities and foundations? What is the changing role of the state? Do shifts towards 'informal' family structures place increasing burdens on state welfare, and how do states respond, in moments of fiscal austerity, to ideological change or total war? The collection develops an emergent strand of historiography which aims to examine the family simultaneously as an agent, not a subject or object, of change within national imaginaries and in everyday life. It looks across Europe, specifically with case studies of Britain, Ireland, France, Spain, East and West Germany. This is, of course, far from a 'complete' portrait of all families across all of Europe. Nonetheless, these specific case studies can offer a new analysis of liberal democratic, fascistic and authoritarian states. They can provide some assessment, also, of how religion, generation and national histories complicated the politics of ongoing government regimes and their effects on family life. These case studies also enable us to better understand how European internationalism co-existed with new strands of nationalism over the mid-to-late twentieth century. Far more significant work must continue to be done in this area, and this special issue is intended as a contribution to an ongoing and new conversation, with the hope that rich new case studies from new research will continue, in subsequent years, to challenge and nuance its findings. In framing its analysis around the 'mid-to-late twentieth century', this collection emphasises that the post-war period did not appear without precedent. Rather, the changes of this period were gradual, with a continuous history dating back towards the early twentieth century. Military families, as Grace Huxford's article suggests, experienced 'the cycle of reunion and separation' long before 1945, and military memories of the 'poor reception' for soldiers after the First World War fuelled anxieties following the Second. As Elisa Chuliá explains, legal restrictions on Spanish women's lives, likewise, preceded the entwined work of Franco and the Catholic Church and, especially, the male dominance in household finances had been included in civil legislation since at least 1889. Lindsey Earner-Bryne's article, similarly, shows how significant events from the 1930snotably papal warnings about the familycontinued to exert influence over debates into the late twentieth century. As Elisa Chuliá points out, such long-term analysis is especially significant for considering families: family dynamics, 39 Stokes, 'An Invasion of Guest Worker Children', 374-5. 40 Ginsborg, 'The Politics of the Family', 444. in particular, were shaped by intergenerational discussion, and social and class inheritance. In family life, also, the influence of 'war' has persisted long beyond the 'wartime' period and into 'peacetime', while the influence of sudden 'permissiveness', for example, has often not had an impact for many decades. Nonetheless, certain features of the 'mid-to-late twentieth century' were distinct, and make it particularly ripe for analysis with regard to ideas of family, democracy and welfare. The mid-to-late twentieth century marked an unprecedented era of family separation through war, migration, evacuation, incarceration and genocide. For some families, this period also marked new periods of reunion. Because of this, families were constantly being mobilised and rethought, internally and externally, as political and social units. In this time, policy makers analysed throughout this collection pinned new hopes on subsequent generations of children to embody a distinctly 'new' or 'modern' sense of selfhood or leadership. Families were also continually renegotiating and redefining their own family systems, obligations and lives. In the context of family reunification for soldiers, Grace Huxford argues, they were 'learning familial dynamics anew'. Families had new opportunities for collective action together. Women and young people, also, had new opportunities to make visible the ways in which 'families' had wrought injustices upon their lives. This collection proceeds with a broadly chronological structure, spanning different national case studies. As an opening and making use of oral history testimony, Elisa Chuliá's article analyses the everyday resistance of Spanish families in Francoist Spain. Chuliá argues that families were spaces of nurture and care, and also of resistance: mothers, in particular, used systems of education to empower their daughters for 'the future'. The second article is Grace Huxford's analysis of British military families in Cold War Germany. Focusing again on the 1950s, this piece highlights the ways in which these families resisted their projected roles as diplomats or models for a 'British way of life', mixing in complex and varied ways with new local communities. As such, the opening articles grapple with the aftermaths for family life and civil society following the Second World War. While war remains a significant spectre in Chuliá and Huxford's chapters, having increased the role of the state in family life and civil development, the chapters also show the role of family resistance to new professions and voluntary groups, organised around developing frameworks of nation such as national-populism and the East-West divide. In the third article, Lauren Stokes considers emergent concerns in West Germany in the 1970s that a Spanish grandparent may not be a 'functional' and thus 'legitimate' family member. Spanish families in Stokes's account demonstrated 'agency' through voluntary action, although their actions were limited and oppressed by the political, legal and cultural systems in which they lived. Jennifer Crane, meanwhile, considers the ways in which families of so-called intellectually 'gifted' children in Britain were able to resist intervention from national conservative press and European voluntary agencies, particularly from the 1980s through the formation of activist groups and in everyday lifealthough activisms remained shaped by broad visions of what precocity and childhood should mean. In these types of activism from the 1970s and 1980s, consumerist models of living across Western Europe led to new economic and neoliberal definitions of the family, but families simultaneously used this new interest in individual rights to reassert the significance of their own models of happiness and well-being. In the next article, Jonathyne Briggs traces the changing activism of families with autistic children in contemporary France. He pinpoints a shift, from the 1960s until the 1990s, in the nature of this activism, with parents moving from resistance to professionals towards collaboration. For Briggs, this case study reveals the power of 'the family' as an 'agent', but also the fragmented nature of its internal agency, as the preferences of parents have consistently overridden those of the young. In the final article, Lindsey Earner-Bryne analyses agency in the Irish family. She argues that until the 1980s the Irish family was 'still regarded as vital in shoring up the moral and social status quo'. From this decade, she states, 'the age of the inquiry', the family was made 'culpable for the moral cowardice of the nation', with unmarried mothers in particular a target of sustained professional violence. Overall then, while looking across Europe, at diverse national contexts across a broad scope of time, this special issue teases out the contested and changing relationships between family, democracy and welfare in the mid-to-late twentieth century. Notably, it shows that families were not only objects of state governance and welfare, organised 'from above' in authoritarian, fascist and democratic states alike; rather, families renegotiated, reformed and reshaped the running of nation-states, their provision of welfare and their visions of equality and governance. Recognition of the family as an agent of change must be made central to our historical work. The growing power of the family tells new stories about the shifting power of social collectives, the changing cultural norms of generations and the relationships between economic and political structure, and individual life. These stories are, we hope, of interest to readers of Contemporary European History. Cite this article: Crane J . Agents of Change? Families, Welfare and Democracy in Mid-to-Late Twentieth-Century Europe. Contemporary European History 32, 173-185. https://doi.org/10.1017/S0960777323000152

Families have always been vulnerable. They have long been torn apart by the mass migrations of warfare, the oppression of minority groups, the closure of international borders and the refugee crises governed 'from above'. Families have also always been powerful symbols. Nationalist-populist movements have capitalised on fears about familial decline and liberal democracies have built moralistic views of the family into their welfare systems. Yet, this special issue aims to demonstrate that families have not merely been objects or subjects buffeted by political and social change. Rather, families have also consistently acted as 'agents of change'. This is not to valorise the familyfamilies have been patriarchal, damaging and oppressive as well as supportive, empowering and caring. However, this is to say that historical work must take 'the family' seriously as an active participant in shaping historical change.

Introduction Over 6 million deaths attributable to COVID-19 have been reported globally since the onset of the pandemic in early 2020. 1 Beyond this devastating death toll, there is increasing recognition of the widespread disruption of the pandemic on healthcare services, with a far-reaching impact on the care of non-COVID-19 conditions. 2 Excess mortality has been proposed as a key indicator that captures both deaths caused by COVID-19 and indirect deaths attributed to the pandemic more broadly due to interruption in routine care of chronic conditions. 3,4 Notably, many low-and middle-income countries , particularly in Latin America, were vulnerable to the direct and indirect effects of the COVID-19 pandemic due to chronic underinvestment in public healthcare. 5 Though several reports have estimated that rates of excess mortality were disproportionately higher in LMICs Key Messages • Mexico experienced one of the highest rates of excess mortality in Latin America following the onset of the COVID-19 pandemic but the extent to which non-COVID-19 deaths contributed to excess mortality has not yet been characterized. • We conducted a retrospective, municipal and individual-level study using data from 1 069 174 death certificates to estimate mortality related to COVID-19 and to non-COVID-19 causes using ICD-10 codes in 2020 compared with 2015-2019. • There was a 51% higher mortality rate in 2020 compared with the 2015-2019 average, largely attributable to COVID-19 deaths , which occurred primarily in-hospital; conversely, one-fifth of excess deaths in Mexico in 2020 were attributable to non-COVID-19 causes, largely cardiometabolic conditions, which occurred primarily out-ofhospital. • Southern regions and marginalized communities in Mexico carried a disproportionate burden of excess mortality; municipal-level correlates of these excess deaths included lower healthcare coverage, whereas individual-level factors that correlated with non-COVID-19 mortality included lower educational attainment, blue-collar employment and lack of medical care assistance prior to death. • Excess mortality in Mexico in 2020 was attributed to both COVID-19 and non-COVID-19 causes, likely reflecting a complex interplay between a fragmented and under-resourced health system, strained hospital capacity and sociodemographic inequalities further unmasked by the pandemic. following the onset of the COVID-19 pandemic, there is limited insight regarding non-COVID-19 deaths and their contribution to the reported rates of excess mortality in Latin America. 6 Mexico is of particular interest given that it ranks as one of the countries with the highest rates of excess mortality in the Latin American region following onset of the COVID-19 pandemic. 7 A confluence of health and sociodemographic inequalities that pre-dated the COVID-19 pandemic, a high burden of chronic cardiometabolic conditions and a fragmented healthcare system all contributed to a high and disproportionate burden of excess mortality among marginalized communities. 5,8 A descriptive assessment performed in Mexico showed that chronic cardiometabolic conditions, which are highly prevalent among communities of low socio-economic status, were the main causes of death independently of registered COVID-19 deaths in Mexico during 2020. 9 However, whether hospital saturation had ripple effects on out-of-hospital excess mortality, particularly for highly prevalent chronic health conditions across different vulnerable regions, has not yet been characterized. Hence, there is a need to comprehensively assess the extent to which individual-and municipalwide-level socio-demographic inequalities impacted excess mortality to further guide health policies to strengthen existing systems and mitigate ongoing health disparities. In this study, we sought to: estimate the age-adjusted rates of cause-specific excess mortality due to COVID-19 and non-COVID-19 deaths in 2020 compared with the 2015-2019 period, stratified by in-hospital and out-ofhospital setting; evaluate the geographical distribution of cause-specific excess mortality in Mexico in 2020; and characterize the association between municipal-and individual-level socio-demographic inequality measures and non-COVID-19-related excess mortality. --- Methods --- Study design and data sources Based on the work by Lima et al., we conducted a retrospective municipal-and individual-level study using national mortality records from 2015-2019 compared with 2020. 6 Death certificate records of individuals living in Mexico were collected by the National Institute of Statistics and Geography . Briefly, INEGI generates annual mortality statistics from death certificates and vital sociodemographic characteristics issued by the Ministry of Health, which includes the primary cause of death in accordance with the tenth version of the International Statistical Classification of Diseases and Related Health Problems . 10 Complete methodology of the death certification process, validation and collected variables are available in the Supplementary material . --- Variables and definitions --- Outcome variables Cause-specific excess mortality was centred on two primary outcomes: deaths due to COVID-19 and deaths related to non-COVID-19 causes. Overall excess mortality was the sum of excess mortality due to non-COVID-19 causes and all registered COVID-19 deaths. COVID-19 deaths. Deaths attributable to COVID-19 were defined based on the following ICD-10 codes: U071 , U072 and deaths after April 2020 classified as J00-J99 . This aggregation of COVID-19 deaths considers inadequate registration of COVID-19 cases across 2020 as there are an unknown number of deaths that could have been classified with unspecified respiratory diseases in the early stages of the pandemic due to limited SARS-CoV-2 testing capacity in Mexico. 11 Non-COVID-19 cause-specific mortality. All other causes of death were classified as non-COVID-19-related deaths and were coded using the 2020 Mexican list of mortality, which includes 436 specific causes of death. 12 To simplify result presentation, we only display the first 10 causespecific deaths in the main results, with the full list provided in the Supplementary material . --- Excess mortality estimation According to the approach proposed by Karlinsky and Kobak, we estimated excess mortality as the difference between average deaths during the 2015-2019 period compared with deaths registered during 2020. 13 We used average deaths for two reasons: the use of average deaths is a simple approach shown to be a reliable assessment based on sensitivity analyses estimations; and given that we are estimating 436 specific causes of death, predictive methods based on generalized linear models may overestimate the standard error for low-frequency causes. 9 Excess deaths were standardized to age-adjusted rates per 100 000 population with age structures by state, municipalities, and regions per 5-year increments using population projections provided by the National Population Council . Percent increase in 2020 compared with 2015-2019 was also used as a proxy of excess mortality. --- Stratification by setting of death We hypothesized that the COVID-19 pandemic posed a significant burden on in-hospital care, which may have influenced increases in excess mortality, particularly for non-COVID-19-related deaths. To evaluate this hypothesis, we stratified excess mortality according to whether the death occurred out-of-hospital or in-hospital, as registered on death certificates. Out-of-hospital deaths were defined accordingly if the death was not registered in a hospital setting or if they were coded as occurring at the deceased person's home or elsewhere . Deaths with an unspecified setting were excluded across all the analysis. --- Marginalization index To quantify the impact of municipal socio-demographic inequalities on excess mortality, we used the 2020 municipal social lag index estimated by the National Council for Evaluation of Social Development Policy . 14 Since we intended to evaluate social inequalities independently from urbanization and centralized health services, we used residuals of linearly regressed mean urban population density and hospital beds per 100 000 inhabitants using data extracted from CONEVAL to fit an adjusted municipal SLI . We then categorized municipalities into four aSLI categories based on the Dalenius & Hodges method . --- Municipal-level correlates of excess mortality We included the percentage of the population without healthcare coverage and the hospital occupancy due to COVID-19 inpatients as municipal-level factors related to excess mortality. Healthcare coverage was obtained from 2020 CONEVAL estimations. To estimate a surrogate of hospital occupancy, we used the number of hospitalizations with confirmed COVID-19 from the National Epidemiological Surveillance System data set collected by the General Directorate of Epidemiology of the Mexican Ministry of Health, which includes reports of daily updated suspected COVID-19 cases. 15 Complete methodology, the protocol of testing and the variables included are available in the Supplementary material . --- Statistical analysis To visualize differences in deaths over time in 2020 compared with the 2015-2019 period, we first plotted excess mortality per 100 000 inhabitants by month of occurrence, stratified by COVID-19 and non-COVID-19 causes. We then disaggregated excess mortality rates due to COVID-19 or non-COVID-19 causes by state and municipality. Next, to visualize whether the proportion of age-adjusted excess mortality in each municipality increased due to COVID-19 or non-COVID-19 causes, we used choropleth maps classified using the quantile method with the biscale R package. We further visualized the relationship between excess mortality and aSLI using the same method. The median value for the estimated age-adjusted excess mortality and the aSLI were considered as the cut-off threshold. --- Municipal-level factors associated with excess mortality Next, we evaluated the impact of municipal characteristics on increased risk of age-adjusted excess mortality using negative binomial regression models to obtain incidence rate ratios . Models were adjusted for municipal male-to-female death ratio, education percentage, access to medical assistance and urbanization . We also calculated the ratio of out-of-hospital to in-hospital deaths, which was also adjusted for the above outlined covariates. 16 IRRs were plotted using the jtools R Package. --- Individual-level factors related to non-COVID-19 mortality To identify individual-level factors associated with the probability of death attributable to non-COVID-19 causes as compared with COVID-19, we fitted hierarchical random-effects logistic regression models, which included individual-and municipal-level variables . Individual-level variables included sex, education, selfreported indigenous identity, work occupation, access to medical assistance prior to death and social security coverage. We perform a municipal-level adjustment that included living in municipalities with low hospital bed occupancy and municipal aSLI categories. For this model, we used the municipality of death occurrence as a random intercept to account for inter-municipal variability in death registration in the model and to establish a hierarchical relationship between individual-and municipal-level variables. All analyses were performed using R software . --- Results --- Overall and cause-specific excess mortality in Mexico during 2020 We identified 1 069 174 deaths in Mexico during 2020 compared with 686 567 average deaths in 2015-2019. We estimated an age-adjusted mortality rate of 833.5 deaths per 100 000 inhabitants for 2020, with an estimated ageadjusted excess mortality of 282.41 deaths per 100 000 inhabitants; this represents a 51% increase in mortality compared with the average age-adjusted mortality rates in 2015-2019 . Peak excess mortality during 2020 was observed during the May-to-June period. Approximately 76.1% of excess deaths were attributable to confirmed or suspected COVID-19, whereas 23.9% were attributable to non-COVID-19 causes. The main contributors of excess mortality were suspected or confirmed COVID-19 deaths . The five leading causes of non-COVID-19 excess mortality were acute myocardial infarction , type 2 diabetes , violent assaults , hypertensive heart disease and essential arterial hypertension . All excess deaths were recorded after April 2020, with a steep increase after this period for COVID-19, acute myocardial infarction and type 2 diabetes-related deaths . --- Excess mortality according to in-hospital vs outof-hospital death When stratified by the setting of death, we estimated an inhospital excess mortality rate of 112.47 deaths per 100 000 inhabitants and an out-of-hospital excess mortality rate of 162.54 deaths per 100 000 inhabitants; this represents an increase of 45.4% and 55.5% of in-hospital and out-of-hospital deaths, respectively, compared with the average of 2015-2019. When stratified by the specific cause of death, we observed that excess in-hospital mortality rates were primarily attributable to COVID-19 deaths, whereas there was a decrease for in-hospital non-COVID-19-related deaths after March 2020. An estimated 80.96% of all out-of-hospital excess mortality was attributable to non-COVID-19 causes, whereas only 19.03% were attributable to COVID-19 deaths. Excess deaths attributable to COVID-19 occurred predominantly in the in-hospital setting, whereas most non-COVID-19 deaths occurred largely out-of-hospital . Among the 10 leading causes of excess mortality, acute myocardial infarction and type 2 diabetes decreased in-hospital but increased out-ofhospital after April 2020 . S1, available as Supplementary data at IJE online). Further stratification revealed a geographical aggregation of non-COVID-19 deaths caused by acute myocardial infarction, type 2 diabetes, essential arterial hypertension and unspecified strokes clustered in the southern states of Mexico . We also evaluated age-adjusted excess mortality at the municipal level to obtain a more detailed overview of these geographical differences. Excess mortality had a heterogenous geographical distribution and correlated with the SLI in municipalities in Mexico with higher excess mortality due to both COVID-19 and non-COVID-19 causes . At the state level, the highest decrease in non-COVID-19 in-hospital deaths was seen in Oaxaca, Yucatan and Veracruz, whereas the highest proportion of non-COVID-19 out-of-hospital deaths were observed in Tlaxcala, Yucatan and Colima . --- Municipal-level impact of socio-demographic inequalities in excess mortality We observed marked geographic variability in ageadjusted excess mortality across municipalities with higher aSLI . After excluding COVID-19-related deaths, only the southern municipalities displayed the highest combination of excess mortality and aSLI . Stratifying by aSLI categories, age-adjusted mortality rates and excess mortality showed a stepwise increase with each higher marginalization level . Municipalities with very high aSLI displayed both the higher age-adjusted mortality and excess mortality rates in Mexico. --- Municipal-level correlates of excess mortality To evaluate the hypothesis that excess mortality was correlated with social inequalities in healthcare access and hospital occupancy due to COVID-19 at the municipal level, we fitted negative binomial regression models for ageadjusted excess mortality rates. As observed in the geographic distribution of age-adjusted excess mortality , municipalities at high and very high social lag had the highest risk for non-COVID-19 age-adjusted excess mortality in 2020. Municipalities with a higher percentage of the population without social security coverage , higher COVID-19 hospital occupancy and higher social lag categories were at higher risk of excess mortality after adjusting for covariates . We observed an interaction effect for higher risk of non-COVID-19 age-adjusted excess mortality in municipalities with very high social lag and higher COVID-19 hospital occupancy . --- Individual-factor correlates of non-COVID-19 deaths Finally, we explored the role of socio-demographic conditions and healthcare-related inequalities for the risk of non-COVID-19 deaths using random-effects logistic regression models. We observed that women, people who had lower educational attainment and those who worked as craftsmen, farmers, labourers or were unemployed had an increased odds for death attributable to non-COVID-19 compared with COVID-19 causes. Regarding healthcarerelated factors, people without medical assistance before death, people who reported public or unspecified social security coverage or people who lived in municipalities with low availability of hospital beds had an increased odds of death from non-COVID-19 compared with COVID-19 causes. Finally, people living in municipalities with high and very high social lag had the highest odds of non-COVID-19 compared with COVID-19 death . --- Discussion In this study of 1 069 174 deaths recorded in Mexico between 2015 and 2020, we report that 51% of deaths in 2020 were in excess compared with the average reported between 2015 and 2019. Although cause-specific excess mortality during 2020 was largely attributable to COVID-19 , non-COVID-19 causes comprised up to one-fifth of excess deaths in Mexico during 2020. Moreover, we report a differential impact on excess mortality related to the setting in which the deaths occurred; whereas COVID-19 deaths occurred primarily in-hospital, non-COVID-19 deaths sharply decreased in this setting and had a concurrent increase in the out-of-hospital setting. These findings contribute to the growing literature on the far-reaching impact of the COVID-19 pandemic on the health system and suggest both an excess in non-COVID-19 mortality as well as a displacement of these deaths to the out-of-hospital setting in Mexico. We also observed that excess mortality exhibited marked geographical heterogeneity, which was associated with higher social lag; states in the southern region of Mexico had the highest social marginalization and similarly high rates of non-COVID-19 excess mortality. We showed that lower prevalence of population without social security coverage and higher rates of COVID-19 hospitalization, combined with social marginalization, were municipal-level correlates of non-COVID-19 excess mortality. Finally, at the individual level, lower educational attainment, blue-collar employment , unemployment and lack of medical assistance before death were significant correlates of non-COVID-19 compared with COVID-19 mortality during 2020. These findings suggest that excess mortality from non-COVID-19-related causes, which occurred disproportionately out-of-hospital and among populations with social disadvantage, may reflect a complex interplay between a fragmented health system, strained hospital capacity, interruptions in chronic disease care and sociodemographic inequalities further unmasked by the pandemic. 5 This situation is applicable to Mexico, but also to countries with similar socio-demographic profiles in the region or with high rates of SARS-CoV-2 infections. Previous reports have documented the high burden of excess mortality caused by the COVID-19 pandemic in Mexico, with excess mortality rates being estimated from 26.1 to 36.0 deaths per 100 000 inhabitants; 6,13,[17][18][19] moreover, Karlinsky et al. projected that Mexico's actual toll of deaths could be twice the number of deaths registered during 2020. 14 These reports positioned Mexico as one of the leading countries in terms of excess mortality in Latin America and worldwide. 6,13 However, there is limited information regarding cause-specific contributors to global excess mortality rates in Mexico. Besides contributing to the literature on COVID-19 excess mortality in Mexico, our findings also expand this literature by showing that excess deaths were also related to cardiometabolic chronic health conditions, including type 2 diabetes, cardiovascular disease, arterial hypertension and obesity, which had a steep increase in the out-of-hospital setting. 9,20 Excess non-COVID-19 deaths could be attributable to hospital reconversion policies and healthcare restructuration designed to improve care for COVID-19 cases, which may have reduced access to care for people with chronic health conditions who required continuous medical assistance during the COVID-19 pandemic. Other high-income countries have documented the association between hospital occupancy and excess mortality during periods of peak COVID-19 infections. [21][22][23] Explanations related to this phenomenon rely on data on restricted access to healthcare services in places that experienced hospital overload due to COVID-19, reduced out-of-hospital attention due to severely restricted healthcare services and personnel availability, lower insurance coverage and a lower number of healthcare personnel per capita. [24][25][26] Other reported non-related healthcare contributors were social stigma for being treated in hospitals due to potentially acquiring COVID-19 infection and reduced physical activity due to pandemic restrictions on mobility, which could have exacerbated complications due to chronic health conditions. 27 Overall, excess non-COVID-19 mortality could be interpreted as an indirect proxy of the negative effects attributable to healthcare policies that prioritized in-hospital COVID-19 attention over the care of other chronic health conditions. Notably, increased rates of COVID-19 hospitalizations and mortality in Mexico were observed in municipalities with high marginalization independently of urbanization, likely attributed to increased stress on their healthcare systems. 28,29 This phenomenon may explain the disproportionate burden of non-COVID-19 deaths among marginalized municipalities in Mexico. Our results support the hypothesis that populations with social disadvantage experienced the highest impact of excess mortality attributable to hospital saturation, with this impact having an unequal distribution within Mexico. In the European region, countries with high excess mortality, such as Bulgaria, Russia and Serbia, were impacted by diverse social barriers, such as difficulties in fully adhering to social isolation policies. 30,31 Two recent reports in England showed that communities with a high density of care homes, with a high proportion of residents on income support, overcrowding conditions and ethnic minorities were at higher risk of excess mortality and years of life lost due to the COVID-19 pandemic. 32,33 In Latin American countries, socio-economic inequalities widened due to difficulty adhering to lockdown mandates given low stipend support, a high proportion of their population working in the informal economy and lack of access to healthcare, even among healthcare personnel. 3,30,34 Nevertheless, this evidence and the comparison between countries should be interpreted with caution given the variation in COVID-19 dynamics, within-country gradients of socio-demographic inequalities and different epidemiological profiles of high-risk comorbidities. Our results highlight the impact of healthcare-related and individual-level social inequalities in exacerbating overall and cause-specific excess mortality in Mexico. We show that the main contributor to higher non-COVID-19 excess mortality rates at the municipal level was a lower percentage of the population with access to social security health coverage; in Mexico, social security providers condition the type of healthcare access by individuals, which likely also influences received quality of care and healthcare access. Furthermore, the interaction between a lower percentage of the population with access to social security health coverage and social marginalization confirmed the hazardous interplay between social and healthcare inequalities. 35 At the individual level, we showed that certain socially vulnerable occupations experienced unequal risks for non-COVID-19 mortality. The role of individual and socio-demographic determinants in the risk for adverse COVID-19 outcomes has been previously reported. 36,37 Nevertheless, our findings demonstrate that sociodemographic inequalities impacted individuals with preventable chronic conditions, regardless of public healthcare policies aimed at mitigating the impact of the COVID-19 pandemic in healthcare infrastructure and provision. Overall, our results represent an urgent call to action for local authorities to perform a healthcare restructuration, particularly in marginalized municipalities and with special attention to vulnerable populations to prioritize full coverage of hospital bed capacity, well-trained healthcare personnel and availability of primary care services that cover the management of chronic health conditions. These policies could prevent associated complications in the context of future COVID-19 waves or other circumstances that increase stress and reduce access to healthcare in Mexico and other LMICs. Our study has some strengths and limitations. Among the strengths, we highlight the use of 1 069 174 nationwide mortality registries to compare all-cause and cause-specific excess mortality during the COVID-19 pandemic in Mexico in 2020. This approach allowed us to estimate with higher confidence state-and municipal-level excess mortality rates that helped us to study the regional impact of the COVID-19 pandemic and identify vulnerable zones in Mexico that were especially affected during 2020 compared with previous years. Additionally, the use of sociodemographic variables at different levels gave us insights to evaluate municipal-and individual-level correlates of excess non-COVID-19 mortality. Nevertheless, limitations to be acknowledged include the lack of specific clinical information and co-morbidity assessment for correlates known to be key determinants of higher risk of death from COVID-19 and non-COVID-19 causes, particularly regarding management of chronic cardiometabolic conditions. Second, we could not ascertain the number of non-COVID-19 deaths that occurred due to exacerbation of underlying chronic conditions by current or previous SARS-CoV-2 infection, which could increase the risk of long-term complications, including cardiovascular diseases. 3 Third, our COVID-19 death construct included cases that could have been misclassified as atypical pneumonia or severe acute respiratory infections of unknown aetiology, registered after the onset of the COVID-19 pandemic; this was done to reduce the risk of under-reporting or misclassified COVID-19 deaths, but could have led to overestimation of COVID-19 deaths. Fourth, we used a surveillance data set to assess COVID-19 hospitalization as a proxy for hospital occupancy; however, identification of COVID-19-related hospitalizations may have varied according to weekly SARS-CoV-2 testing capacity and adequate reporting. Therefore, the use of this proxy could be biased in municipalities with higher marginalization and reduced access to testing. Finally, our municipal-level factors should be interpreted as structural conditions that displayed an association with higher excess mortality rates and therefore we should avoid an ecological fallacy in determining personal actions in clinical and healthcare management during the COVID-19 pandemic. In conclusion, we show a high burden of excess mortality in Mexico in 2020, largely attributable to in-hospital COVID-19 and out-of-hospital non-COVID-19 deaths. We observed regional heterogeneity of non-COVID-19 excess mortality, with a disproportionate burden on marginalized municipalities in southern Mexico. High hospital occupancy due to COVID-19 and a higher percentage of the population without social security coverage were municipal-wide-level correlates of excess mortality, whereas individual-level lower educational attainment, vulnerable working occupations, lack of medical assistance before death and public or underspecified healthcare access were factors related to higher non-COVID-19 mortality likelihood. Our findings underscore the impact of sociodemographic inequalities on excess mortality related to non-COVID-19 causes in Mexico during 2020 compared with the 2015-2019 period. These results should prompt an urgent call to action to improve healthcare coverage and access, particularly in primary care settings and among populations with social disadvantage. Such policies could reduce health disparities in Mexico in circumstances that increase the stress of healthcare systems, including the ongoing COVID-19 pandemic and beyond. --- Data availability All code, data sets and materials are available for reproducibility of results at https://github.com/oyaxbell/excess_non_covid/. --- Ethics approval This project was registered and approved by the Ethics and Research Committee at Instituto Nacional de Geriatr ıa, project number DI-PI-006/2020. --- Supplementary data Supplementary data are available at IJE online. --- ---

Background: In 2020, Mexico experienced one of the highest rates of excess mortality globally. However, the extent of non-COVID deaths on excess mortality, its regional distribution and the association between socio-demographic inequalities have not been characterized. Methods: We conducted a retrospective municipal and individual-level study using 1 069 174 death certificates to analyse COVID-19 and non-COVID-19 deaths classified by ICD-10 codes. Excess mortality was estimated as the increase in cause-specific mortality in 2020 compared with the average of 2015-2019, disaggregated by primary cause of

Introduction The Chinese nation is a large family composed of 56 ethnic groups, all of which have learned and helped each other in history, and together created a brilliant Chinese civilization. Throughout the long history, each ethnic group has its own unique language, culture, and social customs, and through long-term communication, they have learned and integrated with each other, forming a unique Chinese language and culture. However, due to geographical location, historical and cultural factors, and economic development, there are significant differences in language use and communication among different ethnic groups. These differences not only affect communication and communication between people, but also affect people's identification and belonging to ethnic groups. There is a close relationship between the use of minority languages and the sense of belonging of ethnic groups, and it is also an important factor for an ethnic group to identify with its ethnic group. In this sense, there is an inseparable relationship between language use and a sense of national belonging. At present, research on the relationship between language use and ethnic belonging has achieved certain results in relevant fields. However, these achievements are mostly studied from a certain aspect or perspective, and have not formed a complete system, nor have they been analyzed and demonstrated from multiple perspectives. Language is the most important communication tool in human society. Kessler Greg mentioned that humans live in an era of unprecedented opportunities to communicate with others in real and compelling language and cultural contexts [1]. Nguyen Trang Thi Thuy believed that mainstream language is established as a code of intergenerational solidarity among young members, while mother tongue is considered a power code related to older members of the family and community [2]. Against the backdrop of mainstream languages being the main language users, the language usage environment of ethnic minorities is worth discussing. Serrano Maria-Sierra Cordoba mentioned that the concept of "minority languages" has always been a topic of special concern in linguistic sociology [3]. Rodrigue Landry believed that social determinism has greatly influenced the use of minority languages in society [4]. Giollagain Conchur O believed that the symbolic language policy for ethnic minorities is harmful to threatened ethnic minorities because it ignores the social environment of ethnic minority language ISSN 2616-5783 Vol.6, Issue 15: 86-92, DOI: 10.25236/AJHSS.2023.061514 Published by Francis Academic Press, UK -87-decline [5]. Magdalena Skrodzka examined the impact of minority language use on history [6]. The Peace Hughes Tracey emphasized the importance of families and communities in the revitalization of minority languages [7]. In the contemporary world, language teaching is responsible for preparing learners to interact with people from other cultural backgrounds and teaching them skills, attitudes and knowledge [8][9]. Due to the fact that language is an important tool for people's communication and interaction, there is a need for mutual understanding and communication among different ethnic groups during the communication process, which also promotes the widespread use of minority languages. Since entering the 21st century, with the continuous development of globalization, the connections between countries around the world have become closer, and they are facing opportunities and challenges together. At the same time, with the acceleration of economic and social development in ethnic areas, the increasing frequency of population mobility, and the increasingly close communication and integration of various ethnic groups, many ethnic groups have developed varying degrees of language and cultural barriers. These all affect the communication and development between different ethnic groups. How to enhance the sense of belonging and identification of ethnic minorities in a harmonious and stable language environment has become a noteworthy and important issue in current education work in ethnic minority areas. Therefore, it is of great significance to study the relationship between the use of ethnic languages and the sense of belonging of ethnic groups, which can not only provide a scientific basis for the state to formulate relevant ethnic language policy, but also provide a certain reference value for the development of ethnic regions. --- Language Use and Group Belongingness --- Language Usage In Chinese mainland, most ethnic minority students face the challenge of learning three languages in school, namely their mother tongue, Mandarin, and a foreign language, usually English [10]. The degree of language use also represents the degree of identification with the language used. Language identification to a certain extent affects the direction of language retention patterns. The higher the sense of identification with the mother tongue, the more conducive it is to maintaining the mother tongue. The use of minority languages has various purposes, sometimes to promote the content of minority languages, but more often to attract people's attention to these language contents [11]. People who use their mother tongue as a common language exhibit a preference for their mother tongue [12]. Each ethnic group has varying degrees of emotional identification with their own language, while they have a more reasonable sense of identification with Chinese. Within the same region, there may be subtle differences in the specific manifestations of different races. However, most of them have a natural and profound emotional attachment to their mother tongue and the traditional culture of their own nation, which drives them to do their utmost to protect and protect their mother tongue and culture, thus enabling the long-term maintenance of their own language community. The ethnic languages of ethnic minority students constitute their ethnic connotations [13]. A common mother tongue indicates ethnic relations and trust [14]. In the process of urbanization, most urban minority groups have transitioned from monolingual and quasi bilingual to truly bilingual, multilingual, and multilingual. During this process, while maintaining the emotional identity of their mother tongue, the language communication function and identity of indigenous people have been weakened, and the status of Chinese as such has become more prominent. The common sense understanding of "national language" and monolingual or multilingual language has developed in the social context [15]. Speaking a late acquired second language involves an increase in cognitive needs, which is mainly reflected in young and middle-aged people [16]. For newcomers, the formation of their language identity is also directly related to the degree of urbanization, and their language identity vary with the degree of urbanization in the destination, thereby affecting the use and maintenance of their language identity. In areas with high levels of urbanization among ethnic minorities, people have a lower level of identification with their language communication function and are less likely to retain their mother tongue. Urban ethnic minority groups generally tend to prefer Chinese in language selection and use, but there are significant intergenerational differences in the use of their mother tongue, and the level of mother tongue use is decreasing, especially among the third generation and young people. Consistency and accuracy when used in ethnic languages are important factors in language acquisition [17]. In short, whether it is traditional urban ethnic minorities or different population types of urban ethnic minorities, their recognition of their mother tongue directly affect the maintenance mode of their mother tongue. ISSN --- Expressions of National Belongingness Sense of belonging is a unique psychological characteristic of a social group, and it is also the psychological belonging of social members to the social group they belong to. The sense of belonging plays an important role in group psychology and serves as a bridge for the interconnection and interaction between individuals and groups in social groups. A person can only be said to have a sense of belonging when they have a certain emotional attachment to the social group they belong to and are willing to make a certain contribution to this social group. Ethnic belonging refers to a psychological identification of a certain ethnic group towards the ethnic group to which it belongs. In psychology, people refer to this sense of national identity as a sense of national belonging. Sense of belonging is a broad concept that spans several disciplines of social science. In sociology, the concepts of belonging and identity can sometimes be interchangeably used to capture subjective feelings as part of a social group [18]. In a country, people with the same ethnic group have a sense of belonging, which can be shared experiences of war, disasters, and hardships, as well as shared development and prosperity. Regardless of the situation, there is a sense of identification with this country and nation. People further demand the development of precise measures regarding the need for a sense of belonging and the realization of a sense of belonging, as well as new measures to strive to achieve a sense of belonging through specific channels [19]. Ethnic groups not only represent genetic diversity, but also exhibit strong local characteristics in work and daily life [20]. In the study of ethnic identity, a sense of belonging to a group is an important concept. Ethnic identity is a complex structure that not only includes an individual's sense of belonging to the group, but also includes positive evaluations of the group to which they belong, as well as their investment in group activities. Some people believe that the core of national identity is the attitude towards the inner group, some believe that a sense of belonging and commitment are the core of national identity, while others believe that identification with national language, culture, and history is the core of national identity. By general standards, from the perspective of modern ethnic identity research, developmental psychologists often place self-identity at the core of ethnic identity. Social psychologists often put the shared values, sense of belonging, attitude and behavior of ethnic members at the core of national identity. Sociologists and anthropologists focus on cultural identity, language identity and national history identity in national identity. --- Experimental Investigation --- Experimental Ideas Starting from the sense of belonging of ethnic minorities in Yunnan, the current situation, characteristics, and changes of the sense of belonging of ethnic minorities in Yunnan are analyzed, and the impact of the use of minority languages on the sense of belonging of ethnic minorities is explored. 100 ethnic minority residents from A community in a certain region of Yunnan were selected as survey samples, and the current situation of the sense of belonging of ethnic minority residents in A community in a certain region of Yunnan was studied. Establish the research object as ethnic minority residents of A community in a certain region of Yunnan, and design a survey questionnaire to conduct a questionnaire and pre survey on the sense of belonging of ethnic minority residents in A community in a certain region of Yunnan. --- Experimental Samples The sample structure for questionnaire collection is shown in Table 1. From the sample structure in Table 1, it can be seen that the proportion of residents from ethnic autonomous regions and non ethnic autonomous regions in this community is 79:21, indicating a higher proportion of residents from ethnic autonomous regions. In the survey on the ethnic composition of parents, there were more cases where both parents were from ethnic minorities, with 81% having both parents from ethnic minorities and 19% having only one parent from ethnic minorities. In the survey of the proficiency level of the native language among 100 people, 69 were proficient in using the native language, 20 were not proficient in using it, and 11 were completely proficient in using it. ISSN Statistics were conducted on the ethnic groups of 100 survey samples, as shown in Figure 1. --- Figure 1. Distribution of residents of different ethnic groups From Figure 1, it can be seen that in this community, the Yi ethnic group accounts for the majority, with 19 out of 100 people being Yi, 15 being Hani, 13 being Bai, and only 2 being Naxi. Many ethnic groups live in the gap between the dual cultural collision, namely the collision between Chinese and foreign cultures and the collision between ethnic minority cultures and Han cultures. They not only enjoy the fruits of the dual culture, but also bear the impact of the dual culture. Therefore, the research object of this study is selected as the residents of ethnic minorities in A community of a certain region in Yunnan, and the sense of belonging of the residents of various ethnic minorities in the community is studied. The differences in language proficiency levels among 100 ethnic minority residents in A community of a certain region in Yunnan Province, as well as the differences in various dimensions of belonging among ethnic minorities, are shown in From Table 2, it can be seen that there are significant differences in the sense of belonging among ethnic minority residents with different language proficiency levels in the national language dimension. Ethnic language is an important ethnic characteristic of ethnic minorities, so there may be certain differences in the sense of belonging among ethnic residents with different proficiency levels in ethnic languages. The results of this study indicate that there are significant differences in the proficiency levels of ethnic languages and customs among ethnic minority residents who are proficient in using ethnic languages, not proficient in using ethnic languages, and completely proficient in ethnic languages. In terms of ethnic language and customs, the number of residents who are proficient in using ethnic languages is higher than that of residents who are not proficient in ethnic languages, and the number of residents who are not proficient in ethnic languages is higher than that of residents who are completely proficient in ethnic languages. Because ethnic minority residents who are proficient in using ethnic languages have very distinct ethnic characteristics, they may encounter communication barriers or differences in lifestyle habits in modern cities. Therefore, these ethnic minority residents urgently need an environment where they can use their ethnic languages to overcome learning and communication barriers and quickly align with mainstream culture. The proficiency in the use of ethnic languages also largely indicates that the residents have a high sense of ethnic belonging. Ethnic customs and habits have been maintained from childhood to adulthood, and are difficult to change in a short period of time. In modern cities where mainstream culture is abundant, there are naturally many differences in living habits, which is also a very confusing problem for many ethnic residents today, especially for some ethnic minority residents from mountainous areas, Before arriving in the city, there would not be any significant issues with customs and habits due to the fact that the surrounding people were all of the same ethnic group or ethnic minorities. On the contrary, in modern cities, it would be different. Multiple comparisons of the above two dimensions indicate that the significant differences in variance tests in the dimensions of national language and customs are mainly due to the fact that there is a significant difference between proficient and unskilled use of national language, as well as between unskilled and unskilled use of national language. --- Correlation between Language Use and Ethnic Group Belonging In order to further study the relationship between the use of minority languages and the sense of belonging of ethnic groups, relevant statistics on ethnic identity and language attitudes were conducted, and the statistical results are shown in Table 3. From Table 3, it can be seen that both dimensions of ethnic belonging show a significant positive correlation between the level of ethnic identity and attitude towards ethnic language among the 100 ethnic minority residents in A community in a certain region of Yunnan. The commitment dimension of ethnic belonging also shows a significant positive correlation with their overall attitude towards general communication and other dimensions. This positive correlation may be due to the high frequency and widespread use of minority languages. The reasons for this may be multifaceted. On the one hand, the use of minority languages has a certain historical and cultural accumulation, making them play an important role in inheriting their own culture and maintaining their national identity; On the other hand, as language is an important tool for people's communication and interaction, there is a need for mutual understanding and communication between different ethnic groups during the communication process, which also ISSN 2616-5783 Vol. --- Impact of Language Use on National Sense of Belonging Through the above analysis of ethnic language use and sense of belonging, it can be seen that the use of minority languages has a positive impact on the sense of belonging of an ethnic group. From a practical perspective, in today's society, the level of economic and cultural development in ethnic minority areas is relatively low, but this does not mean that the local ethnic minority groups do not have a strong sense of social belonging. In ethnic minority areas, when people use ethnic languages for communication and exchange, it can fully reflect the local people's high recognition of their own culture and historical culture. These have created a good language environment for the local people. If an ethnic group uses their own national language for communication in this environment, they consciously form a sense of identification with their own national history and culture, as well as their own language and culture, and feel their own existence value from it. These can fully reflect the positive impact of minority language use on a sense of belonging of an ethnic group, and improve the positive impact of minority language use on a sense of belonging of an ethnic group. --- Conclusions This study mainly conducted a correlation analysis between the use of minority languages in China and the sense of belonging to ethnic groups. The results showed that there was a significant positive correlation between the use of minority languages and the sense of belonging to ethnic groups. However, overall, the use of minority languages has a positive impact on the sense of belonging of ethnic groups. Due to the fact that minority languages are mainly used as carriers and media, their impact on the sense of belonging of ethnic groups can be significant. With the increasing communication and interaction among various ethnic groups in China, there is a close relationship between the use of minority languages and the sense of belonging to ethnic groups. If a member of an ethnic group cannot use their own language, they develop a sense of unfamiliarity with their own country and ethnic group, which is not conducive to an ethnic group's identification with their country and ethnic group. Therefore, relevant personnel should pay attention to research on the use of minority languages, so that more people can understand this. To fundamentally enhance the sense of belonging of an ethnic group, it is necessary to strengthen the use of minority languages and fully utilize this favorable condition to promote the identification and belonging of minority groups to their own country and ethnic groups. This is an important guarantee for the development of various ethnic groups and the harmony and stability of their relations.

In order to understand the relationship between the use of minority languages and the sense of belonging to ethnic groups, this article conducted a correlation analysis between the use of minority languages in China and the sense of belonging to ethnic groups. Research has found that in a survey of 100 people's proficiency in their own language, 69 were proficient in using their own language, 20 were not proficient in using it, and 11 were completely unable to use it. There is a significant positive correlation between the use of minority languages and the sense of belonging to ethnic groups, and its influencing factors include the number of ethnic populations, proficiency in language use, and attitudes of ethnic members towards ethnic languages. Due to the large population, wide distribution range, and complex influencing factors of ethnic minorities, as well as the high frequency and wide range of use of ethnic minority languages, their impact on the sense of belonging to ethnic groups is greater than in nonethnic minority areas with smaller populations. The use of minority languages can enhance a sense of social belonging within an ethnic group, and this impact is positive and positive.

Introduction Community engagement is historically associated with the third mission of universities through various labels such as outreach, community service, service learning and community service learning. These labels all carry with them slightly different meanings. Initially, CE was regarded as a philanthropic exercise by universities towards communities in need. Kruss et al. and the Council on Higher Education among others explain that CE is now promoted as a mutually beneficial partnership with a range of actors including business and government ministries. The essence of this change in relationship is an emphasis on the co-creation of knowledge and collaboration between university and partnership members . The notion of community has also received much attention; it is an allembracing term that may be geographical, social or ideological. Hall , for instance, describes community as: <ext> a cluster of households or an entire region, as an organisation ranging from a provincial government department to an NGO, as a school, clinic, hospital, church or mosque or as a part of the university itself. … Obviously, communities are a loosely defined set of social organisations. But community also functions as an adjective, as a qualifier that indicates work that is socially beneficial. <ends> Schuetze describes the concept of community engagement as: <ext> the collaboration between institutions of higher education and their larger communities for the mutually beneficial exchange of knowledge and resources in a context of partnership and reciprocity. <ends> The introduction of 'service learning' as a component of community engagement adds a further dimension. This definition has also evolved slightly since the one widely used by Bringle and Hatcher in the context of the United States, where the emphasis was on how students articulated their learning in communities which would then contribute to accreditation of student programmes, to an arrangement whereby students and communities work together for mutual benefit. An example of this latter definition is provided by Stellenbosch University as follows: <ext> An educational approach involving curriculum-based, creditbearing learning experiences in which students participate in contextualised, well-structured and organised service activities aimed at addressing identified service needs in a community, and reflect on the service experiences in order to gain a deeper understanding of the linkage between curriculum content and community dynamics, as well as achieve personal growth and a sense of social responsibility. It requires a collaborative partnership context that enhances mutual, reciprocal teaching and learning among all members of the partnership . <ends> In the context of service learning, particularly in South Africa, the focus is usually on addressing the needs of marginalized and often impoverished communities that are within travelling distance from the university campus. The impetus for this initiative derives from the nation's post-apartheid efforts to address the inequalities of its divisive history. The responsibility of universities towards community engagement and service learning is enshrined in government policy . The ideology of a mutual partnership relationship in these contexts, however, often comes under scrutiny . Service learning requires extensive negotiation and preparation with attention to the inevitable power differentials that surface between an institution with status and resources and organizations or community locations which are targeted on the basis of defined need . Much has been written about service learning and the nature of student learning in the community engagement process. Concerns have been expressed about how power differentials are addressed in these contexts and there has been some exploration, and critique, of how service learning is managed as a student-focused pedagogy that also services community needs . The community perspective on the engagement relationship has also been discussed . It is apparent that the very structure of many service learning courses militates against students contributing to lasting change in communities . Students often have to manage their engagement activities within a full lecturing timetable and over a defined period of weeks, while community needs and activities do not necessarily coincide with such fixed timetables. Publications that pay attention to the organizational arrangements that facilitate such partnerships and how community learning spaces are created are less common . Comparative, qualitative studies of such experiences are also relatively rare. This article compares the preliminary findings of two universities' efforts to involve their students in their local community by drawing on the skills and knowledge they obtained from their coursework. Twelve case studies were conducted . The case studies were all short term projects lasting between six and twelve weeks. Some of them involved students from dedicated service learning courses; others drew on students from existing degree courses. In most cases the students were required to produce coursework that demonstrated learning from their community engagement project. The notion of service learning, according to the above definition, therefore, was interpreted loosely but required application of academic knowledge to a community-defined problem. An additional criterion for the students in the study was that they were expected to work in teams, thus requiring a sharing of ideas and understandings between themselves as students as well as in their community placement. The UFS university campus was situated in a rural location; the UKZN campus was located on the outskirts of a small city with access to both urban and rural conditions. For reasons of space, only four case studies are discussed in detail in this article, two from the rural campus and two from the urban campus. The theoretical concept of adaptive leadership was adopted as both an overall approach and a lens through which to interpret the findings. This concept was adopted because, in the university-community context, it allowed for a recognition that power differentials are an inherent feature of community engagement and that strategies are necessary to manage these differentials in the engagement process. --- Adaptive Leadership Adaptive leadership is an organizational management term coined by Heifetz . According to Heifetz, Linsky and Grashow adaptive leadership is <ext> The practice of mobilizing people to tackle … challenges and thrive. Adaptive leadership is specifically about change that enables the capacity to thrive. New environments and new dreams demand new strategies and abilities, as well as the leadership to mobilize them. Adaptation relies on diversity. <ends> Heifetz's theory of adaptive leadership provided a valuable contribution to understanding how communities may be engaged in the process of development. In this theory, different strategies are followed in different contexts or situations. Heifetz discusses three distinct situations leading to leadership responses, strategies and approaches. The first and probably the most common one is the 'Type 1' situation where leaders conclude that the challenge requires only their or their team's technical expertise. Burke states that Heifetz referred to this as the absence of leadership. In community development contexts, many communities may not challenge the expert's solution. In some cases, the expert may not have understood the situation to the same level as the members of the community would have understood it. At this point, examples of many urban solutions which have been used for rural ecologies bear testimony. A vivid example is the rebuilding of schools in tornado-prone rural areas of the Eastern Cape Province . In the 'Type II' situations, leaders see a problem as requiring some interaction with the community and in so doing view it as a shared challenge that warrants the involvement of the community. This approach, according to Burke , combines the leader's expertise, persuasive powers, and input of the communities. However, this type is characterized by limited leadership. In many instances, however, leaders and 'experts' do not have readily available feasible solutions to seemingly intractable situations. These are what Heifetz calls 'Type III' situations. Whilst the communities may remain content with quick, tried and tested easy solutions to their problems, Burke , in relation to this context, calls for an 'honest and courageous leader who would demonstrate the need for redefining the problem, changing priorities, and possibly greater sacrifice from the members'. Although this level of leadership in small scale projects is inevitably limited, the Type III situation provides space for dialogue in university-community or community-university relations. This leads to the need for ongoing clarification of competing roles and responsibilities. It also shifts power relations so that the 'expert' becomes a participant who may, like all members of the community, learn from others. For individuals in positions of power, this may not be an easy shift. Although the problems were necessarily small scale, we sought, through the study, to place students in situations that made them full members of the communities with no better status or credibility than that of the communities with whom they were to engage. Their leadership role, therefore, included encouraging community or organizational participants to 'clarify values and make progress on the problems those values define' as a dialogic process . Heifetz summarizes this as 'working within society's own frame of reference'. Heifetz's theory further distinguishes between two types of problems that communities face. Drawing from Heifetz's work Kania and Kramer refer to technical and adaptive problems. Technical problems are those that are well defined; the solution is known in advance, and one or a few organizations may be able to provide that. Adaptive problems are, by contrast, more intricate and complex; the solution is not known, and no single entity may be able to provide the appropriate service. Educational transformation and health renewal may be classified as adaptive problems. In adaptive leadership it becomes extremely important for the leader to be fully 'present' to comprehend what is happening with a view to framing key issues and questions from within the social group. A facilitative, inclusive approach is key to ensuring lasting impact . This is in line with Kania and Kramer's work titled 'Catalytic Philanthrophy', where they make reference to the fact that mobilizing multiple organizations and stakeholders may be messier and slower. For example, a rural community may decline a particular process because it does not augur well or resonate with their ways of existence, unless they are given the opportunity for shared ownership and understanding of change. An adaptive leadership approach may challenge and inspire students and afford them the opportunity to lead and stimulate change in community contexts by adopting context-sensitive strategies for dialogue and mutual problem solving. This requires stepping out of their everyday environment, gaining insights into a range of social issues and 'reframing' their challenges through a process of community engagement. It also entails developing a greater understanding of group dynamics and increasing awareness of one's own behavioural patterns and how they impact on others . It is a trust-building process whereby a diversity of views is respected . Since community projects are often multi-layered in terms of relationship structures, this action research project and its academic facilitators encouraged students to engage with those multiple levels of participants in their projects. In summary, the focus was on encouraging community ownership of the engagement initiatives, ongoing dialogue to clarify competing goals and values, and sensitizing students to the need to respect and value diverse community perspectives with a view to contributing to community-identified development challenges. Community engagement relies on these adaptive leadership principles, and in their service learning, students were encouraged to apply these principles in their community engagement projects. As students and university lecturers engage with and serve communities, they need to observe these principles. For example, communities need to feel that problems and solutions are theirs and not those of the students and/or lecturers. --- Methodology This was an action research project because it was more than simply a data collection process. The research focus was on improving what exists by exploring what works and what could be improved, listening to the views of all partners and taking action to address the challenges raised . The project went through four phases. They were as follows: Phase 1: Consultation with relevant organizations and academic departments to match potential students with relevant disciplinary knowledge to the community-identified problem or task. Phase 2: Discussions between organizations and students in preparation for the task. Phase 3: Ongoing monitoring by research assistants and researchers during the case study phase, followed by interviews and focus group discussions with all participants . Phase 4: Feeding back our findings to the participants and discussion of ways forward. The case studies themselves emerged after several consultations with different NGOs following a public stakeholder meeting at each institution. It was a process of trying to match 'this' problem with 'those' students and negotiating how the student's timetable constraints could interface with the timetable demands of the participating community organization. Ethical clearance was obtained from each university; all organizations gave permission for participants to be interviewed and each individual interviewee signed an informed consent form that promised anonymity whilst allowing the researchers to record and publish findings. Interviews with key community participants and students took place either individually or in focus groups, depending on which process was deemed most appropriate for obtaining honest and frank answers . In each case community members were interviewed in their preferred language and the digital recordings were transcribed into English. The research questions relevant to this article were: 1. How did the communities, university staff and students engage with each other? 2. To what extent did the adaptive leadership philosophy contribute to the engagement relationship and its outcomes? Transcripts of interviews and focus group discussions formed the main source of data, though field notes on the preparation phases were also available. The precise form of data collection varied across the case studies according to the circumstances of the participants through what Barton and Tusting have called 'responsive methodology'. Transcripts for each case study were scrutinized by their respective university research teams for patterns of responses and then coded thematically, drawing on the adaptive leadership framework as an evaluative tool for analysing the data. The cross comparison of findings took place during university team meeting discussions at subsequent conferences during 2014. This entailed a process of verification, cumulation, generalization and application whereby common criteria were used for the engagement process, as stated through the adaptive leadership approach, and then the cases were examined in relation to what was happening at micro-level in order to gain 'insight rather than overview' and generate patterns that could inform our analytical understanding across all the cases. For reasons of space only two case studies from each institution are discussed here. They are introduced under their separate university headings. --- The Case Studies --- University of the Free State, Qwa Qwa campus In the first case study an NGO requested student assistance in conducting awareness raising workshops with parents in order to improve their children's literacy levels. Two meetings were held to discuss ways in which the students could be placed with the organization. After the second interaction between the researchers and two representatives of the NGO, a workshop was conducted to train the students to conduct workshops with parents. The coordinators from the NGOs contacted the school where workshops would be held by the students after the training session. Nineteen students conducted five workshops in Sesotho, the local language. The second case study did not have an NGO to mediate between the university and the school. In 2013 a school teacher asked the university for student assistance in the teaching of Mathematics, Social and Natural Sciences in two grade nine classes. Nine university students were tasked to teach on Saturdays. Separate focus groups were held with the university students and teachers from the school. For the literacy programme, a focus group was conducted with both university students and school parents who participated in the workshops. --- University of KwaZulu Natal, Pietermaritzburg Campus Both projects took place over a period of six weeks. The first project involved two isiZulu-speaking political science students in making a film on poverty and hunger in the wider Pietermaritzburg area. The lead NGO worked with a local film-making organization to produce the film. Students assisted with logistical preparations such as liaising with the various households about filming times, dates and requirements and passing this information on to the NGO and film making crew. Students then acted as translators between the crew and the community and vice versa. At an initial meeting between the NGO, the film crew and students, the NGO director led the discussion on the envisaged role of the students and the nature of the project. Times and days were negotiated. At the end of the six week period, the two students and the NGO director were interviewed regarding the benefits and challenges of the service learning experience. The second NGO trains local members of a semi-rural township to facilitate morning crèche activities with pre-school age children whose families cannot afford the fees of a formal nursery school. Parents are encouraged to attend and join in the nursery activities, though they rarely do more than observe. Two third year students from the education and development disciplines met in the township with the NGO's trainer, the local coordinator and the early child development facilitators to receive some training on the kind of activities that might be employed to support one of the crèche facilitators. At the end of the six weeks, the students, facilitator, coordinator and two observing parents were separately interviewed for their assessment of what benefits and challenges the arrangement had revealed. The findings for all four case studies are presented in terms of key themes that emerged and could be relevant for leading adaptively: building on community assets, paying attention to power differentials, dialogue, shared learning and stimulating change. Respondents and case studies are categorized as follows: --- Project --- Findings --- Building on Community Assets The students took their responsibility seriously, recognizing the existing knowledge of communities, and learned to respect the contributions that community members were making within their own environments, even when it appeared that assets were limited. <ext> The role they [community facilitators] play is … a very empowering role … even though they know that they have nothing at all in life but they see that they can do something with their lives … the parents of the children they are teaching trust them … even though they know that they are not qualified teachers . <ends> Similarly, with case study 2 it was evident that because NGOs work closely with communities on the ground they are aware of the issues confronting communities. The NGO director explained the organization's approach to community knowledge and to creating awareness of community issues at different levels of involvement: <ext> So there are two levels: the level on which the film is telling the story through the eyes of the community people, but [also] external people who have knowledge on the context of food prices . <ends> In the literacy project the students learned from the parents how to handle learners, so that in order to motivate children to do better; the children's responses have to be handled with decency: <ext> I also learnt that you have to address your child in a decent manner. Attempt to use different strategies to help the child discover the correct answer. Do not say the child is stupid [dumb] as this may make or break the child's future . <ends> This included encouraging children to aspire to greater heights in life: <ext> As a teacher you need not be negative when you ask learners about their future careers. You should encourage the child to work harder . <ends> At the same time, in the spirit of adaptive leadership and in recognition of the power differentials between a university and a community setting, students were asked to comment on how they managed to address that challenge. --- Paying Attention to Power Differentials The early child development project students revealed it was not always easy to gauge how to interact with their facilitator: <ext> This was a learning curve that this [project] is their baby so we were afraid to raise some points because maybe she would take it as though you are undermining her position. That was also a challenge for us . <ends> This meant that they had to find ways of introducing new ideas that did not alienate her. So they learned to imagine how the facilitator might feel and be sensitive to her context: <ext> What I learned was the facilitators they take this job very seriously, it is kinda like it's their baby and if someone else from the outside tries to intrude somewhere … you are attacking them personally so … if you want to intervene … do it in a way that … does not seem as if you are attacking them, in a way that we are here to learn . <ends> --- And slowly they built a relationship whereby each would support the other: <ext> We were second facilitators … but we were not superior to her. We tried to work as a team with her … she would let us do our thing and she would continue with her thing. So we tried to play the same role as her even though she is the more educated one when it comes to ECD and children … but in the eyes of the children … we were also their teacher … she would say 'these are your teachers' . <ends> The students in the film project recognized that they were dealing with several layers of participants but that the main decision-makers were the film-making company: <ext> our focus was on the communication and logistics stuff so to write appointments and stuff so [NGO] and the film makers would actually make the big decisions and what we actually did was ensure that everyone was available, you know. Perhaps if they were not available on Tuesday then we would have to settle for another day which was gonna accommodate everyone of us and ensure everyone is there and not left behind so they took bigger decisions like for [film crew], they are the film makers they are the ones shooting so we can't agree on a day on which they are going to deal with something else, we have to hear from them if they can come . <ends> Both the literacy project students and the community parents learnt something from their collective participation. Learning from one another balanced unequal power relations. It was not only parents whose patience was tested; students also felt that the project enhanced their ability to manage patience: <ext> As university students, this project assisted us to gauge our patience towards people . <ends> Parents also appreciated university students' demonstration of love and respect: <ext> We appreciate the respect . We thank you people [university students]; we thank you very much . <ends> A key ingredient for ensuring such a positive working relationship was to engage in dialogue. --- Dialogue This process of building constructive dialogue took time. At first for the students in case study 1, their efforts appeared to flounder. For instance, they built on the trainer's advice to bring in plastic cartons to use as resources for the children, but did not realize that the facilitator would not necessarily have the skills to develop these resources: <ext> We didn't know how to do certain things with her [the community based child development facilitator], how do we engage with her in doing something because … we took the plastics to her and said here are the plastics what should we do? … and she was like 'eish I don't know as well'. <ends> It then became apparent that ideas must be introduced slowly and with time to allow for clarification and competing agendas. Most importantly this required an opportunity for all participants to respond and get used to new ideas: <ext> What we learned … if we have ideas that we want to implement … inform her that okay, we are thinking of this, to do this, is it alright? So that the other person is in the loop … by the time we arrive she knows already that this is what's going to happen . <ends> The students from the film project commented on the ongoing dialogue that happened between all parties to make the project work: <ext> Well in terms of making decisions we all discussed it, there were emails, when we see each other. Like they worked around our time and we worked around their time we all compromised something in like, ja. And then we contributed by availing ourselves or trying to work around each other's time like we couldn't clash, if you get what I am saying . We all compromised time and whatever . <ends> The UFS students who were placed in the literacy booster project commented that they were able to discuss with parents and teachers and as a result they could work with other students whose perspectives, regarding certain issues, were different from theirs. 'My answer should not be the only correct answer', said one student. In conducting workshops, students complemented each other's presentations: <ext> If I were to do the seven sessions alone, it would have been impossible for me. Because of the other members of the team, it is now possible and enjoyable . <ends> Furthermore, in the initiative where grade nine learners were taught by the students, the initial dialogue made students realize the need for personal sacrifice in terms of time. They learnt the need to keep a balance between their academic work and the community engagement requirements: <ext> It was sometimes a challenge. You will find that sometimes you would be writing on Monday, and sometimes you are given material on Friday to prepare for Saturday . <ends> He explained that because of the initial discussion to help the learners, they were continuing with their university work after classes: <ext> After the lesson you will have to come back and continue with the university academic work . <ends> Most importantly, the discussions between different community members ultimately provided opportunities for shared learning. --- Shared Learning A key outcome of the dialogue which engaged in clarifying competing values and purposes was the opportunity to build on each other's knowledge base. So, for instance, the local child development facilitator in the UKZN case study realized the benefits of new ideas, especially when she was allowed to steer the intervention in a way that recognized her knowledge of the children and their learning abilities. <ext> If they have an opinion they would share and I would also do the same when I had an opinion about the children … see they are also educated -they came with the shapes and the robots ... [but] this one day they finished with shapes and proposed to teach them about robots. We told them no, they shouldn't -these children are still young, they shouldn't learn everything at once in a day . <ends> Within the space of six weeks both students felt they were learning: <ext> I gained an understanding that children don't learn in the same way as adults and that children learn in a slow pace and that you need to be patient … also kids learn things better if its visual … it must be … colourful and bright . <ends> And the facilitator felt she was working in a partnership relationship: <ext> We had different ideas and worked in a good partnership … we used to discuss things; agree that they can take over now. They would tell the story they prepared and I would also tell some of their stories that they've told . <ends> Students from the film project learnt directly about poverty and its effects on people, issues that they were dealing with in their university courses which contributed to their linking theory to practice: <ext> There are things we aren't aware of as I live in a kind of incubaterish community, ja. Yes I did gain some information … like on Saturday when the participants are actually talking about their situations you could see that some things went beyond hunger and everything and it was more psychological, ja you could just read it through their body language so it affected them more than, so it did go with my psychology module [more] than my with my politics module . <ends> Another student explained further that this experiential learning is powerful and can be emotional: <ext> It hit me hard emotionally; I think that I put my heart too much into it. I learnt that that it could be visible to society that a certain household is coping…. But in that household things are going wrong that nobody can see. People go days without eating and yet they still work . <ends> From the perspective of the NGO director he felt that they also learnt from the students from the project because the dialogue between participants' knowledge was shared: <ext> It also helped us from the point of view of, mmm, the students themselves because they sometimes suffer the same issues that the people in the community suffer they helped to bring also another dimension of understanding of food insecurity to the script … so because they also took part in the discussion, they also said things which was quite interesting stuff, like what is happening at varsity and how there is hunger in hostels . <ends> In the literacy project one single-minded student reported that she had acquired the knowledge and skill of discussing and working together with people of opinions different from hers. She said that she had learnt to work together with other people and accommodate views different from hers: <ext> I also learnt to take other's points of view into consideration; I should not be the only one who enjoys prominence . <ends> As for the maths and natural sciences case study, our findings revealed that the students were able to link what they were taught at the university to what they were expected to teach at school. The students further experienced that in some instances the university's programmes were not preparing them sufficiently for the workplace: 'When it comes to Geography there is little map work that is being done', commented one student, 'so that places a lot of pressure on us as students when we are in the field. We struggle because we were not equipped well on that part'. Two other students shared similar sentiments: <ext> From the first year to the third year, we haven't learnt anything about map work. We started learning when we were looking at the method of teaching Geography. Luckily, we got some information from one high school so that we could know the formula to calculate the gradient . We were provided with textbooks with no information on the topics to be presented. For an example, we were to teach about the gradient. But only to find that there is no information about the gradient. I then went to Mr Vxxx at S Secondary School. I asked information about the practical. He provided me with the topographical and ortho-photo maps. This made the practical to be possible as some of the calculations were to be taken from the map. We found that at the school there were no maps. We tried to download some maps from the internet, but we could not get the correct one. So I have learnt that if I am given a topic to present I need to search for some information instead of folding my arms . <ends> An important concern of adaptive leadership is the application of such strategies to stimulate change. The projects were small-scale and it was difficult to assess the extent to which lasting change might have been implemented. Nevertheless there were signs that the students were keen to introduce change and in most projects there was evidence of change in attitudes at community level, albeit tentative at this stage. --- Stimulating Change The parents involved in the UFS literacy project began to recognize a number of ways in which they were able to assist their children. This included working with other children in addition to their biological children. <ext>I am able to assist other children other than mine . <ends> Some affirmed that they were better able to care for their children after participating in the initiative: <ext> I did not have that patience for children. Now I am no more impatient with them . <ends> Others felt they were now able to assist their children, taking them step-by-step: <ext> I read and guide my child. When she asks questions I respond. She did not know the meaning of the word 'hygiene' and I explained it to her . <ends> Parents also recognized that they should create time to be with their children in order to assist them when doing their homework: <ext> If there is anything I am doing, I immediately leave it and assist my children if there is a need .<ends> Many parents pointed out that they were able to assist their children at home: <ext> [even with homework] I now understand that the child does not only need to be taught at school, but at home as well . I am now able assist my child to hold a pen, as I have been taught to do so . <ends> They demonstrated that they were now using a variety of strategies to make learning fun: <ext> I did not present what I was reading in an interesting manner, without actions. That is why the children were bored because I was just reading . <ends> The project also created a bond between parents and their children. Parents said that they were now able to discuss matters that they were unable to discuss before participating. For example: <ext> There were things I could not discuss with the children but now I can. The children find it easier to approach me now. This change seems to have cemented the relationship I have with my children . <ends> In the early child development project, from their own perspective the students felt they introduced new ideas: <ext> Since we are students … I think we have like more, fresher ideas of teaching children compared to the facilitators there; even though it was threatening but then we had fresher ideas. We had new ways of implementing the teaching … we tried to have more fun ways of learning for the kids . <ends> The facilitator felt both empowered as a facilitator and, as a result, that her role was being taken more seriously by the children's parents. The very fact that university students were taking part in the project had created a new power dynamic that gave the it new legitimacy: <ext> We were very happy to be with you [students] and the children also saw -and I also saw that I am also important … the children and parents saw that this is a legitimate thing … we wish that you people could come back again because your presence has been noted by the parents.... Your presence helped because some parents thought this was just a game. Some even refused to allow their children to come … now they saw that this thing of teaching from home is serious … because of the students that came and brought some of their things as well and they saw . <ends> In support of this change of attitude, the parents indicated increased interest in the idea of playing a supportive role in their children's crèche. Although this may simply have reflected a desire by the parents to please the interviewer: The NGO director of the film-making project felt that because of issues that students raised about student hunger that: <ext> They [students] enriched the process through their thinking engagement … I could see a possible connection for this for not this year but later on in the year because I have friend of mine in the university who is working on issues of hunger in the hostels so mmm, it might lead on to something like perhaps another DVD or something . <ends> The director was also hopeful that the film produced during this project would be a catalyst for change in the lives of the community: <ext> If this film can get a wide audience and can get a conversation going within a sitting about this phenomenon, about hunger, inevitably different role players would have to start doing something about it. Mmm, it becomes a social justice movement . <ends> --- Discussion Each project entailed similar strategies for organizing the service learning initiative. For instance there were several preparatory meetings between the students and community organizations. Sometimes there were two layers of participation, such as between academic staff and students and the initiating organization. Occasionally there was a third layer of participation, such as for the early child development and literacy projects where the NGO acted as a mediator between university and community facilitators. But the project 'beneficiaries' -at community grass-roots level -were only involved at the evaluation interview stage. This signifies that in these case studies the community engagement relationship remains embedded in a partnership for development 'for' rather than 'with' the community at grass-roots level. The partnership 'with' relationship is with a mediating NGO or school. Such an organizational strategy is perhaps unavoidable for short term service learning projects but it does highlight a potential limitation for community engagement partnerships that aspire to contributing to community change. Nevertheless, the community-based learning spaces provided opportunities for students to understand the context-specific challenges of Heifetz's Type III problems where there are no easy technical solutions. A prime example of this adaptability in action was evident in the natural sciences classes where students had to use their own initiative to acquire the necessary practical resources. At the same time the community spaces provided a unique opportunity for shared learning whereby nearly all the respondents shared how they had learned from each other. This suggests that the non-formalized nature of community environments creates a resource for collective learning whereby new relationship dynamics and forms of dialogue take place. Within these spaces students and community members learn to recognize the value of pooling assets, skills, knowledge and understanding. Within that process of informally re-framing key issues and questions and facilitating learning for mutual problem-solving, students had to step out of their everyday learning environment and build trust with their diverse community contacts . The follow-up interviews indicated that the students had contributed to stimulating change amongst community members, particularly in the projects involving children, but also at NGO-level in relation to the food security film-making project. On the one hand, therefore, these case studies indicated the potential of an adaptive leadership approach to contribute to community change through small-scale service learning projects. They also indicated the need for extensive preparations prior to the engagement activity and the limitations of what can be achieved in terms of working with grass-roots communities -unless a mediating organization can maintain continuity after the student interventions. The extent to which change is sustained could not be ascertained in these case studies in view of the time limited nature of the research study. Nevertheless the comparative element of these four case studies can produce some tentative conclusions. --- Conclusion Each university focused on different contexts but the process of adaptive leadership produced similar outcomes in terms of facilitating shared ownership and development of knowledge which builds on community assets, and is sustained through a process of dialogue. This relationship in turn stimulated motivation for change among students and community members. In such diverse and multi-layered community contexts there is a need for constant vigilance regarding power dynamics between university and community levels. But often this power differential can have positive effects in that the very presence of university interest in community initiatives can stimulate a sense of self-worth and self-respect, which in turn motivates enhanced community activity. However, in order to maximize such potential there must be ongoing dialogue and respect for diverse views.

The purpose of this article is to compare how students and community members learned and applied their knowledge in four small-scale university-community engagement projects during 2013. It draws on the concept of adaptive leadership as an approach and analytical tool in a recently completed community engagement and service learning action research partnership between the University of the Free State (UFS) Qwa Qwa campus and the University of KwaZulu-Natal (UKZN) Pietermaritzburg campus. The project was funded by the National Research Foundation, with additional support from the UKZN Teaching and Learning Fund and UFS Faculty of Education research funds. A total of twelve case studies involved sixty-five students, nine NGOs and four schools. In each case, students worked in teams in response to community requests for assistance. Projects included Saturday curriculum activities for schools, workshops for parents, assisting with film making or archiving, assisting with monitoring and evaluation of rural reading clubs, producing small organic gardens and assisting with a childcare development project. Each case study involved end of project interviews with students and community contacts and some interim observations during the project implementation phase. This article compares four of the case study findings between the two institutions. It outlines how

Introduction Metabolic syndrome is a combination of interrelated risk factors that predispose individuals to the development of cardiovascular disease and diabetes. This includes hyperglycemia, raised blood pressure, hypertriglyceridemia, low high-density lipoprotein cholesterol levels, and abdominal obesity, and is now recognized as a disease by the World Health Organization and other international entities [1,2]. According to a study done across seven European countries, the overall prevalence of MetS was estimated to be 23% using the WHO criteria [3]. In Canada, nearly 25% of the adult population was found to be afflicted with MetS using the National Cholesterol Education Program -Adult Treatment Panel III [4]. In Australia, the prevalence values of MetS using the WHO, NCEP-ATP III and International Diabetes Federation criteria were 21.7%, 21.1% and 30.7%, respectively [5,6]. This points to the fact that prevalence of MetS within the same region may vary according to the definitions employed. Moreover, this variation could be due to the differences in the defined cut offs for its associated metabolic components. As the proportion and distribution of body fat in Asians, in general, was found to be different from the populations in North America and Europe, it became apparent that the definition of obesity applied to Western populations was not be applicable to Asian populations [7,8]. Therefore, the estimated prevalence values of MetS among Asians were found to be increased when Asian-adapted definitions of obesity were employed in the NCEP-ATP III. For example, in the Southeast Asian region, it increased from 13.1% to 20.9% for Singaporean males, and for the Chinese adults, it increased from 10.1% to 26.3% [9,10]. A similar trend was observed among the Malaysians where during the 2008 nationwide survey an overall prevalence of 42.5% from 4341 subjects was reported using the Joint Interim Statement "Harmonized" criteria, compared to 34.3% via the NCEP-ATP III criteria [11]. Like in NCEP-ATP III, MetS according to the Harmonized definition includes any three of the five metabolic abnormalities-central obesity, hypertriglyceridemia, low HDL-cholesterol, high blood pressure and hyperglycemia [2]. However, the Harmonized criteria have defined Asian cut-offs for central obesity and reduced cut-off for hyperglycemia . Ramli et al., using the Harmonized definition reported the prevalence of MetS to be 43.4% in 2013 among 8,836 subjects across East and West Malaysia [12]. This percentage was very close to the 42.5% prevalence reported in the 2008 nationwide survey [11]. The prevalence of MetS is dependent on a variety of non-modifiable and modifiable risk factors. These factors are known to, directly or indirectly, influence MetS among the populations. For instance, Wen and colleagues reported the prevalence of MetS in rural China as 44.3% , 40.7% and 47.7% , amongst a large cohort of 4748 subjects, primarily females aged 50 years and above [13]. From a study in Canada, Liu and colleagues reported MetS prevalence to be higher among the Cree Indians compared to other aboriginal and nonaboriginal Canadians [14]. Similarly, the prevalence of MetS in Malaysia is not different, as according to the nationwide survey in 2008, the prevalence was found to be higher among older age groups, more among females, and most common among the Indians compared to other races in Malaysia [11]. Studies have shown that various lifestyle factors influence MetS. A sedentary lifestyle and physical inactivity are factors that have been shown to contribute to the development of MetS and its components [15][16][17][18][19]. Smoking and alcohol consumption have also shown to have variable influences on MetS and its components [20][21][22][23][24]. Furthermore, diet habits such as speed of eating, dining out, skipping breakfast and late dinners have been found to be associated with increased incidence of MetS [25,26]. These factors are present in most ethnic communities and might provide some insight into how their influence on MetS could be regulated among populations to contain its life-threatening complications. Reports mentioned above indicate the significant influence of lifestyle habits on the prevalence of MetS in a particular population. Malaysia is a unique country in Southeast Asia because of its ethnic diversity, culture, lifestyle choices and dietary intake habits. The influence of differing lifestyle choices and diet habits across the three major races of the country may provide a better understanding of the high prevalence of MetS in the country, along with measures for its containment. There have been very few studies carried out in Malaysia on investigating the influence of lifestyle factors with the risk of MetS among the Malaysian population [27][28][29]. While the two studies by Chu and Moy described the influence of physical activity on MetS among the Malays, the only study which dealt with ethnic differences with respect to physical activity and prevalence of MetS among the Malaysian population was based on the data that were collected more than 13 years ago [27][28][29]. Moreover, in that study the relationship of lifestyle behaviors with MetS among major ethnicities was not reported [29]. Therefore, the objective of the present study was to determine the association of sociodemographic characteristics, lifestyle factors and diet habits with the risk of MetS, overall and among the three major ethnic groups residing in Johor, Malaysia. --- Materials and methods --- Study design and location This was a cross-sectional study, employing a nonprobability sampling strategy, conducted in Kulai and Felda Taib Andak of the Kulai district and Johor Bahru , Ulu Tiram and Kota Masai of the Johor Bahru district of Malaysia. Based on the Department of Statistics Malaysia, Johor represents the other states of Malaysia in terms of ethnic distribution. Furthermore, the selection of the above-mentioned study locations was based on the available percentages of MetS across each major ethnicity of Malaysia, reported in the nationwide survey 2008 [11]. This was to have enough subjects to represent each ethnicity so that data could be available for in-depth analysis for the stated objectives. As a result, the selected sample size from these locations was considered to be generalizable to some extent to the population of Johor, and to population of some of the other states in the country, with similar distribution of the three major ethnicities. --- Recruitment and eligibility criteria Research camps were set up in central locations of Kulai and Johor Bahru districts, which were easily accessible to the target community. Assistance was sought from community elders for making the locals aware of the research camps and to convey our requests for their participation. The inclusion criteria for the study participants were that the subjects should be of ages 18 years or above, of either gender, and had been residing in Johor for at least one year. The subjects were requested to observe a 10 to 12-hour fast before arriving at the medical camp to donate blood samples for accurate assessment of fasting serum levels of glucose, triglycerides and HDL-cholesterol. Exclusion criteria included pregnancy or having any illness which could preclude their participation in the study such as cancer, liver disease, etc. Consented participants were invited to visit these camps for a physical examination and collection of fasting blood samples. After sample collection, the subjects were asked about their lifestyle and dietary habits. Participants who did not observe the 10-12 hour fast were excluded from the analysis. --- Data collection and measurement Data from the participants were collected using a structured questionnaire and a proforma which contained information on anthropometric measurements, measurement of blood pressure, blood sample analysis results and questions on sociodemographic, lifestyle factors and diet behaviors. The questionnaire was designed in English and back translated into the Malay language. In the study, the Malay language version was used. The questionnaire was pretested on a sample of 29 subjects in Kulai before its employment on the main target population. A copy of the questionnaire is attached as supporting information . Body height was measured using Seca stadiometers , while the weight was measured using the InBody 120 body fat analyzer . Steps were taken to ensure the subjects wore light clothing and had no shoes on. The measurement was recorded to the nearest 0.1 cm and 0.1 kg, respectively. Waist circumference was measured using a measuring tape. The measurements were taken at the mid-point, between the lower rib margin and the iliac crest. Caution was taken during measurements that the subject was standing straight with feet together and arms relaxed on either side. Furthermore, it was ensured that the tape was held in a horizontal position, wrapped around the waist, loose enough for the assessor to insert his/her finger between the tape and the subject's body. The subject was instructed to breathe normally during the assessment, with the measurement recorded at the end of a normal exhalation and rounded to the nearest 0.1 cm. Blood pressure was recorded using the Omron digital sphygmomanometers . The subject was provided a 4-5 minute rest, in a seated position, with the arm supported at heart level. At least two readings were taken from each subject, recording the concurrent or highest measurement obtained from the two readings. A third reading was taken in case, the difference between the two readings for the systolic blood pressure was more than 10 mmHg, and for the diastolic blood pressure more than 5 mmHg. Fasting blood samples were collected from the study participants for determining the levels of fasting serum glucose , fasting serum triglycerides and fasting serum HDL-cholesterol . Standard guidelines for phlebotomy were followed throughout the venepuncture procedure [30]. Collected samples were transported in cold chain to the laboratory where these were centrifuged, and the sera samples were separated and placed in identity marked cryotubes or Eppendorf tubes. These were then placed in a -60 degree Celsius freezer till laboratory analysis. The blood analysis for determination of serum levels of fasting glucose , triglycerides and HDL-cholesterol was carried out using clinical chemistry analyser . Its reagents were purchased from Randox Laboratories, United Kingdom. The participants' physical activity status was determined using the validated International Physical Activity Questionnaire [31]. The questionnaire comprised seven questions; the first two pertaining to the time spent on vigorous activities performed, the next two for moderate activities, the next two for mild activities and the last question was on the time spent. Responses were converted to Metabolic Equivalent Task minutes per week according to the IPAQ scoring protocol. The protocol also provides details for data processing, cleaning and truncation. The total minutes spent on vigorous, moderate, and mild activities over the last seven days were multiplied by 8.0, 4.0, and 3.3, respectively, to create MET scores for each activity level. MET scores across the three sub-components were then summed up to indicate the overall physical activity score. These overall scores were then categorized into high , moderate and low level activities. Diet habits included quick finishing of meals, frequency of late dining, frequency of skipping breakfast and frequency of dining out. For quick finishing of meals, the question was asked on the subject's perception on finishing their meals either fast or not fast [26,32,33]. The assessment of the other three diet habit questions were based on the participant's frequency per week; three times or less were considered favorable [26]. "Late dining" was defined as a meal eaten within two hours before bed-time. "Dining out" was defined as a meal consumed by the participant that is not prepared at his/her home [34][35][36]. --- Definition of MetS MetS was defined using the Harmonized criteria as having at least three of the following five risk factors: 1) Abdominal obesity, defined as having a waist circumference � 90 cm for males and � 80 cm for females; 2) Raised serum triglycerides , defined as � 1.7 mmol/L ; 3) Low high density lipoprotein cholesterol , defined as < 1.0 mmol/L for males and < 1.3 mmol/L for females; 4) Raised blood pressure, defined as a systolic blood pressure � 130 or a diastolic blood pressure � 85 mmHg, or current use of anti-hypertensive medications; and 5) Raised fasting blood sugar , defined as � 5.6 mmol/L or current use of anti-diabetic medications. --- Statistical analysis Data entry was performed using EpiData version 3.1. During the process of data entry, 5% of the forms were re-checked for accounting of any errors during entry of data. All data were analyzed using Statistical Package for Social Sciences version 23 . The sample size estimate was calculated using estimates of various components of MetS reported in the 2008 nationwide survey [11]. According to the calculation, increased blood pressure yielded the sample size estimate of 386 at 5% level of significance and precision of 0.05. Frequencies and percentages were obtained for categorical variables. Chi square tests for Independence were used to determine the univariate association between categorical variables. Multiple logistic regression analyses were used to determine the associations of sociodemographic and lifestyle factors with MetS, calculating odds ratios with 95% confidence intervals, while adjusting for confounding factors. Variables, with p < 0.25 on univariate analysis were selected for adjustment in the final logistic regression model. A p < 0.05 was considered statistically significant. --- Results The prevalence of MetS was found to be 32.2% in the study subjects, according to the Harmonized criteria; highest among the Indians , followed by the Malays , and lowest among the Chinese . Abdominal obesity and high blood pressure were more common compared to other metabolic abnormalities. Three most prominent MetS risk factors among the Malays and the Indians were abdominal obesity, high blood pressure and low HDL-cholesterol. Among the Indians however, the percentages of abdominal obesity and HDL-cholesterol were higher than that among the Malays. Prevalence of high blood pressure was more prominent among the Malays compared to the other ethnic groups. Among the Chinese, the third most prevalent risk factor was hypertriglyceridemia. Prevalence of low HDL-cholesterol was lowest among the Chinese. Table 1 shows the summary statistics of sociodemographic, lifestyle and diet characteristics overall, and across the 3 major ethnicities in Johor. The target population consisted of 64.9% females, while the remaining were males. About 49.9% of the population were of ages 50 years and above; from which 25.6% of them were aged 60 years and above. A majority of the participants were married , attaining at least secondary education and were unemployed . Thirty-one percent of the target population were Malays; Chinese were 47.4% while the Indians constituted 22% of the target population. Table 1 also shows the comparative association of sociodemographic, lifestyle and diet characteristics with MetS overall, and across the three major ethnic groups in Johor. Overall, significant differences were observed with age, ethnicity, marital status, education and physical activity . Marital status and education were found to be related significantly with MetS among the Malays, while age and physical activity among the Chinese and age among the Indians showed a significant association with MetS. Among the Malays, 59.4% of the people with primary education or lower were having MetS, suggesting that the Malays having higher education appear to be protected against the risk of MetS . A majority of the Indians appear to be afflicted with MetS at a younger age . Conversely, only 15.2% of the Chinese were suffering from this syndrome in this age group . This shows that the Chinese in Johor are getting this disease at a relatively older age. Table 2 shows the adjusted multiple logistic regression model, the results of which indicate that overall in this population, higher age groups and the Malaysian Indians had increased odds of MetS, while the Chinese ethnic group and those with tertiary education were protected against the risk of MetS. Lifestyle factors and diet habits did not appear to have any association with MetS, overall, in the adjusted model . In view of the contrasting estimates of MetS among the ethnicities, we explored the effect of ethnicity further with sociodemographic, lifestyle and diet factors. Table 3 shows the adjusted logistic regression models among the Malays, the Chinese and the Indians, revealing higher odds for MetS for quick finishing of meals among the Malays and low physical activity among the Chinese . Furthermore, higher educational categories were protective against MetS among the Malays. Among the Indians, older age groups were more prone to developing MetS, while significant odds with respect to age were found among the Chinese older than 60 years of age . --- Discussion In this study, the prevalence of MetS was found to be 32.2%, which was unexpectedly less than that reported in the 2008 nationwide survey, that contained 19% of subjects from Johor [11]. Prevalence among the Indians and the Chinese turned out to be 51.9% and 20.2%, respectively. Comparing the prevalence values reported in the nationwide survey 2008 for these two ethnic groups in Malaysia, the prevalence among the Indians appears to have remained unchanged over a period of nine years, while among the Chinese, the prevalence has reduced considerably from 42.1% to 20.2% [11]. However, among the Malays the prevalence has decreased from 43.9% to 36.7% [11]. This apparent decline among the Malays and the Chinese has been attributed to decreased prevalence values of hyperglycemia, low HDL-cholesterol and hypertriglyceridemia. The present study showed that MetS was more prominent among the higher age groups. This finding has been observed by other researchers as well. He et al., reported a comparatively higher prevalence of MetS among older subjects compared to those aged between 60-69 years among a total of over two thousand Chinese subjects. [37]. In the study by Rampal et al., the prevalence of MetS among the Malaysians was found to be higher among subjects aged 40 years and above compared to those aged less than 40 years [38]. Moreover, in the study by Ramli and colleagues, the odds of MetS, irrespective of definition applied, were found to be higher among higher age groups, and maximum among subjects aged 60 years and above [12]. The nationwide survey also reported higher prevalence of MetS among higher age groups; additionally, higher age groups also had a higher prevalence of central obesity, high blood pressure, low HDL-cholesterol, elevated triglycerides and hyperglycemia [11]. This suggests that higher prevalence of MetS among higher age groups may be due to the accumulated higher prevalence of its associated cardio-metabolic risk factors among elderly subjects. Studies have shown that ethnicity influences the prevalence of metabolic syndrome. For instance, from a study in Canada, MetS prevalence was reported to be higher among the Cree Indians compared to other aboriginal and non-aboriginal Canadians [14]. The Cree Indians also had a higher prevalence of central obesity and hyperglycemia compared to other ethnic groups in the country [14]. Similarly, from a study in Suriname, South America, MetS prevalence was reported as the highest among the Hindustanis , compared to other Suriname races [39]. The prevalence values of high blood pressure, low HDL-cholesterol and hyperglycemia were also high among the Suriname Hindustanis [39]. In a local study among obese adolescents, Indians again had the highest prevalence of MetS, contributed mainly by higher prevalence of central obesity, increased blood pressure and low HDL-cholesterol [40]. In our study, results show that the Indians in Johor are at a greater risk of developing MetS, while the Chinese appear to be less prone to developing MetS. This is in line with the reports from other researchers from Malaysia that the Chinese have lower odds, while the Indians have higher odds of developing MetS [12,38]. More educated adults in the Johor area, especially the Malays, appear to be protected against MetS, probably due to their increased awareness of healthy lifestyle habits, such as engagement in physical activity, smoking cessation, moderate to none consumption of alcohol and adoption of healthy eating habits [26]. This is supported by a couple of studies showing that the individual's higher level of education is protective against diabetes and hypertension, which are prominent risk components of MetS [41,42]. Furthermore Kaur et al., reported that the odds of MetS among Malaysian Punjabis were higher with primary education, compared to those with higher education levels [43]. Ching et al., however, have reported that higher education levels of Malaysian vegetarians with and without MetS were observed to be nearly the same [44]. This could be due to the fact that it was a unique group of subjects with specific dietary habits and the results pertaining to this group may not represent the general population of Malaysia. Literature suggests that excess energy accumulated in the adipose tissues causes metabolic abnormalities, leading to high blood pressure, hyperglycemia, hypertriglyceridemia and inflammation, hence, regular physical activity enhances energy consumption leading to reduced prevalence of obesity, hypertension, diabetes mellitus and also MetS [45,46]. Our results show the prevalence of MetS and its components to be comparatively lower among the Chinese than in the Malays and the Indians, and this could be attributed to better physical activity among them. Chu et al., have shown that longer sitting time and insufficient physical activity have resulted in an almost 4-fold increase in MetS risk among the Malays, and the risk gets reduced by 50% by engaging in moderate to high physical activity [27,28]. On the basis of these reports, it can be suggested that despite having a decreased risk of developing MetS, the Chinese in Johor can still benefit by engaging in moderate to high levels of physical activity. A number of studies have shown a direct relationship of smoking with the risk of MetS, yet in the current study smoking does not appear to be associated with the risk of MetS [20]. This could be due to a small percentage of past and current smokers in this cohort. Similarly, no association was found between alcohol consumption and risk of MetS among those in this population. Again, the reason could be the small proportion of subjects who were reported as alcohol consumers . The association of dietary habits, such as quick finishing of meals, frequent dining out, late eating, skipping breakfast, with MetS has been reported in other studies in the East Asian region [25,26,[47][48][49][50]. For example, Shin et al., reported quick eating as one of the risk factors for MetS among the Koreans [50]. Among these dietary habits, quick finishing of meals was identified in the current study as a new risk factor for MetS in Malaysia, especially among the Malays. According to Dallman and colleagues, fast eaters may consume more food than usual, or be eating under psychological stress which affects hormones regulating metabolism [51]. The underlying mechanism of relationship of such habit with the metabolic health functioning, however remains unclear [50]. There are certain limitations that warrant consideration. First, the present research study was cross-sectional in nature, assessing the exposures and outcomes at the same point in time. In this regard, the findings cannot indicate causality. Second, the selection of study locations harboring subjects was non-random and partly based on the information on the available percentage of MetS across each Malaysian ethnicity in Johor as reported in the nationwide survey 2008 [11]. This was done to have sufficient number of subjects in each ethnic group for better analysis and interpretation, especially among the minority ethnic groups, which had been underrepresented in previous research from Malaysia. Though ethnic stratification was never intended, the in-depth analysis showed the variable influences of lifestyle and diet habits on MetS, especially among the Indians and the Chinese with adequate power, suggesting that the variable effects were more likely attributed to the cultural diversities across the different ethnicities of Malaysia. Moreover, as the information collected was based on recall, misreporting of information cannot be completely ruled out, and this might have added some variability in our results. Despite these sources of potential variability, the results provide credible evidence towards the association of certain sociodemographic, lifestyle and diet factors that affect the disease spectrum of MetS in Johor, and provided an opportunity to further analyze these characteristics influencing MetS across the three major ethnic groups of this state in the country. In this regard, we believe the current research study to be adequate and its findings comparable to similar studies by other investigators using a non-randomized design and exploring associations of various risk factors influencing metabolic diseases in Malaysia. Based on the results of this study, it can be deduced that the population of Johor is diverse in its habits pertaining to lifestyle and diet. Some of these factors are associated with the risk of MetS in certain ethnic groups and modifying these factors would be important for reducing cardiovascular and metabolic health risks among Malaysians. Though not analyzed in detail in this research, effects of other variables determining socioeconomic position, like wealth index, could be explored with MetS in future research. Moreover, further prospective studies delineating the association of various diet habits among different ethnic communities would be imperative to contain the unfavorable effects of this syndrome on the overall health of Malaysians. Since Malaysia is a multi-ethnic country, it would be important to consider this ethnic variation, especially with respect to lifestyle and diet factors, so that intervention programs for addressing behavior modifications would also be tailored across different ethnicities of the country. Increasing awareness among the masses through electronic and print media about the beneficial effects of healthy lifestyle is likely to be another powerful approach to combat the menace of this syndrome in Malaysia. the Monash University faculty and staff, namely Mr Chui Chor Sin, Mrs Savithri Gopal, Ms Pang Pei Ling, Ms Ungku Zulaikha Ungku Omar, Mr Muhammad Daniel Mahadzir, Dr Nor Azim, Ms Kong Li San and Dr Iekhsan Othman for all their support and assistance in this study. Additionally, the authors would also like to thank Ms Harbans Kaur Singara Singh for proofreading the manuscript. --- Ethics Ethical approval was sought from the Monash University Human Research Ethics Committee , which was granted before the start of sample collection. --- Supporting information

Literature shows a high prevalence of MetS among Malaysians, varying across the major ethnicities. Since sociodemographic characteristics, lifestyle factors and diet habits of such communities have been reported to be diverse, the objective of this study was to investigate the association of various sociodemographic characteristics, lifestyle factors and diet habits with MetS overall, as well as with the three major ethnic communities in Malaysia, specifically.We conducted a cross-sectional study among 481 Malaysians of ages 18 years and above living in the state of Johor, Malaysia. Information on demographics, lifestyle and diet habits were collected using a structured questionnaire. Harmonized criteria were used to assess the status of MetS. Multiple logistic regression was employed to determine any associations between sociodemographic and lifestyle factors and dietary behaviours with MetS.MetS was found among 32.2% of the respondents and was more prevalent among the Indians (51.9%), followed by the Malays (36.7%) and the Chinese (20.2%). Overall, increasing age (AOR = 2.44[95%CI = 1.27-4.70] at 40-49 years vs. AOR = 4.14[95%CI = 1.97-8.69] at 60 years and above) and Indian ethnicity (AOR = 1.95[95%CI = 1.12-3.38)] increased the odds of MetS, while higher education (AOR = 0.44[95%CI = 0.20-0.94] decreased the odds of MetS in this population. Quick finishing of meals (AOR = 2.17[95%CI = 1.02-4.60]) and low physical activity ) were associated with increased odds of MetS among the Malays and the Chinese, respectively.The population of Johor depicts a diverse lifestyle and diet behaviour, and some of these factors are associated with MetS in certain ethnic groups. In the light of such differences,

Background HIV/AIDS remains one of the most common public health problem in developing countries like Ethiopia [1]. According to global burden of diseases report, HIV/ AIDS accounted for 1.1 million deaths in the world in 2016. it was the greatest single cause of years of life lost , contributed to 59.8 million years of life lost [2]. HIV/AIDS has been a major cause of morbidity and mortality in resource-poor settings, especially in sub-Saharan Africa where around 25 million people were living with HIV [1]. Ethiopia is among the countries most heavily affected by HIV/AIDS which poses threat to the country's overall development. In 2016, an estimated 740, 251 people were living with the virus and 442,895 of them were females [3]. Moreover, AIDS accounted for an estimated 34% of young adult deaths and 632,670 orphanages in the country [4]. Specifically, In the study area 18,642 people were living with HIV/AIDs in 2016. HIV/AIDS has changed individual's lifestyles and quality of life. It leads to physiological, physical, psychological, and sociocultural problems that are caused by many factors such as symptoms of the virus, side effects of ART, and opportunistic infections [5,6]. Studies have shown that HIV/AIDS has multidimensional consequences: personal suffering such as discomfort associated with the disease's progression, the social impact of the diagnosis, the emotional consequences of dealing with the diagnosis, and related stigma [7][8][9]. Studies have identified disease-related factors such as CD4 count, viral burden, HIV disease stage, infections [6,9,10] and psychosocial factors like social support, coping, and disclosure [7] as predictors of quality of life among people living with HIV/AIDS. Depression has also been associated with changes in general health perceptions, emotional well-being and QOL domains in HIV patients [11,12]. Additionally, several socio-demographic characteristics such as age, gender, education and employment were also factors associated with lower QOL [6,8,13]. Hence, finding ways of mitigating these factors and consequences makes quality of life in PLWHA a relevant issue of health care [14]. The information on gender-specific differences in QOL outcomes has been controversial. On the whole, several studies have shown that females had significantly lower quality of life compared with males [6,[15][16][17][18]. On the other hand, in studies conducted in western countries gender showed no major impact on QOL [19,20]. However, in settings with limited health care resources, heavy HIV-related stigma and discrimination like Ethiopia, women's entrenched economic and social inequality within their relationships with men that could constrain their ability to access to care, treatment and supportive services. These barriers could make women more vulnerable to the physical and psychological burden of HIV [4,21]. A study examined gender differences in quality of life in Ethiopia reported lower quality of life among women compared to men [22] Other studies conducted on QOL in patients with HIV /AIDS have reported overall low QOL scores [5,[23][24][25][26]. However, more specific conclusions about gender differences difficult to be made from these studies due to their limitations that included: lack of comparison group and underpowered since single proportion was used to determine sample size which made the samples difficult to be representative and only over all QOL was reported .It would be important to determine if there are gender differences in QOL domains from comparative studies in order to understand differences in how men and women are affected by HIV/A IDS, and potentially identify areas where interventions could improve patients' quality of life. In general, evidence showed that there appears to be very limited research measuring outcomes in terms of improved health related quality of life in Africa including Ethiopia [27]. HIV/AIDS remains one of the key challenges for the overall development of the country [4] due to high rate of morbidity and mortality as well as its negative impact on individuals well-being and quality of life [3,22]. In addition, the presence of women's economic dependency on men and relatively high stigma and discrimination among HIV infected women could limit their access to treatment and supportive services [21]. Moreover, investigation of contextual gender specific factors associated with QOL would help policy makers to plan gender specific comprehensive and effective HIV care service to improve the lives of PLWHA [28]. However, gender differences in QOL domains and respective predictors not well documented in Ethiopia in general and specifically in the study area. Hence, this study intended to assess sex-specific social, physical, mental, independence, spiritual and environmental domains and their determinants among HIV/AIDS patients in order to understand how people's lives are affected by HIV infection, to help health providers in making the best choices in patient cares and aid to design palliative programs accordingly. --- Methods --- Study design and participants The study used institution based comparative crosssectional study design and the study population were HIV/ADIS patients aged 18 years or older who have been attended ART at public health facilities in West shoa Zone from April to May,2016, western Ethiopia. Sample size was determined using the following formula which helps to compare between two means male versus females with consideration of power 80%, 95% CI, a 1: 1 ratio and Overall HRQOL mean 81.2 for males and 77.1 for females [5]. n 1 ¼ n 2 ¼ Z α=2 þ Z β À Á 2 σ 1 2 þ σ 2 2 ð Þ Δ 2 Δ = /μ1-μ2/ = mean difference between the two respective groups and are the means and variances of the two respective groups n1 ¼ n2 ¼ 1:96 þ 0:84 ð Þ 2 14:2 ð Þ 2 þ 17:4 ð Þ 2 À Á ¼ 236 81:2-77:1 ð Þ 2 After adding 10% non-response rate the final sample size was 520 . From 24 health facilities that provide ART services in west shoa zone, Ambo Hospital and five health centers were randomly selected using lottery methods. Then the study participants were selected proportionally from each health facilities using simple random sampling method using their ART unique number. --- Data collection and measurements HRQOL was measured using the interviewer administered World Health Organization's Quality of Life HIV short form instrument [29]. The WHOQOL-HIV BREF instrument produces six domain scores and contains 31 items. For each item there is a five-point Likert scale where 1 indicates low or negative perceptions and 5 high or positive perceptions. These items contain six domains: physical health , psychological well-being , social relationship , environmental health , level of independence , and spiritual health . There are two items that examine general quality of life: question 1 asks about an individual's overall perception of quality of life and question 2 asks about an individual's overall perception of his or her health. The physical health domain contained information on presence of pain, energy, and sleep. The psychological domain consisted of negative and positive feelings, self-esteem, and thinking. The social domain covered social support, personal relationship, and sexual activity. Mobility, work capacity, and activities were included in the level of independence. Financial issues, home and physical safety and security, and participation in leisure activities were included under the environment domain. The spirituality domain did contain questions about death and dying, forgiveness and blame, and concern about the future. The suggested reference time frame of QOL experienced within two weeks was used in this study [28]. Depression was assessed using BDI-II [30]. It consists of 21 questions scaled from 0 to 3, so the lowest possible total score is zero and the highest possible total for the whole test would be sixty-three. BDI-II: 0-13 points considered no depression and above 13points considered depression. Anthropometrical measurements were taken to assess nutritional status of patients. Weight of the participants was measured in light clothing and bare foots using standard beam balance to the nearest 0.1 kg and the scale was checked at zero before each measurement. Height was measured using the standard scale. The subjects were asked to remove their shoes, stood erect, and positioned at the Frankfert plane with feet together and knees straight. The heels, buttocks, shoulder blades and the back of the head were in touch against the vertical stand of the stadio meter and the values were recorded to the nearest 0.1 cm. Body Mass Index / Height 2 was computed using the weight and height measurements to assess nutritional status. The standard cut-offs were used: < 18.5 kg/m 2 underweight, 18.5-24.9 kg/m 2 normal, and greater than or equals to 25. 0 kg/m 2 was considered overweight. Clinical, laboratory and ART data were collected through reviewing records from ART entry registration book and individual followup form using pretested data collection form. Data on WHO clinical stages, CD4 count, reported side effects, drug adherence, and functional status of participants were extracted by six trained nurses from medical charts in the ART clinics. Data about socio-demographic characteristics, quality of life, depression and family support from family members in terms of psychological, financial, material or ART treatment support were collected through face to face interview using pretested interviewer administered questionnaires. --- Data analysis Data were entered to Epi-Info version 3.5.1 statistical software for windows and analyzed using SPSS version 22 software. The WHOQOL-HIV BREF [29] was used to produce a QOL profile derived from six domain scores denoting facets of an individual's perceived QOL. The mean score of items within each domain were used to calculate the domain score. Mean scores then multiplied by four to make domain scores comparable with the scores used in the WHOQOL-100, a commonly used scale. The instruments' guidelines for checking and cleaning data and computing domain scores were rigorously followed. Participants characteristics were described in terms of mean and median for continuous data and frequency distribution for categorical data. Pearson correlation coefficient was used to evaluate correlation between the domains of quality of life. Independent samples t-test was used to compare domains of quality of life between men and women. Bonferroni procedure was used to declare significance difference in correlated domains of quality of life between men and women in order to minimize type one error. Overall significance level was divided by the number of two-way comparisons made and significance level 0.006 was used declare the significance difference between the two means. The Cronbach α was calculated to determine the internal consistency of the different domains of the WHOQOL-HIV instrument. Mean was used as a cutoff point for quality of life domains because of a normal distribution of the scores. To determine independent predictors of QOL, participants were divided into two groups based on the mean score of the facet because the WHOQOL-HIV BREF denoted facets of an individual's perceived QOL as 1 = Very poor, 2 = Poor, 3 = Neither poor nor good, 4 = Good and 5 = Very good, mean score ≤ 3.0 indicates poor QOL and mean score of domain since mean scores multiplied by four to make domain scores comparable with the scores used in the WHOQOL-100, a commonly used scale, domain mean score ≤ 12.0 indicates poor QOL. Therefore, Participants with mean score > 12.0 were categorized as having good QOL and their counterparts mean score ≤ 12.0 as having poor QOL. Bivariate and multivariate logistics analysis were carried out to assess independent predictors of HRQOL domains in both genders. Goodness of fitness of the final model was checked using Hosmer and Lemeshow statistic. Crude and adjusted odds ratios with their 95% Confidence Interval were estimated and P-Value less than 0.05 was used to declare presence of significant association between HRQOL domains and covariate. --- Results --- Socio-demographic characteristics of the study participants A total of 251 males and 254 females responded to the study making a response rate of 97.1%. The mean age for males was 38.5 years while for females 33.8 years. Majority of the study participants were rural resident; 177 males and 158 females. One hundred eighty-six males and 158 females were married. Regarding education, 163 males and 141 females were literate. Two hundred thirty-one males and 210 females were employed . --- Clinical profiles, nutritional and depression status of the study participants Half of the study participants were at clinical WHO stage III, 124 of males and 122 of females. Median CD4 count was 209 for males and 186 for females. Concerning nutritional status; 49 of males and 75 of females were undernourished. Sixty-five of males and 95 of females were depressed. Twenty-five of males and 33 of females had tuberculosis. Sixty of males and 75 of females were Anemic. 154 males and 138 females received psychosocial support from family members or friends or organizations . --- Internal consistency of the WHOQOL-HIV questionnaire and Beck depression inventory tool The internal consistency of the WHOQOL-HIV tool was high and to measure internal consistency, the Cronbach's alpha was also calculated for each domain of the instrument. Most domains of the WHOQOL-HIV had a high value of Cronbach's alpha ranging from 71 for psychological domain to 83 for environmental domain. But, medium internal consistency was observed in social relationship domain physical . Inter domain correlations showed that there were statistically significant associations between domains. Strong correlation was found between level of physical health and independence , and weak correlation was observed between spiritual and environmental . The correlation between items in the Beck Depression Inventory scale ranged from 0.52 to 0.76 and the internal consistency of the Beck Depression Inventory tool was high . We found strong correlation between the QOL domains and the Beck Depression Inventory scale. Strong correlation was found between the Psychological domain and the Depression Inventory scale followed with social . Physical , level of independence , spiritual and environmental domains with the Beck Depression Inventory scale . --- Gender difference in mean score of health related quality of life domains The result showed that average scores of females for all the six domains and two general questions were lower than the males score on the 4-20 scales. There was significant difference in quality of life between males and females for all domains and two general Questions except the Social relationships and spiritual/personal belief domains . The overall mean quality of life perception was 3.5 for males and 3. 0 for females. Similarly, the mean score for general health perception score was 4.2 for males and 3.5 for females. Mean Physical domain was significantly different between males and females; 15.3 for males and 13.9 for females. Similarly, males and females were significantly different in Psychological Domain; with mean of 14.1 for males and 12.6 for females. Mean Level of independence domain for females 12.1 was also significantly lower than males 13.7. Social relationship was the second lowest QOL domain, with mean score of 12.0 for males and 11.9 for females. Spiritual/PB domain was the highest HRQOL domain with mean ; 15.7 for males and 15.5 for females and environmental domain was the lowest with mean score; 11.4 for males and 10.4 for females . --- Predictors of health related quality of life domains Sex-specific multiple regression analyses revealed that education, occupation, malnutrition, depression, tuberculosis, anemia and family support were independent predictors of health related quality of life domains in both genders. In addition, rural residence was also significantly associated with QOL domains in females. Illiterate individuals reported significantly lower QOL in independence , social and environmental quality of life in both female and male respondents. Similarly, employment was also significantly associated with physical, psychological, social and environmental domains in both genders. Among females, rural residents showed 1.8 and 1.5 times lower QOL in independence and environmental domains than urban residents respectively . Physical, independence and environmental QOL domains were significantly reduced among undernourished patients in both genders. Furthermore, undernourished females had 2.2 times poorer psychological QOL compared with well-nourished counter parts . Similarly, statistically significant differences were found in physical, psychological and environmental QOL domains with anemia status in both male and female patients. In addition, anemic women had 1.9 times lower independence QOL compared with women who had no anemia . For both genders, tuberculosis was the common predictor of physical, psychological, independence and social QOL domains. Besides, TB/HIV co-infected females had 2.0 times poorer environmental QOL as compared to only HIV infected females . Depressed PLWHA had significantly lower QOL in all domains as compared with HIV infected patients without depression in both genders. Likewise, Family support was also a common predictor for psychological, social and spiritual QOL domains in both male and female patients. Moreover, women who received family support were found to have 1.8 times higher physical QOL compared with women who lacked family support . However, age, marital status, CD4 count and WHO clinical stage were not significantly associated with quality of life . --- Discussion This study revealed that females had significantly lower quality of life in physical, psychological, independence and environmental domains as compared with males. This finding was consistent with study conducted in Ethiopia reported that females had significantly lower QOL for all domains except the social relationship domain [22]. The study done in Vietnam also found that females had significantly lower scores than males in environmental and psychological domains [15]. Cross sectional studies conducted in Estonia and Nepal also supported this finding that female had lower QOL in physical, psychological, social and environmental health [16,17]. The possible reason for females scoring lower QOL than males could be that many women living with HIV are burdened by responsibility of child rising [18]. Scholars also suggested that women report poorer QOL because their illnesses may be taken less seriously and therefore they receive less social support than their male counterparts [31]. The result also showed females were less educated, more unemployed and had more opportunistic diseases like TB, depression, malnutrition and anemia than males which might lead to poor QOL. In this study social relationships health domain was not significantly different between men and women which was in line with the findings of other studies [15,18,22,31]. Similarly, Spiritual domain was not significantly different between men and women though high quality of life than the other domains. The possible reasons for scoring high quality of life than the other domains could be that Displayed the gender difference in mean score of health related quality of life domains of PLHIV on ART in public health institutions, western Ethiopia. The result showed that average scores of females for all the six domains and two general questions were lower than the males score on the 4-20 scales. There was significant difference in quality of life between males and females for all domains and two general questions except the Social relationships and spiritual/personal belief domains people tend to be spiritual and religious when confronted with problems that are beyond them; they engage in spiritual and religious reflections [32]. This study also indicated that illiterate individuals had significantly lower QOL in independence, social and environmental domains compared with literate counter parts. This was in concordance with a multi-country study among patients with HIV [6], and a study conducted in Ethiopia [23] stated that the less educated had lower QOL. Another study conducted in northern Ethiopia also reported illiterate People living with HIV/ AIDS had lower physical and social domains of quality of life [22]. This might be due to literate individuals could have better financial resources, work capacity and access to quality health and social care. This study also revealed that unemployed individuals had poor QOL in physical, psychological, social and environmental domains compared with employed individuals. This finding is consistent with other similar studies [13,24,25]. Study conducted in northern Ethiopia also reported that monthly income significantly associated with Physical, independence, social and environmental domains [22]. Similarly, PLWHA who received family support had significantly higher QOL in physical, psychological, social and spiritual domains as compared with who lacked family support. Studies also stated that lack of social support, lower level of education and income had associated with poor QOL [22,33,34]. This result may be attributable to the fact that psychosocial support could have increased personal satisfaction and positive entire effect in having good nutrition, self-care and better health [35]. Among females, residence was one of the significant predictor of their quality of life. Rural resident women showed lower QOL in independence and environmental domains than urban residents. A facility based cross sectional study from northern Ethiopia also found significant association between rural residence and poor quality of life [22]. This may be associated with the presence of relatively poor infrastructures, more financial constraints as well as high social stigma and discrimination in rural areas. In addition, females in rural areas engaged in more physical demanding work like farming as well their economic dependency on men compared to urban dwellers could limit access to better health services [21]. The study identified that physical, psychological, independence and environmental QOL domains were significantly reduced among undernourished HIV/AIDS patients. This finding was supported by the study from Nepal [17] and India [36]; that showed most of health related quality of life domains of PLWHA were significantly decreased as body mass index lowered. A study from South Africa also found that physical and independence health were affected by daily energy intake [10]. This was also strengthen by the outcome that better nutrition promote quality of life [37]. Physical health might be lowered because of under nutrition aggravates severity of HIV. AIDS symptoms. Poor psychological health could be due to fright that underweight may exposed their HIV status and get rejection from their family and social life or because of negative perception about their body image change due to underweight [38]. Poor independence health among malnourished HIV/AIDS patients might be due to decreased activity, mobility and work capacity. In this study anemia co-morbidity in HIV patients was independent strong predictor for poor physical, psychological, independence and environmental health domains. This finding was consistent with a study conducted in India which found that lower hemoglobin level significantly affects five health quality of life domains except social domain [10]. Evidences also indicate that anemia among HIV/AIDS patients was correlated with high degree of disease advancement, diminished health related quality of life and high mortality [39][40][41]. Longitudinal study also discovered that increscent of hemoglobin levels through treatment will reduce blood transfusion necessities and enhance quality of life of people living with HIVAID [42]. The result also showed that tuberculosis co-infection was one of a strong determinant for decreased physical, psychological, independence, social, and environmental health among HIV/AIDS patients except their spiritual health. This find was supported by studies conducted in Ethiopia and in Nepal that stated Co-morbidity with tuberculosis in HIV infected patients deteriorates all six health related quality of life domains [11,23]. Studies also described that HIV patients had poor QOL as compared to the general population [9] and lower QOL values were scored in tuberculosis patients [34,43]. These findings indicate that the double burden from the two diseases have synergetic impact on overall health state of the patient. The study identified that depressed PLWHA had significantly lower QOL in all domains as compared with HIV infected patients without depression in both genders. Studies also reported depression depreciates all health related quality of life categories and has been recognized as main determinant of alterations in overall health perceptions and psychological conditions in HIV infected people [12,16,17,44] together with social care and psychotherapy practices [45,46]. A study conducted in Ethiopia also identified that depression was associated with poor physical, social and environmental health [23]. In addition, lower quality of life scores were found in individuals with common mental disorder symptoms [26,33,47,48]. Evidences revealed that QOL could be influenced by various determinants of psychological morbidity and depression is one of the most important of these factors [11,[49][50][51]. The study had limitations that should be acknowledged. This study was subject to selection bias since the respondents were ones who were actively seeking routine medical care selected from health facilities. They may not represent patients who do not come to the institution from the communities; therefore, generalizability is only for those who are on ART follow-up. Since interview administered questionnaires for HRQOL in the last 2 weeks were used interviewer and recall bias may be introduced to the study though pretested. The cross sectional nature of the study itself limits conclusions on QOL over a period of time and difficult to know cause-effect temporal relationships could not establish the circumstances resulting in low health related quality of life. Since there are many social factors that may alter people's health related quality of life on HAART and might confound the findings of this study. we recommend further longitudinal studies triangulated with qualitative methods. --- Conclusions The findings showed that females had significantly lower quality of life in physical, psychological, independence and environmental domains as compared with males, except social relationship and spiritual domains. The independent significant predictors of health related quality of life domains in both genders were education, occupation, malnutrition, depression, tuberculosis, anemia and family support. in addition, rural residence was also significantly associated with QOL domains in females. Therefore, to improve HRQOL of PLWHA due emphasis should be given to females and co-infected individuals improve socio-economic status and enhance integrated early detection and management of malnutrition, depression, tuberculosis and anemia among PLWHA in Ethiopia. --- --- Abbreviations AIDS: Acquired immune deficiency syndrome; AOR: Adjusted odds ratio; ART: Antiretroviral therapy; CI: Confidence interval; HRQOL: Health related quality of life; PLWHA: People living with HIV/AIDS; QOL: Quality of life; TB: Tuberculosis; WHO: World Health Organization Authors' contributions DY : proposed, designed and implemented the study, did the analysis and write-up. Then, read, critically commented, and approved the final manuscript. KT: Made substantial contributions to conception and design of the proposal, analysis and interpretation of data, critically edit and approved the final manuscript. EM: Made significant contributions in design, interpretation of data, involved in revising and approved the final manuscript. RT: Participated sufficiently in analysis and interpretation of data, involved in drafting the manuscript, critically edit and approved the final manuscript. All authors read and approved the final manuscript. --- --- Competing interests The authors declare that they have no competing interests. ---

Background: Though HIV/AIDS has multidimensional consequences on quality of life, there is a gap in measuring and monitoring health related quality of life of HIV/AIDS patients. Hence, this study intended to measure health related quality of life domains and associated determinants among people living with HIV/AIDS in western Ethiopia. Methods: A comparative cross-sectional study was conducted among 520 HIV/AIDS patients on anti-retroviral therapy in public health facilities in West Shoa Zone, Western Ethiopia from April to May, 2016. Participants were selected using simple random sampling method. Quality of life was measured using WHOQOL-HIV BREF and depression was assessed using Beck Depression Inventory, Second Edition (BDI-II). Data were analyzed using SPSS version 22. An independent sample t-test was used to compare quality of life domains between men and women and logistic regression analysis was used to determine independent predictors. Results: Females had significantly lower quality of life in physical, psychological, independence and environmental domains as compared with males except social relationship and spiritual domains. Depressed HIV patients had significantly lower quality of life in all domains as compared with HIV infected patients without depression in both genders. Malnutrition and anemia were significantly associated with poor physical, psychological, independence and environmental domains. Anemic women had 1.9 times lower independence quality of life compared with women who had no anemia (AOR = 1.9, 95%CI: 1.4, 3.5). Tuberculosis was also predictor of physical, psychological, independence and social domains in both genders. TB/HIV co-infected females had 2.0 times poorer environmental health compared to only HIV infected females (AOR = 2.0, 95%CI: 1.2, 3.5). Family support, education and occupation were also independent significant predictors of QOL domains in both genders. In females, residence was significantly associated with independence (AOR = 1.8, 95%CI: 1.2-3.8) and environmental (AOR = 1.5, 95%CI: 1.1-3.2) domains. Conclusions: Females had significantly lower quality of life compared with males. The findings indicted poor socioeconomic status and co-infections significantly associated with poor quality of life among HIV/AIDS patients. So, due emphasis should be given to improve socio-economic status and enhance integrated early detection and management of malnutrition, depression, tuberculosis and anemia among HIV/AIDS patients in Ethiopia.

INTRODUCTION Tuberculosis remains a major global health problem . TB affects not only physical health, but also the mental wellbeing of patients. For patients with TB, depression is a common mental disorder . The presence of depressive disorder leads to poor treatment adherence , drug resistance, and high rates of community transmission, ultimately increasing morbidity, and mortality . Therefore, it is necessary to identify the related risk factors of depression among patients with TB and accordingly devise effective interventions. At present, studies have begun to explore the influencing factors of depression in patients with TB to take measures to reduce the risk of depression. However, the prevalence of depressive symptoms in patients with TB is still high, as reported in India , Ethiopia , Pakistan , and Cameroon . Much of the existing TB programs were focused on outcomes of mortality and microbiological cure , while mental health-related outcomes were ignored, which may lead to the high prevalence of depression in patients with TB. In addition, inadequate research on the influencing factors of depression may be another key reason. Previous studies have reported that some personal and social factors were associated with depression in patients with TB, such as age, sex, lack of TB knowledge, insufficient social support, and TB-related stigma . However, it is unclear whether these factors act independently or through complex mechanisms. Furthermore, previous studies exploring factors associated with depression mainly relied on analysis via logistic regression or linear regression analyses . Although useful information was provided, these types of analyses considered the effects among the variables that may affect depression independently; However, it is possible that those variables interacted in a more complex way, which was not straightforwardly captured by the regression analysis. Therefore, the complexity of the interrelationships among them is less well-understood .The lack of research on those complex relationships makes it difficult to determine accurately what variables should be targeted in depression interventions . An alternative method to tackle these issues is the use of structural equation modeling , a technique designed to decompose into direct and indirect effects of variables to understand mechanisms and pathways that may explain these relationships . In addition, SEM analysis is generally more robust than regression analyses because measurement errors are considered . Consequently, we applied SEM to address the complex interrelationships between specific personal and social variables and depressive symptoms among patients with TB in China. Understanding complex pathways may increase the effectiveness of interventions to reduce depression risk among patients with TB, which may ultimately decrease the morbidity and mortality. --- THEORETICAL BACKGROUND --- Stigma and Depressive Symptoms Goffman posited that stigma was a social process that reduced individuals "from a whole and usual person to a tainted, discounted one" . During this process, patients tend to conceal their symptoms and withdraw from interpersonal contact for fear of discrimination, thereby isolating themselves to avoid negative public attitudes. TB is usually associated with uncleanliness, and patients are often attached to a disease label and experience social stigma . Previous studies on the impact of stigma among patients with TB suggested TB-related stigma may lead to diminished self-esteem and self-efficiency and has been found to be significantly associated with the development of mental health problems, especially suffering from depressive symptoms . Moreover, a number of studies have confirmed that patients with TB are more likely to have depressive symptoms than the general population and a high level of stigma among patients with TB in different countries has been reported . However, few studies have investigated the status of stigma and their relationship with depression in patients with TB in China. Due to variation in socio-cultural elements, ethnicity, and region, factors associated with TB-related stigma may differ and it is necessary to focus on the stigma and depression of patients with TB in China. Personal and Social Factors and Depressive Symptoms The evidence for the relationship between social support and depressive symptoms has been well-substantiated by previous studies . Indeed, the family is the main source of social support when the disease occurs, whether through tangible instrumental support, such as administering medication and preparing meals, or through emotional support . A-well functioning family suggests that the family could adapt to the crisis including the patients illness and role change and may be effective in lowering the risk of depression, while family dysfunction would suggest that the home environment might be stressful to the patient with TB. In addition, good doctor-patient communication is also a source of social support for patients. Previous studies has demonstrated that the levels of perceived social support received from healthcare workers are negatively associated with depressive symptoms . Jittimanee et al. found that patients with low TB knowledge were more likely to have severe TB disease and to have depressive symptoms . Personal and Social Factors and Stigma Various researchers have explored the relationship between personal and social factors and stigma in patients with TB. The relationship between social support and TB-related stigma has been examined and low levels of social support was associated with high levels of TB-related stigma in patients with TB . Few studies, however, have been published to date on the relationship between family function and TBrelated stigma. Only one study conducted in Pakistan showed that good family function can alleviate stigma in patients with TB . Stewart et al have reported doctor-patient communication may have negative correlation with stigma . Moreover, in our previous study, the relationship between patients' knowledge about TB and stigma in patients with TB has been confirmed. We found that knowledge about TB may have negative correlation with stigma using logistic regression . --- MODEL ASSIGNMENT AND HYPOTHESES In summary, previous studies provided some evidence that TBrelated stigma are influenced by personal and social variables . Previous studies also provided evidence that patients with greater TB-related stigma were more likely to have higher levels of depression . However, the complex interrelationships between these variables have not been explored. Based on the above review, a hypothetical model was obtained. As illustrated in Figure 1, we proposed the following hypotheses: Family function, doctorpatient communication, and knowledge about TB will be negatively related to stigma. Stigma will be positively related to depressive symptoms. Stigma will mediate the relationship between family function, doctorpatient communication, knowledge about TB, and depressive symptoms. --- METHODS --- Participants and Sampling Ethical approval was provided by the Human Research Ethics Committee, Huazhong University of Science and Technology, Wuhan, China. Each participant was voluntary and provided written informed consent prior to participating in the study. A cross-sectional study was carried out between October 1, 2013 and March 31, 2014 in Hubei province, central China. Multistage stratified sampling was performed. Counties across Hubei province were divided into three categories based on their economic development status , and a county was randomly selected from each category. Thereafter, patients with TB who attended TB clinics at the study sites in the selected counties were recruited as participants. The inclusion criterion was a diagnosis of active TB, based on national TB program guidelines, and the exclusion criterion was the presence of psychosis. Some elderly people with TB have more health problems, which makes it very difficult for them to complete the self-assessment questionnaire, resulting in a valid response rate of 84.7%. Thus, 1,430 patients with TB completed a structured, self-administered questionnaire anonymously. Of the collected 1,430 questionnaires, 88 were removed due to a large number of missing data, and eventually 1,342 TB patients were included in the analysis. --- Measuring Instruments The questionnaire consisted of six sections: Demographic Characteristics, Family Function, Doctor-patient Communication, Knowledge about TB, TB-Related Stigma, Depressive Symptoms. The demographic characteristics included age, sex, educational level, and history of prior anti-TB treatment. Patients' family function was measured using the Family APGAR Index developed by Smilkstein in 1978 . It was designed to evaluate the satisfaction with social support received from their family members based on five components: Adaptability, Partnership, Growth, Affection, and Resolve. Each item is scored using a 3-point Likert scale ranging from "scarcely" to "often" . The item scores are summed to provide a total score and a higher score indicates better family function. The Family APGAR Index has been widely used in China, with good reliability and validity . In the present study, the Family APGAR Index showed high internal consistency . Doctor-patient communication was measured by the following four questions: satisfaction with the doctors' service attitude; extent to which the doctor introduces the illness; extent to which the doctor explains the details of taking anti-TB agents; and extent to which the doctor explains adverse drug reactions to anti-TB agents. Patients were asked to rate their answers in terms of satisfaction or detail scale ranging from 1 to 3 for the four questions listed above. The total score ranged from 4 to 12, and a higher score means better doctor-patient communication. Patients' knowledge about TB was measured by six multiplechoice questions which were mainly from the questionnaire used in the National TB Epidemiological Survey of China : the cause of TB, the route of TB transmission, whether TB can be cured, duration of TB standard treatment regimens, common clinical symptoms of TB, and unhealthy behaviors that make TB susceptible to infection. In the six questions, the first four questions have a single correct answer, and the patient scores 1 point for each correct answer; There were four correct answers to the remaining two questions, and the patient receives 0.5 point for each correct choice. If the answer is incorrect or "I do not know, " then the question score is 0 points. Thus, the total score of TB knowledge ranges from 0 to 8 points. The higher the TB knowledge score reflects that the patient has more knowledge of TB. TB-related stigma was measured using the TB-Related Stigma Scale, which was developed by our research group according to the standard method for the development of new scales and has shown good validity and internal consistency . Detailed information on the development and evaluation of this scale was available in our previous study . The scale consists of nine items divided between three subscales . Responses for the items are provided using a 4-point Likert scale ranging from "strongly disagree" to "strongly agree" . The item scores are summed to provide a total score , and higher scores indicate greater stigma. Depressive symptoms were measured using the Center for Epidemiologic Studies Depression Scale, developed to identify individuals who are at risk for depression . The scale consists of 20 items divided between four subscales: Depressive Mood, Somatic Symptoms, Interpersonal Relationships, and Positive Affect. Responses for the items are provided using a 4point Likert scale ranging from "not at all" to "almost daily" . The item scores are summed to provide a total score . The CES-D has been widely used in china and in the current study the scale demonstrated high internal consistency . --- Data Analyses Descriptive statistics comprised percentages, means, and standard deviations . Student's t-test and analyses of variance were conducted to compare depressive symptoms scores between groups. Bivariate correlations were assessed using Pearson's correlation. The above analyses were conducted using SAS 9.4 software . All differences were assessed using two-tailed tests, and the significance level was set at P < 0.05. SEM was conducted using AMOS 17.0 software to assess the interrelationship between variables associated with depressive symptoms in patients with TB. In addition, the significance of direct and indirect effects was examined using a bias-corrected bootstrap 95% confidence interval . The following fit indices were used to assess the overall model fit: root mean square error of approximation , comparative fit index , and Tucker-Lewis index . CFI and TLI values above 0.90 and RMSEA values below 0.08 indicated acceptable fit; CFI and TLI values above 0.95 and RMSEA values below 0.05 indicated good fit . --- RESULTS --- Participants' Characteristics Participants' characteristics and their association with depressive symptoms are presented in Table 1. Participants mean age was 47.72 years and most were aged 45 years or older. The mean score of depressive symptoms was 15.84 and 48.00% of the participants had depressive symptoms. Being older, female, having low education, and a history of prior anti-TB treatment were associated with higher depressive symptoms scores. --- Correlations of the Variables, Test of Normality, and Multicollinearity Table 2 shows the means, SDs, and correlations among study variables. There were significant correlations between all study variables in the bivariate analysis. The correlation between family function and depressive symptoms was high compared to all the other combinations of correlations. Furthermore, skewness and kurtosis were used to examine the assumption of normality. The skewness of all study variables was within 1.90, and the absolute value of kurtosis was within 3.00; therefore, the results satisfy the assumption of a normal distribution . In addition, diagnostic checks for multicollinearity were conducted through variance inflation factors. In this study, the variance inflation factors ranged from 1.07 to 1.14, suggesting that there was no problem with multicollinearity . --- Effect Analysis of Hypothesized Model The First Model Without Mediator In the first model, the direct effects of the three variables on the dependent variable were tested without a mediator using SEM. Figure 2 showed the first model with standardized path coefficients. We found that family function, doctor-patient communication, and knowledge about TB were negatively associated with depressive symptoms . The overall model fit indices of the first model were RMSEA = 0.039, CFI = 0.981, NFI = 0.972, TLI = 0.974, which all indicate that the model fit well. --- The Final Model Figure 3 shows the final model with standardized path coefficients. The significance of all the effects were examined using a 95% bootstrapped confidence interval estimate. Age, sex, education, and history of prior anti-TB acted as covariates. The summary of the results are provided in Table 3. The overall model fit indices of the final model were RMSEA = 0.043, CFI = 0.956, NFI = 0.939, TLI = 0.946, which all indicate that the model fitted well. As can be seen in Table 3, the direct effect of family function, doctor-patient communication, and knowledge about TB on stigma were significant, which supported hypothesis 1. In addition, stigma exerted a direct effect on depressive symptoms, which supported hypothesis 2. The direct effect of family function and knowledge about TB on depressive symptoms were both significant; however, the direct effect of doctor-patient communication on depressive symptoms was non-significant. Further, the indirect effect of family function, doctor-patient communication, and knowledge about TB on depressive symptoms were significant; therefore, stigma partially mediated the effect of family function and knowledge about TB on depressive symptoms and fully mediated the effect of doctor-patient communication on depressive symptoms, which supported hypothesis 3. --- DISCUSSION To the best of our knowledge, this was the first study to test the mediating effects of stigma on depressive symptoms in patients with TB. We examined the interrelationships between family function, doctor-patient communication, knowledge about TB, stigma, and depressive symptoms. More importantly, we elucidated the pathways linking family function, doctor-patient communication, and knowledge about TB to depressive symptoms and confirmed that the effect of those variables on depressive symptoms can be mediated by stigma, which supported our hypotheses. Previous studies indicated that family function and doctorpatient communication were the most critical sources of social support for patients . Our results showed that family function and doctor-patient communication can also have an indirect negative effect on depression symptoms through stigma. Good family function and doctor-patient communication can increase patient life satisfaction and social confidence. Patients with family dysfunction and poor doctor-patient communication are more susceptible to being isolated and estranged. Because of these negative treatments, patients have developed negative feelings such as shame, guilt, discrimination, etc., which may result in a sense of stigma . Moreover, prominent levels of stigma was associated with psychological stress disorder, which can increase their risk of psychological problems, such as depression symptoms . Knowledge about TB was another influencing factor that had both direct and indirect effects on depressive symptoms. This indirect negative effect was also achieved through stigma. Inadequate disease-related knowledge often represented the absence of a correct understanding of TB, leading to impaired confidence in treatment and diminished self-efficacy . Patients with low self-efficacy also experienced greater stigma , which contributed to their elevated levels of depression. The present study is notable in that it disentangled the pathways between family function, doctor-patient communication, knowledge about TB, and depressive symptoms. Understanding these factors and their interrelationships offer an opportunity to effectively intervene in depression among patients with TB and interventions can be tailored to these specific pathways. Specifically, interventions aimed at reducing depression among patients with TB should focus on improving patients' family function, doctor-patient communication, and knowledge about TB. More importantly, the role of stigma should be understood and addressed. Given that stigma was not only a negative psychological reaction , but also mediated the effect of other factors on the mental health of patients with TB, interventions should be combined stigma-reduction measures, which is essential for promoting both physical and psychological well-being and, ultimately, better health outcomes. Specific stigma-reduction measures include routine screening for stigma of patients with TB, giving patients access to education on the concept of stigma, and offering techniques to deal with it. This study had several limitations. First, analysis of crosssectional data limits our ability to establish causal relationships among study variables. In addition, although this was one of the few studies to focus on the depressive symptoms of patients with TB in China, our sample only comprised patients with TB in a central province. Therefore, it is necessary to conduct more representative studies to determine the status of depressive symptoms among patients with TB across China. --- CONCLUSION In sum, we used SEM to explore the pathways linking family function, doctor-patient communication, and knowledge about TB to depressive symptoms and confirmed the mediating effect of stigma in the process. Awareness and management of depression in patients with TB may lead to better outcomes. There is an urgent need to develop effective interventions to prevent and decrease depression of patients. The current findings provide direction and information for depression interventions among patients with TB that improving family function and doctor-patient communication, increasing TB knowledge, and alleviating stigma may help to alleviate depressive symptoms among patients with TB. More research is needed to focus on psychological interventions for patients with TB and to assess the effectiveness of these interventions. --- --- Conflict of Interest Statement: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Objectives: To date, the complex interrelationships between family function, doctor-patient communication, knowledge about tuberculosis (TB), stigma, and depressive symptoms among patients with TB are insufficiently understood. We explored the interrelationships between family function, doctor-patient communication, knowledge about TB, TB-related stigma, and depressive symptoms and examined whether TB-related stigma played a mediating role. Methods: A cross-sectional survey was conducted between October 1, 2013 and March 31, 2014 in Hubei province, central China. Data were collected from 1,309 patients with TB using a structured questionnaire that measured family function, doctor-patient communication, knowledge about TB, stigma, and depressive symptoms. Structural equation modeling was used to examine the interrelationships among the study variables based on the hypothesized model.The proposed model provided a good fit to the obtained data. There were indirect effects between family function, doctor-patient communication, knowledge about TB, and depressive symptoms through stigma (β = -0.048, P = 0.002; β = -0.028, P = 0.001; β = -0.021, P = 0.009, respectively). Stigma partially mediated the effect of family function and knowledge about TB on depressive symptoms and fully mediated the effect of doctor-patient communication on depressive symptoms. Conclusions: This study elucidated the pathways linking family function, doctor-patient communication, and knowledge about TB to depressive symptoms and confirmed that the effect of those variables on depressive symptoms can be mediated by stigma. Those findings provide direction and information for depression interventions among patients with TB.

Introduction Artificial Intelligence -based systems are known to typically perform worse for minorities and marginalized groups . This lower effectiveness might have a concrete impact on the users interacting with these systems, such as allocational and representational harms . One of the research areas where AI-based systems are commonly used and where the analysis of these biases might be particularly relevant is the recommendation domain. Recommender Systems are software tools that help users finding relevant items. Due to their ability to adapt to users' needs, they have been applied in various disciplines . As such, they are one type of AI technique that is being increasingly used nowadays, and hence, may affect society as a whole by amplifying existing biases or guiding people's decisions. In fact, RSs are known to be multi-stakeholder environments , since they affect multiple actors in a direct way, mainly the users receiving the recommendations and those behind the recommended objects . Because of that, research on bias analysis and fairness measurements is needed; in particular, specific definitions, dependency variables, and mitigation approaches beyond those already studied for general Machine Learning . Tourism is a domain where the needs of consumers and the services offered by providers naturally meet in the real world. In the tourism industry, travel guides/blogs have always been used to organize trips. However, while travel portals and travel guides tend to focus on the most popular places , recommendation algorithms should also offer users more novel recommendations, to provide them satisfying experiences . For this reason, tourism recommendation, where AI models automatically support decision-making processes, clearly impacts on society. Hence, it is an area that is particularly sensitive to these effects and biases. Several recommendation tasks related to tourism have been addressed, such as tour recommendation to groups , trajectory recommendation , suggestion of travel packages , etc. Probably, the most important recommendation task related to tourism is the Point-of-Interest or venue recommendation problem, which focuses on suggesting to users new places to visit when they arrive in a city . The POI recommendation problem is usually defined upon data stored in Location-Based Social Networks . These social networks allow users to check-in in venues; thanks to these check-ins, platforms such as Foursquare can provide services to the users, like the possibility to share information between them, together with venue search and/or recommendation. At the same time, based on reviews, ratings, and venue check-ins available in LBSNs, users decide what to buy or consume and where to go. However, generating recommendations in LBSNs introduces new challenges with respect to traditional recommendation, such as different contextual dimensions , and a higher sparsity on the user preferences . From now on, we will refer to RSs that operate in LBSNs as Location-Based Recommender Systems . In this context, it is critical to assess the extent to which LBRSs have a concrete impact on the tourism domain as a whole. Besides the users accepting the recommendations , whose experience in a city depends on these suggestions, the business of venue owners/managers strongly depends on them. Hence, we must think of properties of a RS that go beyond accuracy, to provide equitable suggestions. Thus, RSs might be polarized towards certain undesired properties and this would concretely impact the involved stakeholders in different ways. In the end, not exposing the full catalog of candidate venues to the users might not be fair from a business perspective and may also lead to a lack of novelty and diversity in the recommendations. As a consequence, the most widely known type of polarization in recommender systems is towards item popularity, which means that only a subset of popular items is recommended to the user. Polarized recommendations towards popular venues would worsen user experience, since they might get too crowded, and it might also strengthen inequalities between venue owners/managers. Venue category can also be characterized by a certain popularity, which can impact POI recommendation and society at a broader level. Indeed, users might not be recommended possibly interesting but unpopular categories of POIs and the owners of an entire sector/type of business might be affected as a whole by it. Item popularity may also affect the exposure of the venues, since popular venues are always ranked in higher positions. Hence, these venues would increase their chances of being noticed and selected by the users , while other interesting items may go unnoticed by the user . Finally, a geographical polarization towards far away or close POIs with respect to those the user is currently visiting, might ignore their preferences and previous interactions. This polarization would affect the trust of the users on the recommender system and impact owners of more relevant venues. The problem of under-recommending and under-exposing providers is well known in the recommender systems literature , but to the best of our knowledge, it has never been studied for LBRSs. It should be clear that polarization might be related to the concept of algorithmic bias, which has been widely studied in recommender systems . Algorithmic bias assumes that RSs reinforce a previously existing bias in the data. While a pre-existing bias might be the cause of polarization, our focus is at a societal level, to study the impact of polarization for the involved stakeholders. In other words, it does not really matter in the context of this work if a venue is popular in the recommendations because it already was or because the system made it popular. Heavily polarized recommendations have a negative impact on tourism stakeholders, so we study these phenomena, without any assumption of the prior distribution of the data. To summarize, in this work, we use the term polarization to quantify to what extent an algorithm deviates from what it is observed in the training data. 1 We use the term bias to describe, in a more generic way, the inclination of an algorithm to go towards polarization. As our results will show, in LBRSs, polarization is a phenomenon that appears independently of how the data were generated. This, in particular, includes cases where data is biased towards some algorithms or sensitive features of users or items . In this work, we characterize the four previously mentioned forms of polarization through metrics that have not been used before. Then, we assess if the use of check-ins to capture the interactions of the users with a LBSN to produce recommendations may lead to polarized suggestions from these perspectives. To do this, we consider an evaluation methodology that mimics the real world, by using a temporal split of the user check-ins. We then compare different families of recommender systems to inspect these forms of polarization. In order to show to what extent a recommender might be affected by different forms of polarization, it is useful to characterize these phenomena independently. However, at the same time, mitigating these forms of polarization separately would not be adequate, since the objective is to produce recommendations that are as non-polarized as possible . As previously mentioned, each polarization affects stakeholders in different and negative ways; hence, dealing only with a form of polarization would still lead to negative outcomes. For this reason, we propose two forms of mitigation based on the concept of hybrid recommendation and re-ranking . Both approaches will allow us to deal with multiple forms of polarization at the same time by combining the outcomes of different recommenders. --- Background and related work --- Recommender systems The purpose of a Recommender System is to provide recommendations of different types of items to a particular user by analyzing their interests and tastes . These items vary considerably depending on where we apply the recommender . This wide variety of applications has led to the development of a large number of different recommendation techniques. The most extended ones are the content-based models , which exploit the features of users and items to make the recommendations, and the collaborative-filtering approaches, that can be divided into two different families. The first of them, memory-based or k-nn methods , compute similarities between users and/or items to build recommendations. The second family, known as model-based algorithms , uses the information of the interactions between the users and items in order to create a predictive model. Finally, another popular technique in the area are hybrid approaches. These methods combine different types of algorithms to alleviate the possible drawbacks that each recommender may have independently . Regardless of the recommendation algorithm, normally all of them have to deal with a fundamental problem: sparsity, that is, the ratio between the actual number of interactions made by users on items in the system and the potential number of interactions considering those users and items. Generally, this sparsity is severe, being common to work with datasets with a sparsity higher than 97% . At the same time, in classical recommendation researchers usually make use of the ratings that the users gave to the items explicitly . However, in other recommendation domains such as web, music, or Point-of-Interest recommendation, there might not be ratings available, but rather the number of times a user has visited/consumed an item . --- Location-based recommender systems While POI recommendation has the same goal as traditional RSs, there are aspects that make LBRSs different. First, the sparsity in these domains is considerable; for example, the densities, i.e., the inverse of sparsity, of the MovieLens20M and Netflix datasets are 0.539% and 1.177%, respectively. On the other hand, the Foursquare dataset we use in our experiments shows a density of around 0.0034%. Second, the use of one-sided or one-class information, where LBSNs normally only record positive values indicating that a user has visited a venue. Besides, users may checkin the same venue more than once, something that it is not considered in the traditional recommendation. And third, and more importantly, venue recommendation is highly affected by geographical, temporal , and sometimes even social influences. The former is possibly the most critical aspect to consider in LBRSs, as it is usually assumed that users prefer to visit venues that are close to each other . That is the reason why existing algorithms have incorporated geographical influence for generating recommendations . Each model incorporates these influences differently, and although there are a large number of LBRSs for an experimental survey of the state-of-theart models), many of them use traditional recommendation techniques. For example, Matrix Factorization approaches are used in the IRenMF model , which also takes into account the neighbor POIs of the target one by distance and uses a clustering algorithm to group all the POIs to model the geographical influence. Similarly, the GeoMF method , which uses two additional matrices, one to model the user activity areas by dividing the geographical space in a set of grids and the other to represent the influence of the POIs, and the LRT algorithm , which models the temporal component by factorizing the check-in matrix for every hour in a day. User-neighborhood approaches are also used in some LBRSs, like the USG model , which computes user similarities based on their check-in activities and combines them with the probability of visiting the target venue. LORE and iGLSR are two other userneighborhood approaches, which compute the similarities based on the distance of the users' residences, combined with the geographical influence modeled using Kernel Density Estimation . --- Realistic evaluation in recommender systems When evaluating recommendation quality in an offline setting, the RSs literature usually considers a random split with cross-validation methods to avoid the overfitting problem . However, a RS should be evaluated as realistically as possible, not knowing anything about future interactions, to avoid obtaining unrealistic results and avoid data leakage . Because of this, the community is slowly shifting the offline evaluation towards using temporal splits, where the recommendation algorithms should predict the present user interactions based on their past activity . However, different strategies may arise for performing such a temporal split. We can split by selecting a percentage of interactions to use in the training/test splits. A common approach would be to select the 80% of the oldest interactions to build the training set and the rest would form the test set. Other strategies would be to choose a timestamp, so as to use all interactions that happened after that timestamp for testing the recommenders. In alternative, one can order the interactions for each user separately and assign the most recent ratings of each user to the test set. Each of these strategies has advantages and disadvantages in terms of the characteristics of the training/test splits derived and how close they represent real-world scenarios. Based on these descriptions, the most realistic protocols would be those that allow for a training set temporally separated from the test set, which can be achieved by either using a common splitting timestamp for the entire dataset or by selecting a percentage of the data according to the moment of interaction. This conclusion is in line with recent analyses made by the community regarding data leakage . It is worth noting that, even if some of the existing POI recommenders perform a temporal split , to the best of our knowledge there is no thorough research about the effects of this type of evaluation split on typical recommendation approaches in this domain. --- Impact of recommender systems As described before, RSs analyze users' preferences in order to make personalized recommendations to users. However, it has been observed that sometimes the recommendations of the algorithms can be discriminatory for different groups . This effect can also cause certain types of users to receive the same type of items, isolating them according to these biases ). This was one of the main reasons to propose metrics in the field so that we could measure complementary dimensions beyond accuracy, such as novelty and diversity . One of the most recognizable biases in RSs that has received much attention in recent years is the popularity bias, which shows how the recommendations produced are generally biased to the most popular items, affecting negatively the novelty and diversity of the suggestions. Some researchers have proposed different mechanisms to palliate this problem; for example, Abdollahpouri et al. presented a regularization framework to retrieve long-tail items with a small performance loss in ranking evaluation, whereas Abdollahpouri et al. proposed re-ranking techniques to reduce the popularity bias in recommendations. Alternatively, Bellogín et al. defined two new split protocols to counter the effect of the popularity bias. Additionally, recent work has focused on the theoretical impact of popularity bias on the algorithms . In any case, this is an issue that has been studied in different domains . Our goal is to go beyond the assessment/reinforcement of pre-existing polarized data recorded in a system or biases in algorithms, to study more broadly polarization in POI recommendation. Another related topic associated with the societal impact of recommendations on the users is algorithmic fairness. A recent work by Weydemann et al. studied to what extent LBRSs can provide suggestions to groups characterized by sensitive features. More recently, Sánchez and Bellogín analyzed the recommendations of two different groups of users using LBRSs, i.e., locals and tourists, concluding that the latter suffers from a greater popularity bias. As we introduced in our motivation, polarized recommendations do not impact only consumers, but also providers . In this work, we study a broader phenomenon, which complements and does not overlap with the studies on algorithmic fairness , by providing insights on the polarization generated by different algorithms. Hence, no direct comparison is possible and the connection between this study and algorithmic fairness is left as future work. --- Polarization characterization Given the peculiarities of the POI recommendation problem with respect to the traditional recommendation, it is important to control which forms of polarization occur in this domain. In this section, we explain how to measure different forms of polarization: towards popular venues and categories , regarding the venue exposure , and with respect to the geographical distance between the user and the recommended venues. At the end of this section , we also show several toy examples to better understand the proposed polarization metrics. --- Measuring venue popularity polarization From the multiple definitions that "novelty" has in the RSs and Information Retrieval areas, one of the most commonly used definitions is that something is novel when it is not popular . To measure novelty, Vargas and Castells defined the Expected Popularity Complement metric, by computing the number of users who rated that item, divided by the number of users in the system; then, they proposed to subtract that value to 1, so that values closer to 1 indicate that the items are more novel . A similar metric called Inverse User Frequency , defined in Castells et al. measures novelty in a similar way, but considering the logarithm between the user that rated that item and the total users in the system. However, these metrics are too sensitive to the actual number of ratings, or interactions, in general, received by each item. For instance, if an algorithm always returns the same top-n items but the item distribution is too skewed, we may obtain similar novelty values between that algorithm and another one that recommends more items which have been rated by a similar number of users. Because of this, in this work, we analyze the polarization towards popular venues by analyzing the popularity distribution derived from each recommendation algorithm. In this way, we can compare whether some algorithms are more or less tailored to return more popular items. Moreover, we propose a metric that summarizes such distribution in an empirical value for each algorithm; however, since we cannot assume the inherent distribution of the data, there is no general skewness function to measure it ; because of that, we resort to empirical metrics aware of the domain we analyze. --- Definition 1 The polarization of a recommendation model rec towards popular venues is the probability that a more popular venue is ranked higher than a less popular one, when considering the top-n items recommended to a user. Our proposed metric to characterize the polarization of a model towards popular venues is computed by measuring the area under the curve generated by the cumulative distribution of the recommended items by rec; this is done by approximating the analytical integral by the trapezoidal rule. More specifically, given the unique set of items R returned by recommender rec up to cutoff n, i.e., the length of the recommendation list, for all users, we propose the following formulation to measure the venue popularity polarization: PopI@n = 1 2|m| |m| k=2 F R pop + F R pop where |m| are the items in the training set, ordering them by the number of times they have been recommended by the recommender rec. F R pop measures the cumulative popularity distribution 2 for an item x, depending on whether it belongs to R, in such a way that it is updated only for those items contributed by the corresponding recommender used to create such list R. Finally, to measure the popularity of a venue, we count the number of users who visited it divided by the total number of users that visited all recommended venues. By definition, the larger the area, the more uniform the distribution is. Hence, this metric produces lower values for recommenders polarized towards popular items, which is bounded in [0, 1] thanks to the trapezoidal rule applied on a square [0, 1] × [0, 1]. It is important to note that obtaining a high value in this metric does not imply that the ranking accuracy of the recommendations is higher, it implies that more items with different popularity values are being recommended. Therefore, in order to obtain the "expected" value of this metric, we should compute it with the data available in the test set as it represents the real visiting patterns of the users in the dataset. We shall do this later in the experiments by contrasting the behavior of recommenders against a method that provides suggestions based on the test set. --- Measuring category popularity polarization Intuitively, a user who likes rock music would probably prefer recommendations of groups such as Led Zeppelin or the Rolling Stones rather than classical music. In the case of POI recommendation, users may prefer some venues over others depending on the venue type. In this domain, the venue type is unambiguously linked to the venue category, such as restaurant, museum, public park, etc. Because of this, it is important to consider the polarization with respect to well-known groups of items, such as genres in movies or music, venue categories in POIs, or verticals in e-commerce. Moreover, the interactions between users and these groups of items are not uniformly distributed in typical recommendation systems, and in particular in LBSNs, as we show later for different cities with respect to venue categories. Hence, it is important to distinguish the popularity of a specific POI from that of the associated categories . --- Definition 2 The polarization of a recommendation model rec towards popular categories is the likelihood of recommending venues belonging to categories associated with the highest number of user interactions. We analyze the polarization towards popular categories by grouping the top-n recommended POIs by each category, while sorting the different categories by increasing popularity, measured as the number of interactions each category has received in the entire dataset. Thus, we summarize this analysis in the following metric value: PopC@n = 1 min i∈L bin cat |{cat}| PopC@n = 1 |U | u∈U PopC@n) where, as before, n denotes the cutoff at which we measure the metric , R denotes the top-n recommended items to user u by recommender rec. Note that |L| and n will be, in general, equal, since L = R, except when the recommender has a low recommendation coverage . In case of low coverage, |L| might be smaller than n, that is why we prefer to make this situation explicit in the formulation. Here, cat returns the associated category to each item, and bin returns the category bin, where the least popular category is associated to the first bin = 1) and popular categories are assigned the last bins. In this way, a larger value is obtained for popular categories and we can use this as an indicator of how polarized an algorithm towards popular categories is. We consider the number of categories, |{cat}|, to be fixed. 3In those cases where the category information is not available, an implicit clustering of the venues might be used . The metric is in the [0,1] range, where 0/1 indicates that a model only recommended venues associated with the most unpopular/popular category. As in the previous case, to obtain the expected value, we should compute this metric with the data available in the test set. --- Measuring polarization in terms of item exposure When measuring the quality of a recommender system, in most cases only the users' opinions are taken into consideration, either in terms of relevance or other dimensions such as novelty and diversity. However, the perspective of the items should be equally important because we may be over-representing the most popular items in the recommendations . For several years, researchers in the recommender systems area have analyzed the effect of over-representing the most popular items, observing that the most unpopular items actually belong to the long-tail item distribution . Although a large number of users consume popular items, according to Anderson , vendors should focus on such long-tail items as unpopular items are often more profitable. In the POI recommendation domain, the items are venues, ranging from major tourist sites to minor ones, e.g., food establishments, bars, or small businesses. By recommending less popular venues in the long-tail, we may introduce users to new places that they had not thought they might be interested in, and also make these less popular sites receive more visits, which means that they end up having more customers. As these venues are sometimes businesses that generate trade activity in the cities, a poor exposure of these venues might negatively affect the city's economy. --- Definition 3 The polarization of a recommendation model rec in terms of exposure is the likelihood of the model to suggest a venue proportionally to the number of times the users will consider that venue in the future. While, in the characterization of item popularity, we assessed the probability of recommending a popular item, in order to measure the exposure of venues, we compare the number of times an item has been recommended against its actual exposure . However, differently from the metrics proposed by Ariza et al. , instead of dividing RE and AE, we will compute the squared difference since it is a more common mechanism to measure errors, as in Ekstrand et al. : R E@n = 1 |U | u∈U 1/log 2 ) + 1) j∈R 1/log 2 ) + 1) AE = p I E@n = i∈I -AE) 2 where R denotes, as before, the top-n recommended list by rec for user u, whereas pos denotes the position of item j in a recommendation list L. Finally, π denotes the exposure of the item under the target policy . In this paper, we will work with two different policies-see for an overview of reasonable choices over these policies-, namely: Parity, where we assume that all items should be recommended equally, i.e., following a uniform distribution, and Relevance, in which we assume that each item should be recommended following the same distribution observed in the test set. According to this, I E would denote the final Item Exposure based on a target policy π , for a recommender rec measured at cutoff n. Hence, the lower the I E, the better the recommender is. --- Measuring polarization towards geographical distance According to the first law of geography, "Everything is related to everything else, but near things are more related than distant things" , which is why many LBRSs model geographical influence. Because of this, exposing the polarization towards this aspect might be a critical signal of the type of venues provided by a recommendation algorithm. Despite the fact that geographic influence has been used extensively to make recommendations to users, we have not found many works that analyzes the geographical relationship between the actual recommended POIs . Hence, we consider this analysis an important contribution of the presented work. --- Definition 4 The polarization of a recommendation model rec towards geographical distance is the likelihood of the model to suggest a venue that is close to/far from the current position of the user. As a first approximation, we propose two metrics that consider the distance of recommended POIs in their evaluation. The first one, DistT, shown in Eq. , sums the distance of the recommended POIs as if the user accepted those recommendations and visited those venues in order. 4 With this metric, we account for the polarization towards longer or shorter recommended routes or trajectories, even though this metric could be applied to any type of recommender system, not only for those producing routes. The second metric, DistU, shown in Equation , computes the total distance between each recommended POI and the user historical midpoint, obtained by averaging the coordinates of every venue visited by the user in the training set. In this way, we aim to capture how sensitive each recommendation algorithm is to the history of previous locations of the user. This concept connects to the recent literature on calibrated recommendations , by studying how adherent the recommendations are to the previous behavior patterns of the users, which in our case are modeled by their locations. Note that Equation 8 cannot be used if the user has not checked-in in any venue in the training set. However, in a real environment where a tourist arrives at a city, instead of her midpoint we could make use of the coordinates of the venue she is staying at or the actual geographical position of the user. DistT@n = min i=2 Hav DistU@n = min i=1 Hav where R u,i is i th item recommended to user u, u m is user u historical midpoint, and Hav is the Haversine distance of the coordinates of two geographical points. In order to interpret the geographical polarization values found using these metrics, we need to compare those values obtained by the recommenders we are analyzing with respect to those found using the user's ground truth. Thus, obtaining high values in these metrics might not be intrinsically bad if the users actually exhibit those mobility patterns in the ground truth. However, obtaining very different values from those exhibited by the users in the test set would be a sign that the recommenders are actually showing a geographical distance bias far from the expected one. --- Toy examples In this section, we show a toy example for every proposed metric to illustrate how they work, in order to help the reader to have a better understanding of all the different analyzed polarizations. First, in Fig. 1, we compare the performance of two different recommenders using our Venue Popularity Polarization metric , and we compare it against other novelty metrics like EPC and IUF . As we can observe in that figure, both recommenders would obtain the same values in terms of Expected Popularity Complement or Inverse Footnote 4 continued that the lower an item is ranked, the lower is the likelihood that the user will choose it. So we can assume that the ranking generated by a recommender system is a proxy for the sequence of choices for the users. Fig. 1 Visual example of the popularity polarization of two different recommenders, rec 1 and rec 2 . The second recommender would obtain higher values in our metric due to the fact that it is recommending more different venues, and hence the area under the curve would be higher than in the first recommender Fig. 2 Visual example of the category popularity polarization of two different recommenders, rec 1 and rec 2 . The second recommender would be preferred as it is recommending venues from different categories User Frequency because they are recommending items that have been rated by the same number of users . However, the second recommender is able to recommend both the black and white items while the first one is not. Thus, the area under the curve of the second recommender would be higher, as it is recommending a higher number of items, showing less polarized results. Secondly, in Fig. 2, we show a comparison between two recommenders in terms of Venue Category Polarization. Both of them recommend three different items, but the first one is only recommending items with the feature denoted as "A", which is the most popular one. On the other hand, rec 2 is recommending venues belonging to all categories, and hence obtaining a lower category polarization. Thirdly, in Fig. 3, we show a comparison between two different recommenders using our formulation for Venue Exposure Polarization by applying a relevance-based target policy or ideal exposure. In that example, we observe that the second recommender Fig. 3 Visual example of how the value of Item Exposure changes according to the behavior of recommenders, for a relevance-based policy. Here, rec 1 and rec 2 denote the first and second recommenders, R denotes the recommendations from recommender rec n for user u, while T u 1 and T u 2 represent the test set of the two users. In this situation, rec 2 would obtain a lower value because it is recommending the black item 2 times, as in the test set, and it is not recommending the dotted item, which does not appear in the test set. Hence, as the recommended items from rec 2 are more similar to the ground truth of the user than the ones recommended by rec 1 , the venue exposure polarization of rec 2 would be lower obtains a lower result in terms of exposure than the first one due to several reasons. Firstly, rec 2 is not recommending one of the items , which actually does not appear in any of the test sets. Secondly, this model is also recommending the black item twice, which is the same number of times that item appears in the test set; however, the first method only recommends this item once. Finally, rec 2 is the only model that recommends the item with vertical lines; moreover, this item appears in as many recommendation lists as in the test set. Hence, rec 2 achieves the expected exposure for this item, and the value of IE is decreased since both RE and AE are closer to each other. Finally, in Fig. 4, we show a comparison between two recommenders in terms of our Geographical Distance Polarization metrics . In this example, the second recommender would obtain lower values in the metrics as the recommended venues are closer with respect to the user midpoint and also closer between them than the recommendations produced by the first algorithm. --- Evaluation settings --- Evaluation methodology We performed experiments on the Foursquare global check-in dataset5 used in . This dataset is formed by 33M check-ins in different cities around the world, Fig. 5 shows the 50 cities with the highest number of check-ins. We selected the check-ins from the cities of Tokyo, New York, and London from this dataset and, once we selected the check-ins of all three cities separately, we performed a 5-core, that is, we removed both users and POIs with less than 5 interactions. Next, aiming for a realistic evaluation, we split the check-ins so that the 80% of the oldest interactions were used to train the recommenders and the rest 20% to test them. Fig. 4 Visual example of the geographical distance polarization of two different recommenders, rec 1 and rec 2 . In this example, the second recommender will be preferred as the recommended venues are more geographically related between them and with respect to the user midpoint Fig. 5 Plot showing the 50 most popular cities in the Foursquare dataset before preprocessing. In black, the cities of Tokyo, New York, and London are highlighted The statistics of the datasets and their splits are shown in Table 1. Finally, we removed from the test set all interactions that appeared in the training set and the repetitions, that is, we consider that the users just visit the same POI once in the test set. These evaluation methodology issues, combined with the sparsity of the dataset and the fact that we do not force test users to have a minimum number of training interactions, means that the results in terms of ranking accuracy will be low. However, we decided to not focus only on those users with enough locations visited in their profile, as this would make our experimental analysis too limited. However, we leave as a future work the analysis of cold-start users . Moreover, for the training set, we maintain three different versions due to the intrinsic characteristics of some of the aforementioned models: the one with repetitions , the one adding all interactions , and the one binarizing all possible user-POI interactions . Please note that in this dataset there are no explicit ratings as we typically find in classic recommendation datasets, such as MovieLens. In Foursquare, we only know when a user has visited a certain POI, unlike other LBSNs such as Yelp, 6 where we do find ratings and reviews. Hence, the training set with repetitions is being used by the recommenders that build sequences for performing the recommendations. The frequency training set is being used by the recommenders that can exploit the explicit information, to give more importance to those interactions with a higher score. In this case, by aggregating the check-ins, we can obtain a frequency matrix that can be used in the models as if it was the classic matrix of user ratings. However, these frequencies are not entirely comparable to ratings because they are not bounded at the system level . Finally, the binarized training set is used by both the implicit and explicit recommenders. This final training set will denote with a '1' if a user has visited a particular POI and will present a '0' otherwise. For generating the recommendations, we follow the TrainItems methodology , i.e., we consider as POI candidates for a target user u those venues that appear in the training set but that have not been visited by u. --- Recommenders In order to analyze and characterize the biases that may exist in the Foursquare dataset, we now describe the state-of-the-art algorithms that have been considered in our experiments, grouped in different families: -Non-Personalized: we tested a Random and a Popularity recommender. The latter recommends the venues that have been checked-in by the largest number of users. -Collaborative-filtering: we used a User-Based and an Item-Based collaborative filtering algorithm. We also included a matrix factorization algorithm that uses Alternate Least Squares for optimization from , and the Bayesian Personalized Ranking using a matrix factorization approach from . For the BPR, we use the MyMediaLite library.7 -Temporal/Sequential: we include a user-based neighborhood approach with a temporal decay function , and several algorithms based on Markov Chains: Factorized Markov Chain , Factorized Personalized Markov Chains and Factorized Item Similarity Models with high-order Markov Chains . All three Markov Chains approaches are obtained from . -Purely geographical: we used the Kernel Density Estimation from , and a recommender that suggests to the user the closest venues to her centroid . -Point-of-Interest: we used the fusion model proposed by Cheng et al. that combines the Multi-center Gaussian Model technique with Probabilistic Matrix Factorization , a POI recommendation approach from that uses BPR to optimize the model , a weighted POI matrix factorization algorithm from , and a hybrid POI recommendation algorithm that combines the UB, Pop, and AvgDis recommenders . We also include a perfect recommender that uses the test set as the ground truth, named Skyline. This recommender will return the test set for the user, in order to check the maximum values that we can obtain with ranking-based accuracy metrics . At the same time, it helps to evidence the biases and polarizations that already exist in the test split. --- Metrics Since we have already defined in previous sections our proposed metrics to measure different types of polarization, we will now show the formulation of the metrics used for measuring the item accuracy, novelty, and diversity. -Accuracy: oriented at measuring the number of relevant items recommended to the user . We will use Precision and the normalized Discounted Cumulative Gain : -Precision: P@n = Rel u @n k where Rel u @n denotes the set of relevant items recommended at top n. -nDCG: nDCG@n = DCG@n IDCG@n DCG@n = n k=1 2 rel k -1 log 2 where rel k denotes the real relevance of item k in the test set. In a rating-based dataset, this real relevance would be the rating that the user gave to that item in the test set. In our case, as we only know whether a user has performed a check-in, we fix this ideal relevance to 1 as long as the venue appears in the test set of the user . Higher values of P and nDCG imply a better recommendation quality. -Novelty: oriented at measuring the number of popular venues, since they are inversely related to novel venues . We use a simplified version of the Expected Popularity Complement metric: -EPC: E PC@n = C min i=1 ) (12 ) where C is a normalizing constant (generally C = 1/ min i=1 ). In our case, p = |U i | |U training | , with U i being the number of users that checkedin in venue i and U training the set of users in the training set. Higher EPC implies better recommendation novelty. -Diversity: oriented at measuring how many different venues we are recommending to the user . We use the Gini Coefficient to measure the diversity. -Gini: Gini@n = 1 - 1 |I| -1 |I| k=1 p p = |{u ∈ U|i ∈ R s u,n }| j∈I |{u ∈ U| j ∈ R s u,n }| where p is the probability of the n-th least recommended item being drawn from the recommendation list generated by s, that is, when considering all rankings @n for every user. In this paper, we will use the complementary of the Gini Index proposed in Castells et al. , as defined in Vargas and Castells . Higher Gini implies better recommendation diversity. -User Coverage: aims to measure whether the recommender system covers all the users or items in the catalog . We focus on the User Coverage , that accounts for the number of users to whom at least one recommendation is made. This metric is useful because there might be some models that are not be able to recommend to all users of the test set . Higher user coverage means that our model is able to recommend to more users. --- Polarization assessment Tables 3,4, and 5 show the results of the aforementioned recommenders in terms of accuracy , novelty , diversity , and our metrics to measure popularity polarization , item exposure , polarization towards geographical distance , and user coverage . Recall that higher values indicate better accuracy, novelty, diversity, and coverage. On the contrary lower values of popularity polarization , exposure, and distance measure the optimal situation with less polarization. The parameters tested of the recommenders can be found in Table 2. We selected the best configuration of each recommender according to nDCG@5 obtained in the test set 8 . In order to validate the different forms of polarization we presented, we performed three sets of experiments: 1. Impact on accuracy metrics. Before assessing polarization, we evaluate the models shown in Sect. 4.2, considering the metrics presented in Sect. 4.3. This will allow us to assess the behavior of these models from accuracy and beyond-accuracy perspectives, to then contextualize it to the polarization these models generate. 3 123 2. Measuring recommendation polarization. We address to what extent the considered recommendation models are polarized towards the four perspectives considered in this work , by measuring the metrics proposed in Sect. 3. 3. Polarization mitigation. We evaluate the capability of hybrid and re-ranking mitigation strategies to counter polarization. Since no validation set is used in these experiments, the reported performance is an overestimation. Such an experimental setting is not uncommon in recommender systems, especially when dealing with temporal splits as we have here . In what follows, we analyze these perspectives in depth. --- Impact on accuracy metrics The analysis of these results highlighted some interesting behaviors. First, we observe in Tables 3,4, and 5 that the Skyline does not have full coverage for the users and it is not obtaining a value of 1 in the accuracy metrics. This is because we follow the TrainItems methodology and therefore the items that did not appear previously in the training set cannot be recommended. Besides, there might be some users that have a smaller number of relevant items than the used cutoff. These two reasons could prevent some metrics from obtaining a perfect score. Regarding the rest of the algorithms, we observe that one of the best performing recommender is the Pop recommender in all cities, even though in Tokyo the TD model and in London the GeoBPR and PGN models obtain a slightly better value than Pop. This could be due to several causes, including the high sparsity found in the datasets, the test set that only contains new interactions , and the temporal evaluation methodology, as there could be users in the test set that do not appear in the training subset . This is an interesting conclusion, because it is a clear sign that this algorithm, despite its simplicity, is able to beat more complex models that incorporate temporal and/or geographical influences. However, this is somewhat surprising, because despite being such a competitive baseline, it is not so common to analyze the performance of this baseline in POI recommendation . Indeed, the authors of IRenMF and the FMFMGM did not test their approaches against the Pop recommender. With respect to the POI algorithms, we observe that, in terms of accuracy, their performance is very similar to other classical approaches, like the UB or the BPR. This may be due to the high number of both hyper-parameters and parameters that these models have, making it sometimes difficult to find a good configuration of hyperparameters that obtains a decent performance. In fact, it is interesting to highlight the low values achieved by the FMFMGM algorithm in New York and London, while in Tokyo it is competitive against other models. This demonstrates that although we might find good configurations in terms of accuracy, the parameter settings in some circumstances is critical. In the end, classical proposals such as those based on neighbors, might be easier to explain and optimize due to its simplicity and lower number of parameters . This also affects the PGN recommender since, despite its simplicity, its performance is rather high. In New York and London it is the best recommender of the POI family and in Tokyo it has a very similar performance to IRenMF. The low number of parameters of this recommender, combined with the fact that it merges different sources of information such as popularity and geographical influence, may be the reason for this behavior. --- Measuring recommendation polarization When measuring the distance , we observe that both Rnd and Pop algorithms obtain high values, showing us that the recommended venues of these models are far from each other. Analyzing this geographical information is also important because, as we observe in the Skyline, users tend to visit POIs that are relatively close to each other , meaning that the distance between the relevant items, and also between the recommended items and the user's center, should be low. Nevertheless, the geographical influence alone is not enough to obtain high values in terms of relevance, as evidenced by the poor performance of the pure geographical algorithms . At the same time, if we analyze the rest of the recommenders, we observe that, although all of them seem to perform personalized recommendations, regarding PopI, PopC, ExpP, and ExpR metrics we observe a pronounced popularity bias. Let us focus, for example, in the PopI and exposure metrics. The only recommenders with decent values of accuracy that seem to obtain high values on these metrics are PGN and IB, while the rest only obtain results slightly higher than Pop. In fact, when analyzing the exposure metrics , the random recommender obtains lower values in terms of ExpP than all algorithms due to the fact that it recommends items in an arbitrary manner, without overrepresenting any subset of items. Similarly, this recommender obtains good results in the ExpR metric because it is recommending almost all the venues in the system, so it is very likely that within those recommendations there are relevant venues. However, what the Rnd recommender fails is in recommending the relevant venues to the correct users, as discussed before regarding the accuracy metrics. Hence, we conclude that most of the recommenders suffer from a great popularity bias, evidencing the difficulty of finding good representatives for all metrics. Therefore, among all the experimented recommenders, we consider IB and PGN to be of particular interest, since even though they do not perform as well in terms of accuracy as Pop, they obtain competitive results in terms of other metrics like novelty, diversity, and item exposure; this is a direct consequence of suffering less from the popularity bias. Let us now analyze the effect of the popularity and the categorical polarization more in detail. Figure 7 shows the cumulative plot of the cities of Tokyo , New York , and London of the most representative recommenders shown in Tables 3,4, and 5, showing the 30% of the most popular venues. For this selection, we considered those models with better values in any evaluation dimension that belong to different families. By considering those results, we observe that some of the most competitive recommenders like UB, TD, and BPR are just basically returning the most popular POIs . At the same time, those recommenders that are able to obtain a higher area under the curve than the one obtained by the Skyline are the worst in terms of performance . This is a worrying result that departs from the results previously reported for some recommenders in terms of classical accuracy metrics, which slightly differed from the Pop algorithm. However, when the recommended items are analyzed, a clear, strong popularity bias is observed. In order to better visualize this effect, in Fig. 9, in the left column, we show the distribution of the top 30% most popular venues in the three different cities. As we can observe, despite showing only 30% of the most popular venues, most of the checkins are concentrated in the most popular ones, leaving a large number of other venues in the long-tail unexplored. If, for example, we analyze the same distributions at the user level , we can observe how the distribution is not so unbalanced, although we can find that there are a considerable number of users who have made very few check-ins. Nevertheless, we believe there is potential in combining different types of algorithms to see if it is possible to maintain an adequate level of accuracy while increasing at the same time the performance of other metrics such as novelty, diversity, or item exposure. 9); due to space restrictions, they will be presented using their numerical IDs. We first show in the right column in Fig. 8 the distribution of the venue categories in the training set of the three cities. With this image, we want to show that the categories are not distributed uniformly and that venues related to both transport and food are the most numerous in these cities, while the number of check-ins in residences is negligible. Taking this into account, we show in the right column of Fig. 8, the distribution of the categories of the recommended venues by our models using a cutoff of 5, that is, only the top-5 items recommended by each of those models are considered when measuring PopC. In these figures, we observe that the popularity of a category is not always associated with the number of POIs that share that category; more specifically, category 7 concentrates the largest number of check-ins in the city of Tokyo, while Fig. 8 In the left column, we represent distribution of the categories that appear in the top-5 recommended items for each algorithm in the training set for Tokyo , New York , and London . In the right column, we show the distribution of the categories of the venues in the cities following the same order. The category bins in the latter case are ordered by increasing category popularity category 3 is the second most popular category; however, since this category covers a large number of different venues, those recommenders with a strong item popularity bias recommend almost no POIs from this category, since its corresponding items are not globally popular. A similar behavior is observed in New York, where category 3 is the most popular one in the number of check-ins but most personalized recommenders do not suggest as many items belonging to that category as those from categories 7 or 1. Interestingly, the analysis of the category bias allows discriminating between those recommendation methods that seem to have the same popularity bias, according to Fig. 7. For instance, it is now more clear that Pop and BPR are recommending practically the same items. At the same time, IRenMF and UB also include some of the least popular categories, evidencing different patterns on the recommendations that, as we will discuss later, prompts different effects on the accuracy of these algorithms. Finally, those techniques with a less pronounced category bias exploit very different sources of information: Skyline uses the test directly, KDE exploits the geographical coordinates, IB computes collaborative similarities between items , and Rnd. This is an indication that the mitigation of these types of biases requires additional information sources. These additional sources should, in any case, be balanced with relevant recommendations, since the risk of providing not interesting items is higher for less popular categories; for instance, Skyline and Rnd show similar plots but have very different accuracy levels. --- Polarization mitigation As we observed in the previously reported results, it is impossible for one algorithm to obtain the best performance in all reported metrics. In fact, the Skyline, which would represent the best recommender in terms of accuracy, performs worse than the Rnd recommender in terms of novelty and diversity. For that reason, and considering accuracy as one of the most critical dimensions to optimize, we aim to combine several algorithms to create models that obtain decent levels of accuracy while overcoming the analyzed polarization measurements: popularity, exposure, and geographical distance. In order to do so, we propose two different but complementary approaches to mitigate the aforementioned biases. As a first approach, we create hybrid recommenders by combining several models ; we apply simple models based on weighting differently each of the combined recommendation algorithms. By means of these weights, we will be able to enhance the quality of the recommendations by balancing the contribution of the different models depending on the evaluation dimension that we are interested in maximizing in that particular moment, either ranking accuracy, novelty, or diversity. In our second approach, we make use of reranking techniques popular in the Information Retrieval and Recommender Systems fields to address the tradeoff between accuracy and diversity . In our context, we use these techniques in order to rearrange the top-n recommended items by an algorithm according to another recommendation technique. The objective of both proposals is to generate new recommendation lists that are capable of maintaining acceptable levels of accuracy, while improving performance in other dimensions, such as novelty , diversity, or geographical variability, thus mitigating some of the desired biases. It should be noted, however, that all these measurements depend on having a test set as reference. Such a set may contain biases itself, hence limiting the generalization and impact of the proposed techniques. Collecting and using unbiased datasets is out of the scope of this paper, but it is a direction worth exploring in the future. To define our hybrid approaches, we assume we have collected the top-n lists of a set of recommenders, denoted as R, and a weight vector W , so that R j ∈ R denotes the recommendations for all the users of the j-th recommender, and w j ∈ W denotes the weight for that recommender. As every recommender may have a different range , we first combine all the recommendation lists using the min-max normalization. The final score user u has for item i is computed as: s = |R| j=1 w j s -min max -min 123 where s provides the score of item i within the recommendation list L, whereas min and max denote the minimum and maximum score of the list for user u by recommender R j . Moreover, instead of using all the recommended items from each method in our hybrid formulation, we decided to use the top-100 items of each recommender being considered. This top-100 selection is only used for generating recommendations, i.e., it is independent of the cutoff used to measure the quality of the recommendations. On the other hand, we base our re-ranker approach in the xQuAD framework . Considering this, our proposed model can be formulated as follows: f obj = λ • f R j + • f R k where R j and R k are the two RSs to be combined , f obj is the objective function to be maximized. Consequently, the final score of item i is a combination of the ranking position in the original recommender R j and the second recommender R k used to re-rank using the combination parameter λ. In both cases, we use a score derived from the one presented before for the hybrid approach, that is, f R = rank, R u ). Then, a new ranking is created by sorting the combined scores obtained through the objective function. As in this case we re-rank a recommendation using another algorithm, we need to restrict the number of items even more. Otherwise, the second method may push items that are not very relevant since they were originally very low in the ranking. Thus, we consider the top-20 items from R j . Even though both approaches may seem similar, there is a substantial difference between them. While in the hybrid approach we combine two independent recommendation lists, in the re-ranking approach the candidate items come only from the first recommender, i.e., the re-ranked items by the second recommender belong to the first model. Additionally, we can only apply the second approach to a pair of RSs; hence, for the sake of comparability, we restrict the size of the set R to hybrid recommenders of size two, although in the future, we would like to investigate how to combine larger pools of recommenders. Hence, based on the proposed approaches, we present in Tables 6,7, and 8 the results for the cities of Tokyo, New York, and London of the following recommenders: Pop, UB, TD, IRenMF, and PGN. We decided to select these recommenders because they are the ones that achieve the best values according to the accuracy metrics. For each recommender, we show three configurations regarding the hybrid approaches denoted as H, where each model is combined with the IB recommender with different weights. These weights are designed to balance the contribution of each model in the final recommendations. As there might be a large number of possible configurations, we decided to focus on three weights: 0.2, 0.8, and 0.5. These weights allow us to explore the effect in the recommendations when giving less importance to the first recommender , the same weight to both models , and more importance to the first algorithm . Thus, for example for H, the final score of every item is created from Pop recommender and IB recommender contributing 20% and 80% to the final score, respectively. We also include one re-ranker configuration, denoted as RR, where, as explained before, the IB recommender is used to re- 6 rank the top 20 recommended items from each method. The reason why we selected the IB approach is straightforward: it is the personalized recommender that achieves the best values in novelty, diversity, and exposure while not being the worst in terms of accuracy. When analyzing these results, we notice some interesting outcomes. In New York, we observe that the best recommender in terms of accuracy is still the pure Pop model, however, when using the hybrid IB with a weight of 0.5 we reduce the popularity bias while improving almost in half the exposure values. Better mitigation results are obtained when the weight on IB is higher, but in that scenario, accuracy metrics decrease by more than a 37% . For the rest of the models in this city we do observe that using a hybrid with a weight of 0.2 in the IB component allows us to alleviate most of the biases while also obtaining slightly higher values in terms of accuracy. This is particularly interesting because we are able to maintain similar levels of accuracy while improving significantly the results obtained in terms of novelty, diversity, and polarization mitigation using such a simple technique. With respect to comparing the performance of the re-rankers with the hybrids, we can observe that, in general, re-rankers obtain comparable results to those of using a weight of 0.5 for the hybrids, which might be reasonable since the IB re-ranker can only modify the ranking of the top-20 items returned by the recommender, so the original recommendations still maintain a strong effect in the final ranking. It is important to note that, regarding the geographical polarization, we observe that in the case of New York we are able to reduce this bias when using a weight of 0.8 with the IB approach in the hybrid model or when using the re-ranker . However, the reduction of the bias in these metrics is still far from the values reported in the Skyline of Table 4. In fact, it should be noted that any reduction of this bias would be surprising considering that the IB recommender does not include any geographical component. Regarding this, we performed experiments considering the KDE as a candidate algorithm to build the hybrids and the re-rankers. However, we observed that when we reduced the distance of the recommended venues to the user, the accuracy of the recommendations decreased significantly. For example, in New York, we observed that when using our reranking approach, the performance in terms of ranking accuracy decreases, for all recommenders, more than a 50%, evidencing that the KDE is not a good method to be used with these mitigation proposals. The results for the Tokyo dataset, shown in Table 6, confirm a very interesting case where the best algorithm in terms of accuracy outperforms the best recommender reported in Table 3 . Here, the best performing configuration is the PGN with the IB re-ranker. Although this is a promising result, we observe that in this case, the re-ranker is obtaining lower values in terms of novelty and diversity while suffering from a larger popularity bias . Nevertheless, there is one example that shows a very good tradeoff among all the metrics: H. In this case, it also obtains a higher performance than the pure PGN; more specifically, we are able to improve the accuracy a 5.88% in terms of P while reducing the ExpP and ExpR by a 30.9% and a 33% respectively when compared against the result obtained by the PGN. In the case of the city of London, we observe in Table 8 that the best performing configuration in terms of P is the Pop algorithm with the item reranker, and in terms of nDCG is the PGN combined with the IB with a weight of 0.2. However, the most important conclusion about these cases is that we again managed to improve performance in terms of ranking accuracy , maintaining similar values in novelty and diversity while reducing exposure polarization. This indicates that, as long as we have a test set available, we are able to increase the performance of the different models in other dimensions without degrading the accuracy ranking dramatically. The geographical polarization, on the other hand, is more difficult to improve, as discussed for the New York city. However, all these examples confirm that it is possible to find configurations where better results than the original recommenders are obtained, either in terms of accuracy while keeping similar polarization values, or reduced polarization measurements while keeping comparable accuracies. --- Conclusions and future work Research on the characterization of biases in Artificial Systems in general, and Recommender Systems in particular, is an area of growing interest. In this work, we have focused on polarization, that is, how far an algorithm deviates from what was observed in the training data. We have characterized four types of polarizations in Location-Based Recommender Systems, a specific type of algorithms that suggest points-of-interest to users, by exploiting their preferences and other inherent characteristics from the touristic domain, such as location and item categories. This type of suggestion is one of the main means for users to explore a city and the business of venue owners is directly affected by them, hence providing equitable recommendations is a key aspect that may have a concrete impact on society. In detail, we have analyzed the popularity polarization , the exposure of venues, and the polarization related to geographical distance. After the characterization, in the experiments, we have assessed these different sources of polarization by comparing several state-of-the-art recommenders. Our results show that popularity polarization is prevalent in many of these recommendation algorithms, both in generic or tailored approaches for location-based recommendation. In terms of exposure and distance, there is a difficult tradeoff to satisfy with respect to accuracy. This is, as discussed in the paper, tied to the test set available which may itself contain bias. Finally, we propose two techniques based on combining recommendation algorithms that have demonstrated promising results to mitigate the analyzed polarizations. In particular, for some cases, these approaches are able to improve accuracy while reducing the observed polarization. However, this effect depends on how the recommenders to be combined are selected and also on the test set used to analyze the quality of the recommendations. That is why, in the future, a deeper analysis is necessary to be performed so that other families of algorithms are also included. In particular, more dynamic approaches based on sequences or other contexts available in the tourism domain might have different levels of sensitivity to these biases. Similarly, we believe the polarization assessment performed herein should be extended to analyze how it affects groups of different users, for example, according to sensitive attributes such as user gender, age, or ethnicity. In the same way, a more automatic approach to detect which recommendation algorithm should be used to be combined with when using the proposed techniques, needs to be analyzed to scale these approaches to larger datasets or other recommendation tasks. At the same time, we would like to explore other strategies for reducing the polarization of recommendations without the need for the users' ground truth, so that the polarization reduction is not so dependent on the test set. Indeed, besides the algorithmic bias discussed in the introduction, another popular source of bias is the fact that the data could be collected in a biased way, or that users interact with the system in such a way that biased interactions are recorded . In this paper, we aimed at understanding how biased or polarized the recommendations depending on the algorithm are, since, even starting from the same data, some recommenders may output more polarized results than others. However, this only relates to training data, but this could also affect the test data, since the original data from where the training and test splits are generated are the same. To the best of our knowledge, there are not many feasible and realistic solutions to this aspect, and the community is still working on it. One possibility would be to collect complete and unbiased datasets. This has been done for specific domains , evidencing the very high cost it is required for such constructions. We may also focus on specific subsets of users or items , however this is not guaranteed to reduce the bias in the data, and may have generalization problems. A potential solution that would require further analysis and proper formalization is the use of simulations to generate synthetic data without biased ground truth. However, this alternative would depend on the possibility of generating realistic user interactions, which is something quite challenging, even more for location-based information . Finally, other ways to mitigate these and other biases should be explored in the field of Point-of-Interest recommendation, beyond exposure polarizations, such as selection biases-where the observed interactions are not a representative sample of all the interactions-and feedback loop effects-where the exposed items by the recommender are used as training data for the same recommender, intensifying the biases over time- . --- Data availability The dataset used in this paper is publicly available. Likewise, the cleaned data will be provided once the paper is accepted. --- Code availability The code will be available once the paper is accepted in the following Bitbucket repository https://bitbucket.org/PabloSanchezP/BiasMitigationLBRs/. --- --- Appendix As already discussed in Sect. 5.2, in Fig. 9 we show the distribution of the top 30% most popular venues in the three different cities . On the right column of this figure, the check-in distribution performed by users is depicted. It is remarkable how strong the long-tail effect is in both situations, meaning that there are items and users that concentrate most of the check-ins. Fig. 9 In the first column, we show the distribution of the 30% of the most popular venues in the cities of Tokyo , New York and London . In the second column, we show the distribution of the check-ins performed by the users in the same cities Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Location-Based Social Networks stimulated the rise of services such as Locationbased Recommender Systems. These systems suggest to users points of interest (or venues) to visit when they arrive in a specific city or region. These recommendations impact various stakeholders in society, like the users who receive the recommendations and venue owners. Hence, if a recommender generates biased or polarized results, this affects in tangible ways both the experience of the users and the providers' activities. In this paper, we focus on four forms of polarization, namely venue popularity, category popularity, venue exposure, and geographical distance. We characterize them on different families of recommendation algorithms when using a realistic (temporalaware) offline evaluation methodology while assessing their existence. Besides, we propose two automatic approaches to mitigate those biases. Experimental results on real-world data show that these approaches are able to jointly improve the recommendation effectiveness, while alleviating these multiple polarizations.

Introduction Since the advent of the Internet, the quantity of digital data describing our behavior has inflated, offering to scientists an unprecedented opportunity to study human interactions in a more quantitative way. This opened the field of sociology to data-analysis and from the hard-science community, came the tacit idea that several aspects of the complex human behavior can be modeled [1][2][3][4][5][6]. With the rapid development of mobile technologies a lot of effort was first put in trying to capture the patterns of human mobility . A more local picture of our everyday social interactions can be obtained using dedicated proximity sensors. Following a pioneering experiment that equipped conference participants with pocket switched devices [8,9], the sociopatterns collaboration developed some wearable sensors that allow to register the complex patterns of face-to-face interactions [10,11]. The radio-frequency signal is only recorded if two individual are in front of each other for a duration of a least 20 s . We note that, from a sociological point of view, a distance below 1.5 m covers the traditional private , personal and social zones. The goal is not only to analyze social interactions but also to understand how information spreads over a real dynamical network [12][13][14][15]. Those sensors were worn by volunteers in several work-related environments: scientific conferences [10,12,13], a hospital ward [16], an office [15] and at school [17,18]. As part of a UNICEF program, they were also used to characterize social exchanges in small villages in Kenya and Malawi [19,20] and for ethological studies on baboons [21]. It has been known for a long time that the overall distribution of the duration of contacts in face to face interactions is "broad" [8] and presents some "similarities" when observed in different environments . However, those comparisons were performed on data taken in some similar sociological environments, which are typically occidental, educated and often with a scientific background . Here we wish to extend the study of face-to-face interactions by comparing them to some very different datasets that were originally designed for other aims. The fist one are the data taken in the rural Malawi village. The second one concerns interactions among baboons in a primatology center. Moreover, there is more information in the data than what was previously presented [10,11]. Indeed, one has access to the full timeline of interactions for each pair of individuals separately . This allows to study the mean-interaction time per relation and, most importantly, deviations of the contact duration from it, which reveals the underlying relation dynamics. We will show that they are surprisingly similar in all the settings. After describing our data selection and methodological differences with some previous studies in Sect. 2, we will focus on the details of the temporal interactions in Sect. 3.2 after showing rapidly that social interactions among the participants are obviously very different in each environment. We will introduce the concept of contrast of the contact duration and show that the distributions are extremely similar on each dataset and for each relation individually. In the Discussion part, we comment on the utility of using the robust contrast distribution in improving agent-based models, and conclude summarizing the results and highlighting some possible future extensions. Some extra information, referred to in the text, is given in the Suplementary Information document in Additional file 1. --- Material and methods --- Datasets We have chosen four datasets from the sociopatterns web site, sociologically most dissimilar. 1. hosp: these are early data collected over 3 days1 on 75 participants in the geriatric unit of a hospital in Lyon [16]. Most interactions involve nurses and patients. 2. conf : these are also some early classical data from the ACM Hypertext 2009 conference that involved about a hundred of participants for 3 days [13] in Torino . The audience is international with a scientific background. There exist also some data taken at another conference in Nice in 2009 with more participants, but we prefer to use the former which has a number of individuals comparable to the other datasets. However we have checked that we obtain similar results with the SFHH data. 3. malawi: these proximity data were taken in a small village of the district of Dowa in Malawi where 86 participants agreed to participate for 13 days. Interestingly those data contain both extra and intra-household interactions, although we will not distinguish them here. This community consists essentially of farmers. 4. baboons: those data were taken at a CNRS Primate Center near Marseille where 13 baboons were equipped with the sensors for a duration of 26 days. The goal was to study their interactions, and study how conclusions reached from data-analysis match those provided by human observation. With that choice, we span very different sociological environments. We have also analyzed a few other datasets collected at the SFHH conference, an office and a high-school. They give similar results but we consider them as sociologically closer to the conf one. We have chosen to focus on the sociopatterns data since they provide a consistent set taken with the very same devices, minimizing possible sources of systematic errors. --- Differences with previous studies Previous studies considered the overall temporal properties of interactions, i.e. without differentiating the pair of people interacting. In this work we will put accent on the temporal properties of each pair separately. Probability distribution functions are often estimated by histograms, i.e. by counting the number of samples that fall within some bin. But for heavy-tailed distributions the size of the bins is delicate to choose. With a constant size binning, several bins end up empty for large values. Using a logarithmically increasing binning is neither a solution since it supposes that the distribution is constant on the wide range of last bins. Following [24], we will use instead the probability to exceed function which is computed simply by sorting the samples and plotting them with respect to their relative frequency. In this way, one does not need to define a binning and the distribution is easier to apprehend. --- Results --- Interactions between individuals Since it is not our primary goal to study the social structures in those very different communities, we just highlight visually some differences on Fig. 1 which shows 24 hr timeaggregated graphs of the relations between individuals. The graphs for the hosp and especially the conf datasets show a strongly connected core. The malawi one is much sparser, while the baboons one is almost complete showing that each animal interact with all the others. Table 1 gives a more quantitative view of some of the graph's properties. The number of different people met per day is about 20 in both the hospital and the conference environments. As is apparent in Fig. 1, it is much smaller in the rural community . But the interaction times are longer which reflect different sector of activities . The strength of the relation represents the total time per individual spent interacting with others per day. It is essentially the product of the mean number of people met per Figure 1 Aggregated graphs of interactions over one day for our 4 datasets. Vertices represent agents and there is a link if there was at least one face-to-face interaction for more than 20 s. The first day from the datasets is used, but very similar results are obtained with the others Table 1 Properties of time aggregated graphs on each dataset per day. Uncertainties are the standard deviations between the days. T is the number of days in the dataset. N the number of interacting agents. k is the mean degree, i.e. the average number of agents each individual interacts with during one day. w is the mean weight where the weights specify the total duration of a single relation [25] The comparison to the baboons dataset should be handled with care since there is a much smaller number of agents . Since each baboon interacts essentially with each other ), the mean degree is bounded to k N . On the other hand, their small number possibly increases their interaction duration so that the strength of their relation is finally similar to that of the human groups. The goal of this short section is not to dwell into the topological details of these time evolving graphs, but to illustrate that, as expected, these heterogeneous sociological groups show some very distinct interaction patterns between individuals. --- Face to face temporal relations We are interested in the duration of the contacts in those different networks. Figure 2 shows a classical distribution, that of the duration of contacts. We emphasize that such a representation mixes all the interactions of all the participants in the same plot. As well known, these distributions are "heavy-tailed"; most interactions are of short duration but some may drift up to an hour. Interactions for people in malawi tend to last longer than for all the others. The baboons' duration of interaction is similar to the human ones , although there are some sizable differences at short times, somewhat squeezed by the logarithmic scale. Overall, although there is a common trend, some differences appear too. The new aspect of this work concerns the detail of each relation separately. For a given data-taking period, each relation consists in a set of intervals measuring the beginning and end times of the interaction at the resolution of the instruments . There is a varying number of interactions per relation, that we call N int . In the following we will consider the duration of the interactions that we note {t i } i=1,...,N int . They are thus variable-size timelines expressed in units of the resolution step. The number of registered interactions for a given pair depends on the total duration of the experiments but we may compare them just for one day. The distribution of this variable is shown in Fig. 3. It is clearly different for each group. People at the conference tend to interact less often. In 65% of the cases it is only once per day, against 25% for the hosp and malawi datasets, and 3% for baboons. The mean interaction time per relation t = 1 N int N int i=1 t i , is shown in Fig. 3. Here again distributions are heavy-tailed and different. There is a marked difference between animals and humans, the former interacting for shorter times. We are now interested in studying the deviations of the contact duration from the mean value for a given relation. Indeed, in physics the dynamics of a process is often revealed by such a quantity. For instance in cosmology, one uses the "density contrast" that represents the galactic density divided by its mean value. It is the fundamental quantity which traces the dynamics of the underlying field . Inspired by this example, we propose to study what we call the "'duration contrast", or simply "contrast" which is the simplest dimensionless quantity we can form to study deviations from the mean-value δ i = t i t , where r recalls that the quantity varies for each relation. The contrast represents our tendency to spend more or less time than usual with a given individual. Note that "usual" is meant as the mean-interaction time between the two peculiar agents and varies for each relation. For a small number of samples, the arithmetic mean ) is however a poor estimate of the true mean-time and also strongly correlated to the individual samples. Taking the ratio leads to a very noisy estimate of the true contrast variable. In the following we will then apply a cut to keep timelines with a sufficient number of samples. Since the distributions are very broad we require at least N int > 50 contacts in a relation. We will study later the effect of this cut on the results. On the complete datasets, we are left with respectively 57, 26, 91 and 70 timelines for the hosp, conf, malawi and baboons datasets. We show the p.t.e distributions of the contact duration contrast for the 4 groups in Fig. 4. The tails look now very similar up to 10 times the mean-time. The same distribution is observed on data from another conference, an office and a high-school . Thus, a similar distribution is observed on 7 independent datasets. To be more quantitative and assess the level of compatibility between the distributions, we use a Monte-Carlo method. For each dataset, we numerically invert the empirical distribution functions to construct the inverse cumulative function F -1 . We then draw N numbers u from a [0, 1] uniform distribution, transform them with F -1 and reconstruct the p.t.e. The procedure is repeated 100 times and all distributions are plotted on top of each other on Fig. 5. One sees that the distributions are indeed all compatible in the 0.6 δ 10 range, where the upper bound comes from the limited sample size of the hosp and conf datasets, and the lower one from slight differences for low values. This will be our range of interest in the following. Since the data-taking periods are very heterogeneous we have split the data day by day and verified that no particular one particularly affects the results . We have also removed randomly a fraction of the agents , i.e. we removed all relations involving those agents, which did not affect the contrast distributions in a sizable way . The interaction mean-time corresponds to a value of 1. We then see for instance that the probability for an interaction to last longer than its mean-time is around 30%, but, rarely, it can exceed 10 times the mean-time Another option for studying deviations from the mean is to use the z-score z i = t i -t σ , where σ represents the standard-deviation of the duration values. The results obtained with this variable are very similar to the ones with the contrast and we did not notice any difference on the tests that are presented later. Since the contrast variable is somewhat simpler we only focus in the following on it. We consider the impact of applying the N int > 50 cut. First, we note that similar results are obtained with a lower cut value as N int > 30 . We then show that we can still reproduce the contrast distribution without any cut, using only the distributions with the cut . To this purpose we perform Monte-Carlo simulations. For a given dataset, for each relation , we draw N int random numbers following Fig. 4 distribution to obtain δ i=1,...,N int contrast values. Those samples are obtained from the distribution with the N int > 50 cut, so with precise mean values that we call μ. We may mimic the statistical fluctuations due to any N int value, by using the ratio δ mes i = δ i 1 N int i δ i = t i /μ 1 N int i t i /μ = t i t since μ actually cancels out. We compare the measured contrast distribution to the one observed on data, this time without any N int cut, in Fig. 6 for the conf dataset. We reproduce correctly the whole contrast distribution using only the Fig. 4 one obtained with 1% of the data . Similar results are obtained on the other datasets . This shows that the contrast distribution obtained from the large sample statistics is sufficient to reproduce any number of interactions, including small-sample ones. In other words, the N int > 50 cut only cleans the data without affecting the underlying "true" contrast distribution. To check that the contrast distribution is not artificially produced by the procedure of dividing the timelines by their mean value, we use the hosp dataset to retrieve the set of interacting agents and their corresponding characteristics N int and t. We then draw N int random numbers following a Poisson distribution of parameter t and recompute the contrast. The result is shown in Fig. 7 which is clearly different from the results observed on the data. The shape of the observed contrast distribution is nontrivial. It is neither of exponential nor of power-law form. A stretched-exponential form is neither satisfactory. Empirically, we could obtain a reasonable fit in the 0.6 δ 10 region, by combining both a power-law and an exponential function p = 0.3e -0.2δ /δ 1.1 . ( )5 The denominator is here to enhance short contrasts, while the exponential term describes the long ones. This could be an indication of the existence of two regimes, one for short times when communications are more informative and a longer one when real conversations form [27]. At this point, we have shown that the combined contrast duration follows a very similar distribution. We now consider each relation separately and show in Fig. 8 a superposition of the contrast duration distributions with the N int > 50 cut (similar results are observed without it but are, as expected, more noisy (see SI Appendix, S6. --- 2). They all follow rather closely the common contrast distribution. In other words, while the choice of individuals we meet , the interaction rate ) and mean-time spent together ) varies strongly with the environment, the propensity to spend more time than usual with a given individual, is remarkably similar. This points to the idea that once a face-to-face contact is triggered it follows its own dynamics, out of the sociological context. For the sake of completeness, we note that we found no sizable correlations between the contact duration within the timelines . This indicates one can draw independent samples using Eq. . We also considered the inter-contact time in the relations to see whether its contrast reveals features similar to the duration ones. This is not the case as shown in --- Comparison with a model The contrast distribution can be used as a new metric when studying face-to-face temporal graphs in order to test and improve existing agent-based models designed to reproduce the full evolution of a set of individuals. For instance, the "force directed motion" model is successful in describing several key features of observed face-to-face interactions [6]. Based on the idea of attractiveness between some agents performing a random-walk within a bounded perimeter [4,28], the model further includes the concept of "similarity" between two individuals [29], known as homophily in social sciences. The similarity s ij influences the time two agents spend together and the way the random-walk is biased. The model assumes that the contact duration between two agents is exponentially distributed with a rate s ij /μ 1 , where μ 1 is adjusted on the data to reproduce the overall duration of Figure 10 Comparison of the contrast distributions obtained with the hosp dataset to the result of the "force-directed motion" model [6]. We used the parameters provided by the authors and their dataset . The FDM curve is the combined result from 10 simulations contacts. We have run the code provided by the authors with their setup corresponding to the hosp dataset, to test the distribution of the contrast variable. Figure 10 shows that the model distribution falls too steeply. We have tried adapting the parameters and some parts of the code but could not find a configuration giving a better contrast distribution . 2Modeling correctly the tails of the contact duration is also essential in epidemiological studies since the spread of a disease happens mostly during long interactions. For a given mean-interaction time, Eq. allows to simulate a much more realistic duration of contacts than a Poissonnian one. This can be used in SIR-like statistical inference, or using agent-based models, for the precise modeling of long interactions. --- Conclusion We have compared face-to-face interaction data taken in some very different environments; some were recorded in a European hospital and during a scientific conference, others in a small village in Africa. With the original intention to pinpoint differences with the results concerning humans, we have also included data on baboons' interactions in an enclosure. Although the topological structures and the mean-time spent together are clearly dependent on the sociological environment, it appears that the deviations from the mean-time for each pair follow a very similar distribution, including for baboons. We tend to interact most often for much less time than "usual" with a given individual and sometimes, but rarely, much longer. What is striking is that the distribution for this quantity, which we call the "relation contrast" looks universal. It is the same for people at a scientific conference or farmers in a small Malawi village , see Fig. 4 . These results suggests that, once a face-to-face contact is triggered, it follows its own dynamics independently from the social context. This is maybe not a big surprise to a sociologist in particular working in the field of Conversation Analysis [27] where it is postulated that each conversation follows some rules independently from the social context . 3 But to our knowledge, this was not noticed by physicists and may help disentangling the topological and temporal aspects of face-to-face interactions. The possible universality of the relation contrast must be challenged with more data. On the animal side, one should consider groups of animals with strong social interactions, that can be identified and followed individually. Hominids, as baboons, are known to have social behaviors close to ours, which probably explains the similarity of the contrast distribution with the human's one. Chimpanzee or bonobo's data should show similar characteristic. Concerning mammals, we could think of tracking individuals in elephant herds or wolf packs but it's difficult to acquire precise data in the wild. The most promising approach concerns the study of social insect networks [31]. Details about ant interactions is probably the most feasible since recent techniques allow to tag and follow each individual separately [32]. On the human side, we need to check whether the contrast is influenced by age. Since children perceive time differently from adults, following the contact patterns of young children in a nursery could provide a valuable insight into this question. --- Supplementary information Supplementary information accompanies this paper at https://doi.org/10.1140/epjds/s13688-023-00444-z. --- Data availability • The datasets analyzed during the current study are available in the sociopatterns repository, www.sociopatterns.org • The FDM code was downloaded on 10 June 2023 from https://bitbucket.org/mrodrflr/similarity_forces • The python3 software used to produce the results is available from https://gitlab.in2p3.fr/plaszczy/coll • The graph-related computations and Fig. 1 were obtained with the graph-tool software [33]. --- --- Competing interests The authors declare that they have no competing interests. --- ---

The analysis of social networks, in particular those describing face-to-face interactions between individuals, is complex due to the intertwining of the topological and temporal aspects. We revisit here both, using public data recorded by the sociopatterns wearable sensors in some very different sociological environments, putting particular emphasis on the contact duration timelines. As well known, the distribution of the contact duration for all the interactions within a group is broad, with tails that resemble each other, but not precisely, in different contexts. By separating each interacting pair, we find that the fluctuations of the contact duration around the mean-interaction time follow however a very similar pattern. This common robust behavior is observed on 7 different datasets. It suggests that, although the set of persons we interact with and the mean-time spent together, depend strongly on the environment, our tendency to allocate more or less time than usual with a given individual is invariant, i.e. governed by some rules that lie outside the social context. Additional data reveal the same fluctuations in a baboon population. This new metric, which we call the relation "contrast", can be used to build and test agent-based models, or as an input for describing long duration contacts in epidemiological studies.

INTRODUCTION Syria's rich socioreligious diversity and inherited norms have shaped the experiences and expectations of Syrian women. 1 2 Women's historic participation in the public sphere included Nazik al-Abid establishing the Syrian Red Crescent Society in 1920, 3 resistance to the French Mandat pour la Syrie et le Liban occupation 4 and Syria being first regionally to recognise women's political --- WHAT IS ALREADY KNOWN ON THIS TOPIC ⇒ There has been minimal examination of women's experiences in healthcare spaces in oppositioncontrolled areas in Syria and virtually no amplification of their voices in research. ⇒ The Syrian conflict is complex and has both helped and hindered gender equity and women's space for action. --- WHAT THIS STUDY ADDS ⇒ This is a first attempt to document some of the complexities of women's lived experiences as healthcare providers and service users during protracted conflict in Syria. ⇒ Women's lives are affected by physical insecurity, overwhelming workloads, and normative gender roles and social expectations that have been exaggerated by ongoing multiparty conflict, from local travel restrictions and risks to early/arranged marriages and potentially severe stigma or punishment for being harassed/raped. ⇒ Increased employment opportunities do not appear to have enabled increased social empowerment or leadership opportunities. BMJ Global Health rights in 1949. 5 Similarly, during the 2011 Syrian uprising, women participated effectively in non-violent activities challenging the discriminatory and patriarchal norms. 5 6 As peaceful uprising morphed into complex, multisided revolution/civil conflict/proxy war, discriminatory violence against women has been amplified by all conflict actors. 5 7 A decade of armed conflict has decimated Syria and its health system, fragmenting both among several areas of military control . We use the term 'opposition-controlled areas' to refer to areas controlled militarily by several armed opposition groups 1 8 9 or Turkish forces. The ongoing multiparty conflict has reduced OCAs boundaries in the last 5 years, for example, Eastern Ghouta is no longer opposition controlled. 10 Autonomous Administration forces control parts of Syria's northeast, and the remaining approximately 60% of territory is controlled by the Syrian government supported by Russian and Iranian forces. 1 8 11 12 Healthcare became politicised and weaponised through systematic human rights violations, including targeting of health workers and infrastructure mainly by the government and its allies. 9 13-15 Protracted conflict reinforced gender inequalities, posing disproportionate risks for Syrian women and girls. 5 16 Many have been widowed, divorced or separated as men were killed, detained or left the country, and women and children constitute the majority of internally displaced Syrians. 6 16 Women experience gendered risks during conflict, including gender-based violence , persecution and imprisonment, 17 18 in addition to lower pay, fewer leadership opportunities and responsibility for most unpaid household/caring work. 19 However, despite increasing literature on health and the Syrian conflict, health worker and service user voices-if included at --- BMJ Global Health all-are primarily male. 9 14 20 The limited literature on Syrian women's healthcare experience in OCAs generally consists of reproductive health or COVID-19 issues, without explicitly considering gender or related power dynamics. [21][22][23][24] This study aimed, as initial corrective, to explore women's lived experiences and perspectives of healthcare in opposition-controlled Syria during ongoing conflict. Objectives were to: describe women's experiences as health workers and health service users in this insecure setting; and consider opportunities for greater health system equity for women as health workers and service users. --- METHODS --- Study design and methodological approach We chose a qualitative study design, drawing from in-depth interviews with women health workers and health service users in OCAs, focusing on commonalities of women's lived experiences. We adopted an interpretative phenomenological approach, informed by Smith et al 25 , to examine how the experience of conflict and involvement in the Syrian uprising affected the space for women in healthcare, both as providers and recipients. --- Research question The research question was intentionally broad: 'How have women experienced their roles as healthcare providers and service-users within the OCAs health system during the conflict?'. --- Sampling and recruitment Eligible participants were women who had: provided healthcare services for at least a month in OCAs since the start of the Syrian uprising in 2011; or received healthcare services at least once in these areas and time period. Given inherent difficulties and inappropriateness of identifying biological sex for this study, we chose the term 'gender' throughout for expressed identity. 26 We purposively selected interviewees to provide a range of roles and experiences within opposition-controlled health facilities. MA and YD drew on their professional contacts in OCAs to recruit eligible women. Interviewees were further snowballed for a broader range of providers and service users. --- Data collection MA obtained informed consent and conducted in-depth interviews in Arabic, between July and August 2019, using separate topic guides for health workers and service users. We developed topic guides, information sheet and consent form in English and translated them into Arabic. Topics included perceived health system challenges, solutions and adaptations, participation, and transparency, allowing scope to explore emerging concepts. MA did not have prior relationships with interviewees, and no incentives were provided. After introduction by a mutual acquaintance, MA conducted an informational call with each potential interviewee to allow time to review study information sheet and consent form, discuss study objectives and researcher background, ask any questions and decide on participation. Only two eligible interviewees chose not to participate and did not provide reasons. As only four interviewees were able to send electronic copies of signed consent forms, MA recorded verbal informed consent for all interviewees prior to interview. Confidentiality and anonymity were ensured by not recording names, removing any identifying data during transcription and using identity codes in all transcripts and outputs. Privacy was ensured by conducting interviews remotely using internet apps at times chosen by interviewees. 27 MA conducted interviews in colloquial Syrian Arabic, taking detailed notes, and recording all but one-who refused-using an encrypted digital audio recorder. Interviews lasted approximately 60 min. MA had excellent access, so due to time and resource limitations, we determined data saturation when no new topics or conceptualisations arose in interviews. 28 MA reassured interviewees that they could stop at any time and skip any topics. We stored data in encrypted passwordprotected institutional servers only accessible to our team. --- Analysis MA transcribed interviews in colloquial Arabic and analysed them manually with AH according to the six phases described by Smith et al: reading and rereading; initial noting; developing themes; searching for connections across themes; moving to the next case; and looking for patterns across cases. 25 MA developed themes, connecting them with the help of AH and NH, using a combination of abstraction, subsumption, polarisation and contextualisation according to question guide topics and interview notes. 25 Analysis focused on commonality of lived experiences of conflict among women, with relevant quotes translated to English by bilingual coauthors. Themes were critically reviewed by NH, and discrepancies were agreed between investigators. Reporting adheres to Consolidated criteria for Reporting Qualitative research criteria. 29 --- Reflexivity This interpretative study emphasised empathy, openness, questioning preunderstanding and adopting a reflective attitude. 25 The process of examining gender is itself imbued with power relations, including who collects and analyses data, who participates, when and where data are collected, who is present and where data are published. 30 --- BMJ Global Health during the conflict. This increased access, trust and rapport-building with interviewees, who were able to share their experiences in their own dialect with a women researcher who shared their religiocultural background yet was not so close that interviewees were worried about confidentiality within their communities. This balance elicited rich descriptions of lived experiences and sensitive issues. Given MA's similarities with interviewees, she used reflective writing and team discussion throughout to examine theories and assumptions and maintain a critical stance. 32 This is distinct from bracketing, as assumptions are part of sociocultural understanding and should not be automatically discarded. 33 Please check Supplementary Appendix SA1. --- Patient and public involvement We developed and implemented research guided by an advisory committee of Syrian academics, healthcare practitioners and laypeople. All members had experience of the OCAs health system as providers or patients and helped reflect public priorities and preferences. We included service user interviewees, conducted Arabic dissemination webinars for OCAs, ensured open access publication and will host translated materials at https://scahr.org/. --- Ethics Due to the difficult circumstances of women interviewed, we emphasised the voluntary nature of participation, provided additional opportunities to interrogate the study and researchers, and remote referrals to an Arabicspeaking, London-based psychotherapist, experienced in trauma counselling and paid from research funds. Of the three referred, two subsequently cancelled and one disengaged after two sessions to flee increased bombardment. 27 --- FINDINGS Interviewee characteristics and themes Table 1 provides interviewee details. To preserve anonymity, we only reported aggregated characteristics. Seventeen were based in OCAs , while two providers and one service user had recently migrated from OCAs to Turkey and Germany, respectively. Interviewee ages averaged 34 years , 16 were married and 14 had children. Ten were health workers employed in more than one facility. Findings are organised under three overarching themes: gendered effects of the conflict, gendered healthcare employment concerns; and challenges within foreign humanitarian responses. We found, as most interviewees were health workers and all were potential service users, that reporting perspectives separately by group was unhelpful. Instead, we highlighted any differences noted. --- BMJ Global Health Gendered effects of conflict Two inductive subthemes were: changing social norms; and GBV. --- Changing social norms Providers explicitly described changes in social attitudes towards working women and the complexities this added to women's social roles. Patriarchal norms remained, such as underestimating women doctors' qualifications, underestimating nursing as a profession or considering it inappropriate for women. However, the conflict imposed additional needs that favoured working women. Increasing demand for women health workers was due primarily to major health worker losses because of deaths and emigration and partly to increasing radicalisation among armed groups, resulting in some husbands not wanting, or not allowing, their wives to be examined by a male doctor. Women would die of some disease because their husbands wouldn't allow them to be checked by a male doctor… SU4 Due to the low numbers of women gynaecologists and high demand for women to be examined by women, healthcare managers intentionally recruited female doctors whenever possible. This new attitude increased women's employment chances. They [managers] prefer female professionals in health facilities. HP4 Some providers also described how the conflict had forced them to develop strong leadership skills so they could support their families. Two dissenting interviews described patriarchal Syrian traditions positively, in that men usually prioritised women and helped them as a way of demonstrating protection. We as women don't suffer from oppression. I have my own car. There is no harassment, just respect. I don't feel anything bothering me [as a working woman]. Currently, it's not acceptable for men to work as gynaecologists. Women work in jobs that are suitable for them and vice versa. There is balance. HP7 --- Gender-based violence Views on what constituted sexual harassment or assault and ways these had changed or increased were mixed. An interviewee mentioned a young girl who was raped and later found drowned, without clarifying whether this was suicide or murder. Another discussed a girl raped by boys in a displacement camp, suggesting overcrowding was to blame rather than perpetrators. Women described varied management reactions to sexual assault in hospitals. In one case, the woman married her perpetrator, though it was unclear whether this was her choice. In another, management initiated a formal investigation and the perpetrating doctor was subsequently prohibited from working in any OCAs health facility. One described the challenges of reporting harassment, as it would normally be considered the woman's fault, while another described how the conflict enabled new opportunities to hold perpetrators accountable, while demonstrating victim blaming by suggesting women should not 'make mistakes' and conflating expressing opinions with achieving accountability: I think a woman is able to report harassment if she was the victim, I mean if she didn't make a mistake. Previously, they used to blame women for sexual harassment. The whole perception has changed. Now, it is better. You can express your opinions… HP15 Interviewees described stigma and fear associated with reporting sexual violence or seeking related healthcare. You know the environment we live in and the type of people here. In war zones, these things increase. It is difficult to activate a reporting system for such things. HP14 Women normally felt pressured to conceal GBV experiences when seeking health services. For example, if an unmarried woman were raped and needed to check if she got pregnant, she might tell the doctor during examination that her husband had died recently for fear of criminalisation or death for out-of-marriage sex. A form of GBV most women noted had increased during conflict was early marriage. For families, protecting daughters from hunger, besiegement, kidnapping and rape were major responsibilities they were increasingly unable to meet during conflict and many chose early marriage for them instead. Some women community health workers risked themselves by advocating against forcing girls to marry and thus criticising extremist interpretations of shari'a . Another suggested that NGO campaigns telling girls to avoid early marriage were not helpful given they had no choice about marriage and should instead target decision makers with religious and health education. A 17-year-old widow with a child… what does she know about her health or about her child's health? SU2 Several providers discussed the negative health consequences of child marriage, such as more girls requiring caesarean sections. Can you imagine a 15-years-old girl who has had two caesarean sections already? HP7 Gendered healthcare employment concerns Three emergent sub-themes were: living with daily insecurity and restrictions; overwhelming 'double' workloads; and insufficient women health workers. --- Living with daily insecurity and restrictions All providers identified physical insecurity as the most challenging aspect of providing healthcare in OCAs, describing multiple stressors. Conflict-related psychological stressors affected women's mental health. Providers described horror, anxiety and fear as pervasive emotions affecting their work, particularly when hearing about another health facility bombing. People don't understand that we are tolerating more than we can bear. They think we are made of iron… HP4 Several women highlighted the perceived insecurity of the new 'mixed-society' in Northern Syria, as foreign combatants and their families moved into Syrian communities. Interviewees used the term 'mixed-society' to describe influxes of foreign combatants, affiliated with foreign powers and using very conservative interpretations of Islam to justify their involvement in Syria. I can't leave my daughters home alone. They are always with me. There is no safety here. People [foreign combatants] came from different ethnic backgrounds… Each is free to do whatever he wants. HP15 Interviewees described foreign combatants imposing extremist ideology, including restricting women's participation in public spaces. One provider mentioned struggling with not knowing women patients' names, as extremist restrictions stipulated women should not be publicly recognisable. Regular aerial bombardment by government and allied forces, combined with 'mixed-society' extremism, made roads insecure for all women, whether healthcare providers or service users. Combatants recruited boys for combat and extremist indoctrination. When the regime brought those people [foreign combatants], it made the situation much worse […]. They recruited 15-year-old boys at their checkpoints to tell us a woman driving a car is 'haram' [i.e. forbidden in Islam]. HP4 Women health workers experienced restrictions on their appearance that men did not. For example, facility managers required women health workers to wear loose clothing and avoid makeup to reduce confrontations with increasingly radicalised combatants. You feel trapped. You don't have the freedom to behave normally. HP15 Indirect consequences of conflict were often gendered. An interviewee in besieged Eastern Ghouta reported that when hospital management stopped providing staff transportation during the prolonged blockade, to allocate scarce fuel for crucial electricity generators, women were most affected. She noted that men could ride bicycles or motorcycles, a socially unacceptable option for women, while she had to walk four miles each day during aerial bombardment experiencing greater risk of injury or death. Every day, we go to work filled with fear. Would we reach there or not? HP4 Health worker shortages and financial hardships meant many women worked in multiple facilities, increasing their risks of kidnapping, rape or murder. All were common while travelling, especially of people considered sufficiently important or wealthy . We were living in a place that looked like a state, but it wasn't a state. HP3 Women's coping strategies included normalising death because living while anticipating death seemed senseless, trusting God's will and interpreting any death while providing health services as honourable. The regime forces were targeting the hospital with barrel bombs while we were operating on some patients. These bombs were so powerful, the dust went into the open wounds. We didn't stop. We cleaned and continued operating. It happened many times […]. When it comes to death, we didn't mind dying while working […]. Praise God that we are alive to deliver our voices… HP11 Overwhelming 'double' workload Women described 'double' workloads of providing financially for their families alongside unpaid household responsibilities. We must deal with many tasks. It's like holding two watermelons in one hand… SU5 Providers all described increasingly overwhelming daily workloads, due to continuous bombardment, complex injuries and mass influxes of displaced people. Today I did 11 caesarean sections. The other day I did 14. It's exhausting. HP14 Many described working 24-hour days, with days off a luxury due to staffing policies and shortages. This took a heavy toll on their well-being and productivity, with the only choices tolerating it or quitting their job. Besieged areas were additionally challenging. Providers reported not having specific work hours and expectations of always being on standby. Some health facility managers lived at their facility. A senior manager described the additional workload required of her promotion. In besieged areas, numbers of health workers were particularly limited. Women were traditionally underrepresented outside nursing and midwifery, which only worsened during conflict. Before the revolution, the proportion of female to male doctors was unequal. Women used to consider studying pharmacy or dentistry rather than medicine. HP3 Female surgeons were particularly rare. Of course, there are more male health providers, for example, 75% more. We really lack female doctors. Most female doctors left the country. We have no female surgeon in [this town]. HP4 Many women with non-medical backgrounds received nursing and midwifery training, while others were requested to perform tasks outside their role due to severe staff shortages. While necessary, accelerated training and task-shifting were sometimes brutal. --- Challenges within foreign humanitarian responses Women highlighted how multiple competing responsibilities created implicitly gendered employment policy implications. Two inductive subthemes were: adverse employment policies; and mitigation measures women enacted or advocated. --- Adverse employment policies Many health facilities at which interviewees worked were remotely financed or operated by international non-governmental organisations . Working for international organisations was new for most women. INGOs provided capacity-building trainings, particularly for nursing, midwifery and community health workers. Providers described these positively, as crucial for women from non-medical backgrounds. However, they noted that INGOs typically lacked health worker incentive structures, instead relying on penalties such as warnings, salary reductions or dismissal. Only one mentioned incentives, that is, monthly well-being days at her facility. Most women agreed that INGO policies prioritised service user needs , while health facilities preconflict had prioritised providers. For example, INGOs usually allocated representatives at each facility to help supervise workflow and address service user complaints. Insecure employment was the main challenge described by all and particularly affected women breadwinners who had less geographical and temporal flexibility than men due to household commitments. INGOs could halt funding at any time without notifying facility managers or staff. Women described working in INGO health facilities as challenging and not always fair. For example, the requirement for 24-hour shifts caused high turnover for women health workers with caring responsibilities at home. Is any human being able to work 24 hours per day? We are not robots… HP12 Reasons for such INGO policies were unclear, though most women suggested it was to reduce expenses by having fewer employees work longer hours. They always make sure to reduce the expenses as much as they can. [Hiring more staff] is costly. HP12 Women primarily attributed facility staff constraints to low budgets rather than lack of qualified candidates. Attitudes to INGO employment policies were mixed, with several providers unsure about INGO regulations or whether they had employee rights. Many described 'unfair policies', such as not hiring additional needed staff. INGOs who claim to be humanitarian, they impose work conditions incompatible with their claims. A salary of one more employee wouldn't affect the INGO. Salaries of ten more employees wouldn't affect the INGO. HP7 Employee leave policies were inconsistent across facilities and employers and particularly difficult for women. One mentioned an INGO having a clear employee leave policy, which management ignored. Another described a nurse in her facility being fired because she got pregnant and there was no maternity leave policy. A women was given 15 days leave after the death of a relative, while another received none because nobody could replace her. --- BMJ Global Health Some women reported working without contracts, as income needs required risk-taking, though it was unclear how widespread this was. However, even health workers with fixed-duration INGO contracts could be dismissed at any time. Despite these issues, major population needs inspired many to continue working even voluntarily. Even though we don't get paid when no donor sponsors the facility, it is very hard for us to leave the population without healthcare. HP7 Mitigation measures women enacted or advocated Women were resourceful and described several working condition improvements they were able to enact. For example, some successfully lobbied facility managers to provide a practical solution for 24-hour shifts by dividing the workday into three shifts with two health workers sharing one contract. Women gynaecologists reported that midwives provided tremendous contributions to healthcare in OCAs and a midwife suggested developing a midwives' association to organise and support their work. Doctors are protected. We midwives are the vulnerable group. Any medical mistake, they would blame us and ignore all our great efforts… HP7 Another suggested life insurance schemes for health workers in conflict-affected areas, supported by INGO funding. As nobody had sufficient authority to stop health facility attacks, having access to affordable life insurance could help ensure their families would be provided for if health workers were permanently disabled or killed. --- DISCUSSION To our knowledge, this study is the first to amplify Syrian women's descriptions of their lived healthcare experiences in opposition-controlled areas of Syria. This is important both given the dearth of such research in Syria and because it was Syrian led, unlike much global health research that excludes or minimises 'local' coinvestigator contributions. 34 35 We argue that research investigating health sector experiences of women in Syria was needed, as research in conflict-affected settings often ignores the socially constructed power relations and gender norms that lead to different health system experiences, needs and outcomes. 36 Our findings showed the significant role women play in the OCAs health system and the physical, mental and professional overwhelm and precarity they experienced. This supports findings from other settings, for example, Afghanistan and Somalia, which described health worker shortages, preferences for women health workers, gender inequalities, multiple employments and donor dependence amid ongoing insecurity. [37][38][39] Many challenges women health workers encountered were similar for men, but impacts often affected women disproportionately. [40][41][42][43][44] Women described complexities and challenges in how conflict influenced professional and social gender norms, including increased but more precarious work opportunities, harassment and violence. Differential treatment by managers, often based on INGO policies, increased work stress and precarity. However, many continued to volunteer after INGOs stopped funding their facilities, choosing to help where they could. Identifying conflict-imposed changes in gender roles requires accounting for pre-existing inequities, for example, Syrian women always had greater household caring responsibilities and more socially restricted access to education or employment opportunities. 45 46 OCAs remain insecure, having undergone frequent governance shifts and fragmentation during years of conflict. 1 9 11 43 47 48 Women's accounts suggested potential opportunities in healthcare spaces, particularly in increased demand and broadened roles for women health workers. The increased demand for women health workers versus their limited space for action and leadership in health facilities reflected the tensions inherent in the new 'mixed-society' they described. Findings are comparable with those in Afghanistan, with increasingly repressive patriarchal structures increasing demand for qualified female staff to provide for women. [49][50][51] Further reforms may thus be possible in OCAs after conflict, as women described sociopolitical norms as malleable, to improve gender equity within the health system. 52 However, if changes remain linked to radicalised 'mixed-society,' excess workloads, or insecure contracts, improved gender equity seems unlikely. Importantly, more job opportunities do not automatically increase women's empowerment, as conflict is transforming Syrian norms in complex and fragmented ways. Further intersectional inquiry is needed to avoid assumptions about 'Syrian traditions' or gender that fail to capture diverse experiences. 53 Examining health impacts of gender inequity is often undermined by a lack of consensus regarding data interpretation. 54 However, discussion and engagement with those affected is crucial. 52 Syrian women, as health workers and service users, offered context-driven priorities with implications for policy, practice and further research. There is an urgent need for a legitimate authority to organise health responses across OCAs and coordinate international and Syrian actors. In the meantime, a functional and equitable humanitarian response is essential. Women often play an essential role during conflict and 41 45 but interviewees were essentially penalised for taking on additional responsibilities . International donors and NGOs should further coordinate in promoting equitable working conditions for Syrian staff, so women health workers have contracts aligned with international human rights and employment standards. Improving gender equity in healthcare during conflict is not limited to celebrating women's achievements in adversity but must support employees' additional responsibilities and enable men and women to share in strategic decision making. Our findings on GBV support the literature indicating disproportionate impacts on women and girls during conflict, due to erratic rule of law and breakdown in social norms. 5 55 Under-reporting and the lack of a unified safeguarding system for sexual harassment are particularly problematic given the fragmented health system space for action at health directorate and facility level. Women deserve safe spaces to report and address sexual assaults, which is a responsibility of both international and local authority partners. Child marriage is a form of GBV, which negatively affects Syrian girls' health and well-being, and existed to some extent preconflict. 56 For example, UNICEF found that in 2002-2011, 13% of Syrian women aged 20-24 years were first married before age 18 years. 57 58 While it remains challenging to estimate current prevalence, 57 the complex conflict, forced displacement, financial challenges and weak protection systems likely drive increases in such marriages. 59 Conflict-imposed complexities, including physical and socioeconomic hardship and insecurities, require multisectoral support by Syrian and international partners to address. 24 Given fragmented local authorities, community outreach campaigns on the harmful consequences of child marriage, initially targeting religious leaders and fathers, could help. 43 60 Engaging fathers is essential, as girls seldom control such decisions. 56 Any approach must take a sensitive and nuanced stance, incorporating inquiry and intervention beyond individual and familial levels. Solely investing in campaigns in the absence of intervention at multiple levels simply decontextualises systemic violence, portraying it as an interpersonal issue, instead of challenging structural power norms. 61 Interventions and research must incorporate intersectional narratives to strengthen gender equitable initiatives. 62 63 Further researching intersections, for example, between gender, occupation, ethnoreligious identity, socioeconomic position and stage/area of conflict, would help explain different lived experiences of women in healthcare. For example, disparities between the experiences of midwives and clinicians we interviewed were significant but could not be explored in depth. Thus, considering women homogeneously without acknowledging these differences can oversimplify overlapping categories of marginalisation and oppression. 64 As Hankivsky et al argue, existing structures provide benefits and opportunities for men, 65 from default health sector and INGO decision-making positions 52 66 to inclusion as default key informants in research on Syria. 14 If, as findings thus far suggest, the tragedy of the Syrian conflict may open policy or social spaces to shift gendered norms and power dynamics, intersectional research and engagement can contribute to these processes. 65 However, further research is needed on potentially gendered effects of INGO policies for national health workers. --- Limitations We approached participants through trusted contacts, as security concerns made recruitment and rapportbuilding challenging. MA's background as a Syrian health worker during the conflict was crucial but also contributed emic/etic tensions and complex subjectivities to interactions . 67 . For example, interviewing women with similar predispositions reduced perceived researcher impartiality, while familiarity with OCAs issues contributed feasibility and depth to our research. Data collection was challenged by bombardment, poor internet connectivity and scheduling difficulties due to women's multiple roles and overloaded schedules. Including real names and workplaces was inappropriate, and findings should be judged accordingly. Despite reassurances of not including any identifying information, some participants were particularly concerned about their families in governmentcontrolled areas being targeted. This required great care and constant checking about what could and could not be written, which may have reduced some richness but was necessary for safety. --- CONCLUSION Women participate actively in opposition-controlled healthcare spaces and contribute to governance processes in Syria. However, while the demand for women health workers is increasing, women are not routinely included in healthcare leadership or decision making thus minimising their voices in policy debates and making further sociopolitical transformation crucial. --- Author affiliations --- BMJ Global Health the manuscript with help from AH. NH revised for critical content and serves as guarantor. All authors approved the version for submission. Funding Data collection was funded by a Chevening MSc Scholarship. Manuscript development and publication were funded by NH's Medical Research Council Health Systems Research Initiative foundation grant . Funders were not involved in study design, data collection, analysis, decision to publish or manuscript preparation. --- Map disclaimer The inclusion of any map , or of any geographic or locational reference, does not imply the expression of any opinion whatsoever on the part of BMJ concerning the legal status of any country, territory, jurisdiction or area or of its authorities. Any such expression remains solely that of the relevant source and is not endorsed by BMJ. Maps are provided without any warranty of any kind, either express or implied. Competing interests None declared. Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details. Patient consent for publication Not applicable. --- Ethics approval The MSc research ethics committee at the London School of Hygiene & Tropical Medicine in the UK provided ethics approval as there was no legitimate ethics review board in Syria able to provide approval at the time of research. 27 Participants gave informed consent to participate in the study before taking part. Provenance and peer review Not commissioned; externally peer reviewed. --- Data availability statement No data are available. Not sharable due to ethics restrictions. Supplemental material This content has been supplied by the author. It has not been vetted by BMJ Publishing Group Limited and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations , and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

et al. 'We are trying to live in a normal way, but nothing is normal about us anymore…': a qualitative study of women's lived experiences of healthcare in opposition-controlled areas of Syria.

Introduction Older adults age 70 years and older are at higher risk for increased disease severity and mortality from the coronavirus disease virus . Older adults' family and formal caregivers are also at high risk of contracting COVID-19 because of their work with vulnerable older adults . All levels of governments in countries around the world have responded to the pandemic by altering practices, programs, and policies to keep community-dwelling older adults and their caregivers physically safe, we contend, at the detriment of their mental and social health . Promising practices to address social isolation before the pandemic were employed at individual, community, and societal levels; however, none considered the context of a pandemic in which older adults and caregivers are experiencing isolation . Examining the lived experiences of social isolation among marginalized older adults and their caregivers such as those living in rural, underserviced areas where community and health services are strained, is critical for identifying effective practices, programs, and policies to mitigate the negative impacts of isolation during the pandemic and afterwards . This participatory action research complements existing research as a timely approach for uncovering lived experiences of older adults and their family and formal caregivers during the pandemic and proposing contextual solutions to social isolation. Our study leveraged long-term relationships among a network of over 50 individuals, representing academia, businesses, community health and social service providers, elected officials, and older adults who volunteer with Age-friendly Peterborough . During the early weeks of the pandemic, this collaborative group in Peterborough, Ontario, wanted to examine the health experiences of community-dwelling older adults and their family and formal caregivers to tailor their service adaptation appropriately. Peterborough with a population of 83,651 in 2021 is a small city in central Ontario with 24.2 per cent of the population ages 65 years and older. Peterborough County, which includes many rural municipalities and two First Nation reserves, recorded a population of 147,681 of which 25.2 per cent of the residents are 65 years or older . Unlike other pandemic social isolation studies such as reviews and qualitative research , this PAR offers a rigorous, virtual approach, which can be adapted by community coalitions and decision makers to mitigate the negative impacts of social isolation among older adults, family caregivers, and formal caregivers. Our study is particularly significant because it includes those living alone, with lower incomes, in rural areas with service and Internet challenges, and Indigenous peoples who have been historically marginalized. --- Literature Review Social isolation and loneliness are intersecting, yet distinct phenomena. Social isolation refers to a lack of high quality and quantity of social contacts, whereas loneliness is defined as a subjective feeling of isolation, despite the size of an individual's social network . Courtin and Knapp's scoping review showed that little is known about causal pathways relating to social isolation and loneliness, with gaps being identified in only one qualitative study that pointed to the need to understand access to services and other contextual factors of health. Pre-pandemic, Freedman and Nicolle estimated that social isolation impacted 6 per cent to 43 per cent of Canadian older adults. The Canadian Longitudinal Study on Aging reported pre-pandemic prevalence of older adults age 75 years feeling lonely at least some of the timefor women, 30.83 per cent, and for men, 19.41 per cent . When looking at the intersection of loneliness and social isolation, 32 per cent of adults age 55 years and older were desolate and more likely to be living alone and have an income under $50,000 . The pandemic prevalence of social isolation and loneliness increased, in the first phase of the pandemic, where Savage et al. reported that 43 per cent of community-dwelling Ontarians age 65 and older were experiencing loneliness. Of the 43 per cent experiencing loneliness, 8 per cent reported these feelings always or often in the preceding week . Colucci et al. examined feelings of social isolation and loneliness over time among older adults in Quebec. Reduced well-being was reported among participants, which was predicted by older age, lower energy, and social isolation . Moreover, a larger decline in well-being was explained by feeling more isolated by January 2021 by 17.3 per cent . Poverty is also an influencing factor of social isolation, and older adults living alone are more at risk of living in poverty, therefore, facing additional barriers to accessing technology and services . Together, there is a call for a greater understanding of the context of older adults during a lockdown resulting from a pandemic and targeted interventions to mitigate the negative outcomes. Generally, these studies paint a diverse landscape of social isolation and loneliness increasing during the pandemic that requires further understanding of the experiences of older adults and the contextual factors of social isolation. We know that community-dwelling older adults, of which 25 per cent are adults over 65 who live alone, will require more support with age from family caregivers or formal caregivers such as visiting volunteers or paid visiting home care providers . It is estimated that 28 per cent of Canadians are family caregiversamong this group, 26 per cent caring for older adults and 45 per cent caring for those with dementia report experiencing distress including disruptions in social and family relationships, which can contribute to social isolation . One in five family caregivers has experienced social isolation as one of the negative consequences associated with caregiving . Social isolation was also documented amongst paid and unpaid caregivers of older adults, which has also increased among these groups during the pandemic . The majority of paid personal care in the home is provided by unregulated health professionals, mainly women, largely because of relationship and non-monetary rewards, but they face challenges of poor wages, inconsistent scheduling, lack of training, opportunities for advancement, and feeling isolated . The challenges of retention are exacerbated in rural areas where relationships are complex, and the context of the care needs to be considered by health service provider organizations, going beyond the provision of education and training . One study based on home support worker experiences suggested team-based approaches, increased control over scheduling, reduction of risks taken, and isolation experienced to enhance retention . Even when considering the importance of the health and social service workers to enable older adults to age in place, there have been only 16 studies identified in a scoping reviewnone of them PARthat focused on the retention of home support workers. The majority of studies focus on education interventions, with very few addressing organizational approaches, digitalization, and scheduling, ignoring the more complex contextual, gendered, and psycho-social challenges . Before the pandemic, caregivers experienced feelings of isolation, being undervalued, being exposed to risks and illnesses related to stress, but this was amplified during the pandemic. Given the documented health impacts associated with social isolation , several studies have evaluated effective and proposed interventions to address social isolation among older adults. Interventions to facilitate social connection include arts and technology-based programs, facilitated interaction, physical activity, and education . Promising practices demonstrating positive outcomes include community gardens, recreation/exercise programs, regular and robotic pet programs, and computer literacy programs . Technology-based interventions and programs using multiple types of interventions were most successful in alleviating social isolation and loneliness . Far less research has examined interventions for social isolation among family caregivers and formal caregivers, but providers in Quebec made recommendations, including collaboration among service providers to identify those households at risk, enhancing volunteer supports for respite and social interaction . Existing studies have explored Internet and technology-based tools such as virtual support groups and blog exercises for families caring for older adults, in particular, caregivers of older adults with dementia to regain a sense of social belonging . Recent studies have investigated social isolation among older adults during the pandemic. A scoping review revealed that older adults whose social contacts were based outside the home are disproportionately impacted by social isolation and loneliness . Several reviews recommended the use of virtual interventions that could be offered remotely during the pandemic. Telephone-based interventions, including cognitive behavioural therapy, interpersonal psychotherapy, befriending programs, and illness management courses, reduced feelings of social isolation and loneliness . Similarly, web-based interventions included computer and Internet tutoring courses, weekly video calls with family, and mental health modules for issues such as depression and anxiety . Technology has been useful to address social isolation; however, contextual challenges need to be considered. Among health and social service sector providers for older adults, early in the pandemic, cross-sectional online survey results identified regional urban/rural differences, for example, a greater ability to maintain social support services in urban areas compared to rural areas . However, essential services were given priority, interrupting social support services and putting those more vulnerable older adults with dementia and their caregivers in a more precarious position and challenges in identifying those most at risk . While existing interventions may mitigate the negative impacts of social isolation among older adults and caregivers, many of these interventions were not developed through consultations with older adults and caregivers to ensure that the services meet their unique needs . Social isolation interventions and policies are needed to recognize the diversity of older adults and their caregivers. The Agefriendly Communities Outreach Program in Ontario provided an environmental scan of practices, programs, and policy guidance to mitigate social isolation and facilitated knowledge exchange opportunities among community service providers and mobilizers to meet the needs of more marginalized older adults and caregivers during the pandemic . Collaboration among diverse stakeholders as a key approach is recommended to address the real intersecting challenges of older adults, family caregivers, and paid caregivers in the context of their community, during a pandemic . The scarcity of PAR studies to investigate social isolation and related practices, programs, and policies affecting older adults, family caregivers, and formal caregivers reflects a lack of collaboration between service providers, older adults, and caregivers to address social isolation . PAR enabled co-learning between researchers, older adults, care providers, and decision makers and facilitated engagement of those more vulnerable service users and providers with managers and decision makers . Similar to Poulin et al.'s call for older adults and their caregivers to be essential partners in prioritizing and adapting services consistent with their values, preferences, and experiences, this PAR study was developed with people as co-inquirers and co-learners, who sought to understand, explain, and collaboratively act on real problems with those who usually have the power to influence programs and policies . --- Methods --- Study Design The purpose of this PAR study was to better understand the health implications of social isolation among community-dwelling older adults and their formal and family caregivers during the pandemic and to make mitigation recommendations. Hand, Rudman, McGrath, Donnelly, and Sands argued that PAR facilitates the development of collaborative relationships between stakeholders as well as strategies to address social isolation that is based on the lived experiences of older adults, two components that are critical to mitigating social isolation. PAR is therefore well-suited for investigating the contextual aspects of the COVID-19 pandemic that have impacted social isolation among older adults and their caregivers. We anticipated that a PAR study would also strengthen and build new partnerships, facilitate discussions between decision makers, older adults, and caregivers, providing practice, program, and policy suggestions to mitigate social isolation both during and post-pandemic. Every phase of our research was informed by the principles of PAR, which included a commitment to collaborative and equitable partnerships, community empowerment, co-learning among all partners, capacity building, a balance of research and action, and long-term involvement with a commitment to sustainability . To that end, at the beginning of the pandemic, weekly teleconference participants comprising AFP members and stakeholders were invited to join a task force to address social isolation. The TF developed had 17 stakeholders, including local older adults, caregivers, members of non-profit organizations, private companies, researchers, public health, regional health planning bodies, and First Nation community members. Their role was to guide a smaller research team comprising a rural dwelling professor emerita, an older adult volunteer, a city and county government employee, an academic, and a graduate student. The TF met with the research team at specific points during the project between March and December of 2020 to guide the development of the research question, facilitate participant recruitment, advise on document selection, support data analysis, and participate in knowledge exchange strategies. The main research question was, "What are the health implications of social isolation among community-dwelling older adults and their familial and formal caregivers in Peterborough City and County?" Additional objectives were to identify relevant best practices, programs, and policies that will address social isolation among community-dwelling older adults and their family and formal caregivers during and post-COVID-19 pandemic; and collaboratively develop effective knowledge translation approaches that will foster multi-stakeholder engagement and actions to prevent social isolation during and post-COVID-19. Consistent with the tenets of PAR, an integrative knowledge translation approach was utilized in this PAR study through bi-weekly teleconference meetings with the TF from March-June 2020, then shifting to monthly July-August 2020, and bi-monthly with the broader collaborative of AFP, a network made up of 24 older adult volunteers and 66 community-based organizations or businesses. IKT shifts from a paradigm of the researcher as the expert to one where researchers and knowledge users collaborate on issue-driven research with the expectation the research will generate implementable solutions to long-standing problems . Therefore, we involved members of AFP who could use the research evidence of lived experiences of older adults and guiding documents related to the pandemic to inform practice and policy responses in a very timely manner. This approach required a meaningful partnership in the research process, where the smaller research team worked closely with the TF and the broader AFP members to further the refinement of the study findings. In December 2020, a virtual knowledge exchange event hosted by AFP with 55 participants met to discuss a set of actionable recommendations based on the research findings. Once finalized, these recommendations were distributed to the four AFP Working Groups, community members, and decision makers who have been working on identified action priorities in 2021-2022. --- Data Collection Two types of data were collected and analysed: interview transcripts and documents. Semi-structured interviews were conducted with older adults and their caregivers remotely by telephone or Zoom to collect demographic data and rich, in-depth narrative data to better understand the health experiences and proposed solutions related to social isolation. Discussions with TF members, particularly older adults and providers living rurally and in First Nation communities, identified the importance of recruiting frail older adults in a variety of housing settings, those living alone or with fragile support systems, those living more rurally and seeking an Indigenous perspective. The partnership with our TF collaborators enabled decision makers in local organizations to identify federal, provincial, and organizational documents that were guiding pandemic service changes, providing context to the experiences of older adults and caregivers during the initial phases of the pandemic. The documents were also purposively selected and analysed to triangulate our data, thereby increasing the trustworthiness of the study and developing a comprehensive understanding of social isolation among participants . Moreover, the convergence of information in the two data sources served to enhance the validity of themes drawn from these data , as well as minimize the possibilities for biases . --- Interviews Community-dwelling, frail older adults were recruited as they are particularly susceptible to both social isolation and negative impacts of COVID-19 . There are diverse definitions and screening tools for frailty, but common elements established by the National Institute on Aging include multimorbidities, existing difficulty performing activities of daily living, reducing mobility, social engagement, and cognitive abilities . Participants were recruited by TF members through their strong relationships and trust established with older adults and caregivers in the care circle, often overlooked by the health care system. Moreover, social isolation research often focuses on older adults themselves but tends to exclude family caregivers and formal caregivers, therefore we sought to recruit these two additional groups to investigate the relationship between caregiving and social isolation amid a global pandemic. To obtain approval from the Trent University Research Ethics Board, key informants from the AFP social isolation TF helped develop the purposeful sampling framework previously described. Eligible older adults were those identified by TF members as frail, community-dwelling adults age 70 years and older and their family and formal caregivers living within the City or County of Peterborough. Lay language describing frailty based on NIA consensus was used in a letter of information to aid in recruitment, including one or more of the eligibility criteria: limited ability to get around, care for oneself or get groceries, and may have been losing weight, have less energy, have fewer social contacts than they desire, or experience loneliness. Older adults with advanced dementia were excluded given ethical concerns regarding telephone interviews with this population documented in previous research . Residents in long-term care facilities were not included as another Task Force was addressing this population and was beyond the scope of this project. Informed consent was sought after potential participants had reviewed the letter of information. Older adults with self-disclosed early-stage dementia were included in the study, and consent was retrieved from these individuals as well as a proxy. Older adults, family caregivers, and formal caregivers were recruited until there was a redundancy of information or a saturation of themes . In total, 31 participants were recruited for interviews, which is considered an appropriate sample size to ensure data adequacy . Trained research assistants conducted semi-structured interviews on Zoom or by telephone between June and September 2020. The duration of each interview was between 30 minutes and 1 hour, 45 minutes. All participants were asked basic demographic questions and their impression of their physical, mental, and social health, before and during the pandemic; the availability and adequacy of social supports before and during the pandemic; and the types of supports that were needed to improve well-being. Family and formal caregivers were also asked about the care they provided to seniorschanges to caregiving responsibilities during the pandemic and factors that helped or hindered their overall well-being or ability to provide care. Older adults and their caregivers were interviewed separately due to the sensitive nature of caregiving, unless the older adults requested assistance from their caregivers due to mild cognitive impairment. Field notes were documented immediately after each interview. The interviews were then transcribed verbatim and cleaned by separate research assistants. A live transcription software was used to facilitate the transcription process. --- Documents The collection of relevant documents began before the interviews and continued concurrently as interviews were conducted, based on TF suggestions. Table 1 provides a summary of the 10 documents selected for analysis in this study. Purposive sampling was used to select and retrieve relevant documents that provided context to the direct or indirect influences on the health experiences of participants. This included documents from international, national, provincial, regional, and organizational levels to ensure there was representation of different levels of authority/ decision-making that could impact the older adults and their caregivers related to the pandemic and social isolation. Consistent with PAR principles, the key AFP TF members also identified federal and provincial documents that were directing organizational and regional pandemic responses and documents. This information led to the retrieval of additional documents that specifically targeted vulnerable older adults, their families, and friends who assumed caregiving roles and formal caregivers experiencing challenges because of the pandemic. In selecting specific documents, we considered several factors, such as authenticity and meaning . Confirming the authorship, date, and location of publication can be a way to attend to issues of authenticity. The number of documents a researcher should select cannot be determined before starting a study of this nature. Sampling for this study involved reaching a point of redundancy when new insights after collecting new data were not found, and analysing more data did not help develop a new theme . After data were extracted from each document, individual summaries and a chart presenting data from all 10 documents were created. Specifically, data extracted from the documents included: author, date of publication, guiding principles of the document, issue addressed by the document, framing of the issue, practices enabling the issue, and proposed solutions. These data were charted to facilitate reflexive thematic analysis . --- Data Analysis Both interview and document data were analysed, using an RTA approach . RTA is considered a reflection of the researcher's interpretive analysis of the data conducted at the intersection of the data set, theoretical assumptions of the analysis, and analytical skills/resources of the researcher . Lived experience transcripts and contextual documents were analysed through an interpretive/constructivist lens by the smaller research team, with preliminary analysis refined by the AFP TF of older adults and decision makers. RTA emphasizes that researcher subjectivity is a resource rather A planning document to be filled out by Behavioural Supports Ontario clinicians/ team members in collaboration with the persons with dementia and their family members when planning a transition of care during the COVID-19 pandemic. than a problem and highlights the researcher's active role, considering personal knowledge, values, and decisions affecting findings and knowledge production . For example, our "insider" perspectives as older adults, nurses, and caregivers shaped our interpretations of the data and dialogue with the TF members. We were also alert to the social, political, cultural, and economic factors that influenced our perspectives as well as those of the participants . RTA was an ideal method that was congruent with a participatory approach because the TF and participants were able to be involved with theme refinement. Three research team members coded the interview and document data. Multiple coders were beneficial because it allowed the exploration of multiple assumptions to achieve richer interpretations of meaning . An inductive coding process produced a codebook, which was uploaded to NVivo 1.3 and used to code all interview transcripts . Line-by-line coding of the interview transcripts led to the development of an initial coding scheme, which was applied and refined using an iterative reflective process until we arrived at consensus evolving to focused codes: Physical Health, Social Health, Mental Health, and Supports. This reflexive coding process evolved, which helped the research team visualize patterns by looking at the frequency of codes and exemplar participant quotes indicative of themes. Like the coding process of the interview transcripts, RTA of the documents helped identify codes related to the problem/issue, the practices that enable the problem, the implications of the problem within practice, and the opportunities that exist to address the problem. Initial coding by two research team members through an iterative process led to focused coding and themes guiding the analysis of each document. Making comparisons between and among the documents helped shed light on the themes reflecting patterns of shared meaning . The subsequent analysis of both interview and document data involved diagramming, a process that facilitated a broader conceptualization of major themes in the two data sets . Diagramming was a useful analytic tool that helped organize our data and illustrate conceptual relationships . Most of all, diagrams helped explain our preliminary findings to participants and the social isolation TF in very systematic and organized ways. Following a preliminary analysis by the research team, older adult, family caregiver and formal caregiver virtual discussion groups were facilitated separately. The focus groups promoted interaction among participants, enabling a deeper understanding of the major health experiences and supports proposed to mitigate social isolation, enhancing the credibility of the study findings . Diagrams and quotes facilitated further dialogue around sub-themes of mental strain, reduced social and physical health, and supports accessed. Furthermore, the use of diagrams in conjunction with quotes facilitated the refinement of dominant sub-themes represented in the diagrams and emphasized points of convergence between the two data sets. For example, Figure 1 depicts four pairs of principles that are valued in the documents used to facilitate discussion with the participants and other TF members, around the dominant theme of "valuing principles." Participants concurred that health was central to the purpose of all the documents, with much attention focused on providing guidance regarding safety and protection and the means to achieve that through communication and collaboration. The more dominant themes were at the centre of the concentric circles and graphically larger. The dashed lines recognize the fluidity and influence of principles on each other. More implicit were the values of responsibility and support, particularly at the individual level, and more broadly societal values of inclusion and equity, which were less frequently seen in documents. In three separate virtual group discussions, following ground rules and consent, group members discussed points of alignment of documents with the supports that were needed. Each focus group was recorded, and summaries of the discussions were developed and discussed by the research team. Participants who were unable to attend the focus groups but wanted to provide feedback on the preliminary findings discussed during the focus groups provided their input via e-mail. This information was synthesized and presented at a facilitated virtual discussion with the larger AFP network. --- Findings --- Sample A total of 13 older adults, nine family caregivers, and nine formal caregivers were interviewed. There were 23 female and eight male participants. The median age for all participants was 71 years, for older adults it was 86 years, for family caregivers, 68 years, and for formal caregivers, 59 years. Across the 31 participants, 32.3 per cent live alone. Half of the older adults, 22.2 per cent of family caregivers, and 11.1 per cent of formal caregivers live alone. Twenty-six participants reported annual income, and 54.0 per cent of older adults and 67.0 per cent of family caregivers make less than $50,000 a year, whereas 67.0 per cent of formal caregivers made over $100,000 a year. Twenty-eight participants provided information on racial/ ethnic identity and 89.0 per cent identified as white/Caucasian, 7 per cent as Indigenous, and 3.5 per cent as "Other." --- Interviews Four integral themes related to participant experiences during the pandemic emerged from the interview data analysis: sacrificing social health, diminishing physical health, draining mental health, and defining supports. --- Sacrificing social health Much of the dialogue during the interviews centred around the impactspredominantly negativeon participants' social health. Overwhelmingly, participants described feelings of loneliness and loss because they followed the public health guidelines to protect their physical health to the detriment of their social health. Older adults described how their social relationships were severely limited by their decisions to follow public health guidance. In tandem with personal decisions to reduce social interactions, older adults' and caregivers' formal social supports were halted by organizations to prevent the spread of COVID-19, thereby decreasing their opportunities for social participation and enhancing their risk of social isolation: I'd feel lonely sometimes and also, I would feel overwhelmed like I thought I just can't do this… I would feel overwhelmed for having to do everything and then I would feel sad about the fact that [my husband] was dead, and I was on my own. -Family Caregiver #1 Notably, the sacrificing of social relationships was experienced by two distinct groups of participants. The first were participants who were previously experiencing social isolation before COVID-19 6 Ann MacLeod et al. and whose feelings of loneliness and isolation were exacerbated by the pandemic. The second were participants who were socially active and experienced minimal to no self-reported isolation or loneliness before the pandemic but were newly socially isolated due to pandemic restrictions. One family caregiver to her mother with dementia, vision, and hearing loss described how she gradually became socially isolated before the pandemic and how her social interactions were sacrificed during the pandemic: Even, prior to the pandemic, we were socially isolating just out of necessity. Little by little, I mean, people don't want to be around when a person has dementia. It kind of scares off friends and family. They don't want to be around that. It's too hard for them. They don't know. They don't want to have to deal with it. So, you just kind of get used to it… It's pretty much the same as before [the pandemic] as it is after the pandemic. Other than the fact that I had more people coming over to the house and visiting with me, prior to the pandemic. Now, I don't hardly see anybody. I miss the kids… There are times when it's very lonely. -Family Caregiver #5 --- Diminishing physical health The second theme identified was that of diminishing physical health during the pandemic experienced by all participants. For family caregivers, the decline in their physical health often was a result of the increased caregiving responsibilities due to the withdrawal of formal caregiving supports. Older adults who had several co-morbidities experienced further decline. Diminishing physical health for older adults commonly included a decline in their mobility and was also characterized by their inability to attend regular doctor appointments. Diminishing physical health for formal caregivers was less severe and primarily centred around a decline in physical activity: But I was seeing seniors that were a lot more ill than they had been when I was seeing them before [the pandemic]. And that's not something that I can really prove categorically but just from an experience perspective, and the time and the number of seniors that I was seeing in the time that it was taking to look through their situation, they were much more complex, and a lot of them needed to be admitted to a hospital. It wasn't as easy to send them home because they had been struggling and doing the best they could at home. But they weren't thriving at all. They weren't doing well because they're isolated for so long. -Formal Caregiver #7 --- Draining mental health Participants experienced a serious drain on their mental health. Many family and formal caregivers felt mental exhaustion as they were trying to manage caregiving stressors that increased over the multiple waves of the pandemic. All participants were experiencing strain on their emotional capacities due to the lack of social participation available as described in the first theme. Older adults' and caregivers' mental health was also strained as they tried to cope with the worry and distrust that they felt towards others as they related to the spread of COVID-19. Family caregivers, usually older Canadian Journal on Aging / La Revue canadienne du vieillissement adults themselves, were particularly stressed about not having an alternative care plan for their family member if they became ill. Participants' experiences demonstrated how the prolonged nature of the pandemic does not allow for any emotional/mental respite from the anxieties that accompanied living through a global pandemic as a frail older adult, and as a family or formal caregiver lacking adequate personal and professional resources. One family caregiver described a sense of helplessness in caring for older adults when adequate resources were not available: I --- Defining supports In conjunction with the social, physical, and mental health experiences described by the participants, specific supports that were needed to address the health-related challenges they faced during the pandemic were also defined during the interviews. These supports were categorized into three levels: individual/familial, organizational, and governmental with the view that responsibility can be assigned at the different levels to ensure these supports are provided to older adults and their caregivers. At the individual/familial level, older adults, family caregivers, and formal caregivers emphasized the need to prioritize self-care and access the necessary resources to ensure they are taking care of themselves, if possible. Similarly, at the organizational level, enhancing telephone check-in programs and outreach was frequently mentioned by participants as a helpful service. Providing funding to support older adults and their caregivers during the pandemic, tailored to what they needed, was highlighted as a major support that governments should consider implementing. The most requested supports were those that could be collaboratively provided at all three levels, which included technology, caregiver supports, and two-way collaborative communication. Older adults specifically cited the need to be provided with the technologies and related training used to connect virtually : We Meanwhile, family caregivers needed educational resources and better communication from formal care organizations. Communication with family caregivers as part of older adults' circle of care was not prioritized during the pandemic, and so all participants commented on the need for better communication between organizations, older adults, and caregivers: What I found really hard was there was no communication or no contact for months [between us and] our community nurse or social service administrator, and [from] the health administrator with the client. The client had no clue what was going on. -Formal Caregiver #4 Moreover, participants reinforced the need for incentives from both organizations and governments to work in home care . Older adults, family caregivers, and formal caregivers recognized the value of the work that home care providers do, particularly during the pandemic, and emphasized the need for their wages to reflect this value: I got no support. I got no financial support other than the [temporary] $4 raise. But like I have friends who weren't eventhey were getting more money than I was, and I was working my ass off and in the middle of it [the pandemic] and so that to me didn't make a lot of sense. I felt like holy shit the grocery store workers are getting a raise right away like it wasn't even brought up for the nurses in the beginning… -Formal Caregiver #3 I don't think they [PSWs] are paid what they are worth. I think that's something we should look at especially since all of you are going to get old. If you're all going to live into your 90s you sure as heck have to make sure there is somebody there that will at least give you a helping hand. But those PSW people they earn their keep. They really should be appreciated and valued more than I think they are. -Older Adult #1 Resources provided by both organizations and governments that enable older adults to age at home were another focal point of the discussions. Many older adults in this study felt that they do not currently have adequate financial, transportation, or home maintenance supports to age at home: I know that they have, or had started a bus service to get in and out of Peterborough but the bus service is useless to people who live [far], I mean you have to get from your house to where that is. So, it may be helpful for people trying to get in as I understand to [the university] or go into shop. But, you know, 70, 80 and 90 year olds, don't want to go to the mall. They want groceries and I know that some organizations were starting to help that you know if you need groceries or something, let us know, but that took a long time. And so, people in this area have actually created their own little links, which is a good thing. -Family Caregiver #8 While many participants easily defined the support they felt could address the health-related challenges they faced during the pandemic, we observed a hesitancy among older adults and older family caregivers to criticize existing supports directly to the organizations. There was a tendency to express appreciation for existing supports even if participants also acknowledged that their needs were not being met with these existing supports. This same hesitancy was not as prominent among family and formal caregivers, perhaps because they perceive themselves as less dependent on organizations: I just appreciated what I was receiving. And in this regard, sometimes the more you do that way, the more you're cared for. And I've learned often if you keep quiet, some people will just do more for you. And, it's a fact of just showing your appreciation in some ways. And if you're smart enough you'll find a way to show your appreciation. -Older Adult #11 Family caregivers understand their challenges in caregiving and have clear ideas of the supports that would be helpful to them to provide not only respite but also safe social opportunities: I have a vision ofit goes with dementia-friendly communitiesbut I'd like to somehow work with the community, like the art galleries or the museums or anywhere that you would go for an outing leisure activity and have dementia-friendly times. So, even a grocery store and providing a bit of staff training or we provide the volunteer that's trained to maybe stay at the door looking at the things that might cause a barrier for people to engage their community. And I also had the idea for a restaurant so if we found like six restaurants that twice a year or once a month they offer a dementia-friendly dining experience…I think that if we could do that, it would make caregiving roles easier -I think people living with dementia, their quality of life would improvebut I think for caregivers, it would reduce that stress of being so alone. -Formal Caregiver #6 These quotes reflect the need for improved timely, two-way communication required between the decision makers in many sectors that interface with older adults and their caregivers to shape services that are responsive to their needs. Many of the guiding documents created by those in power could have been better informed by including older adults and caregivers in discussions at the beginning of the pandemic, to better understand challenges and supports required. --- Documents Four themes were identified across the 10 documents analysed in this study: valuing principles, identifying problem, setting priorities, and making recommendations. Valuing principles was the main overarching theme, which shaped the problem identified, the prioritization observed, and the resulting recommendations to address the problem. --- Valuing principles This refers to specific values that guided both the intent and content of the documents. Four interrelated pairs of principles were common across all documents: safety/protection, collaboration/communication, equity/inclusion, and care/responsibility. While the concept of health was not defined in many of the documents, it was central to the purpose and problems that the documents were addressing. Safety and protection were principles that were always observed in the context of maintaining health, dominating the content of the eight documents published during the pandemic . The pandemic documents framed the virus as a threat to people's health, and so the protection of health through different safety measures was highly valued in these eight documents. For example, Document 10 included guidance on how to protect the physical, social, and mental well-being of older adults transitioning from one care setting to another, utilizing different COVID-19 safety precautions: Include a summary of strategies to assist in keeping the person well and meaningfully engaged during the first 14 days of their move while they remain in isolation. Consider activities that promote social, sensory, kinesthetic, and intellectual needs… Identify the methods by which family/friends will be able to connect with the person following the move . Identify whether these methods may be applicable for the first 14 days or following a 14 day isolation period… Next, the principles of care and responsibility were observed together across the documents. Assigning responsibility for carerelated actions and resources underpinned the content of many of the documents. In all documents, assigning responsibility was important for establishing accountability in the processes of health care delivery, care planning, and caregiving. For example, Document 3 from Brown and Graham focused almost entirely on fiscal responsibility as it related to establishing new physician billing codes for virtual care. Comparatively, Document 6 placed the responsibility on family caregivers of people with dementia to document their care instructions and get an adequate supply of their medications. Document 5 from a local First Nation community also assigned responsibility to its community members to follow COVID-19 safety measures so that their community could re-open safely, including allowing PSWs back into older adult homes on the First Nation reserve: We have done very well in keeping the virus out of our Community, keeping the risk of transmission low, and the number of COVID cases low in our local area and in the province and across the country. In saying this, we must remain vigilant in our efforts for when the second wave hits. How hard this wave hits will depend on all of us and our diligence. We still have a way to go with this pandemic and each and everyone of us must take responsibility and continue to practice all measures put into place, including social distancing, hand hygiene, and sanitization to keep each other safe. Communication would often be linked with the principle of collaboration in the documents given that effective collaboration should involve communication. The content of many of the documents emphasized the need for communication most often from decision makers to individuals . Communication was a key principle given the need to establish communication channels for individuals to get the most up-to-date information about the pandemic. Many of the documents endorsed collaboration as an important component of organizational or individual responses to the pandemic. However, while collaboration was often promoted as the ideal approach to respond to the pandemic, very few documents provided meaningful guidance to implement a collaborative COVID-19 response. These paired principles emerged in documents from public health and health care organizations as they sought to communicate information about changes to health care services, develop collaborative strategies for COVID-19 prevention, and inform about individual physical, social, and mental health care management. This is reflected in two recommendations from Anderson of the provincial Ontario Health, Document 4: 1. Regions, organizations, and health care workers should work together to ensure delivery of services that support patients'/ clients' full continuum of care. Collaboration should aim to avoid unintended community-wide consequences of resuming care as well as seek to improve the integration of care between sectors and across regions. 2. Communicate regularly with patients/clients and caregivers. It is up to each health care provider to determine how they will communicate to patients/clients and caregivers during this pandemic. Take this opportunity to talk with patients/clients and caregivers about their needs, wishes, and values. Lastly, the principles of inclusion and equity were identified, although to a lesser extent; not surprising, they were emphasized more in the pre-pandemic documents. These principles highlighted the need to, first, identify marginalized individuals or populations who are most affected by the issues discussed in the documents and, second, respond to their unique needs. Inclusion was valued as it related to including older adults in decision-making discussions about their health and well-being. Equity was identified as an important principle to consider in relation to access to different resources for all populations but more so for marginalized older adults. Document 2 which pre-dated the pandemic, demonstrated the valuing of equity: Older adults, particularly those who experience vulnerability or marginalization, may require specific supports to access services depending on their circumstance. All older adults should have access to the services they require regardless of income, ability, or geography. --- Identifying problems The identification of the problem necessitating the creation of the documents was a common theme observed across all 10 documents. The problems identified were broader in scope in three documents and more context and population-specific in the other seven documents . The principles underpinning each document informed the identification and framing of the problem within each document. For example, as mentioned above, community responsibility is valued in Document 5 from the local First Nation community reflecting their self-governance and cultural context; therefore, the problem in this document is framed as a community-wide issue. Included in the content of each of the documents is a statement of the problem. Document 6 presented the problem as challenges: The current situation with coronavirus presents some unique challenges for people living with dementia and their caregivers and families. Social distancing, self-isolation and the associated changes in routine can lead to increased feelings of stress, anxiety, and confusion for people with dementia, and can even make the person's dementia symptoms worse. Many people with dementia also depend on others to help with their daily needs, making social distancing a challenge for everyone involved. --- Setting priorities The third theme identified was the "setting of priorities." This refers to a clear prioritization of one or more aspects of the problem identified in the document. This prioritization is then observed in the specific recommendations laid out in each document. Many of the documents included the principles of safety/protection for physical health. These values are then reflected again in the prioritization of protecting physical health by preventing the spread of COVID-19. Meanwhile, measures to protect social or mental health were not given the same attention in the documents. We see this in Document 4 --- Making recommendations The final theme identified in the documents, "making recommendations," responded to addressing the issue or problem identified in each document. These recommendations were significantly influenced by the principles described in the first theme. The principles may be embedded and explicitly stated in the recommendation themselves, such as checking in/communicating with caregivers for persons with dementia. The principles were also implied through the recommendations, such as assigning responsibility of specific actions/roles to respond to the problem of the document. Recommendations were targeted at governments , organizations, sectors , communities, and individuals measures and public health guidelines. However, we observed a disconnect between the reality of older adults' and caregivers' experiences that were shared during the interviews and the proposed practices, programs, and policies intended to support older adults and caregivers that were recommended in the documents. For example, the lack of two-way collaborative communication between health care organizations, older adults, family caregivers, and formal caregivers was a common concern reported in the interviews. Meanwhile, collaboration/communication was being promoted as an important part of the recommendations to respond to the identified problem, but clearly not being put into practice. Practices, programs, and policies that facilitated virtual supports without physical contact helped protect physical health, but participant supports for training and access to technology were not provided to older adults and their caregivers, suggesting hybrid mechanisms for care to consider the context are needed. Policies to remunerate physicians were facilitated, allowing hybrid delivery of both virtual and in-person visits to clinics; however, the lack of twoway communication between older adults and caregivers failed to acknowledge the financial investment needed for hybrid virtual and in-person health and social care needed to maintain their physical, mental, and social health. The points of divergence were developed further by AFP during the Working Together virtual meeting into recommendations for action by working groups of the AFP and its network members. --- Discussion Our research builds on existing studies about insights on the health effects of social isolation among older adults and their caregivers and provides new insights about how multisectoral collaborations can use PAR as a potential approach to mitigate the negative impacts of social isolation in the context of a global pandemic. While many people eagerly anticipate a return to "normal," our study findings have demonstrated that the norm was insufficient for many older adults, family caregivers, and formal caregivers. The focus for decision makers was to reduce mortality and morbidity during the COVID-19 pandemic, produce guidelines for practice to mitigate the spread of the virus, provide incentives to organizations and practitioners to reduce contact, replace in-person care with remote technology-based visits, and institute policy changes to distribute some of the human resources required to support older adults and caregivers self-isolating for much of the pandemic. Unfortunately, these decisions and government policies that were intended to protect the population's physical health against COVID-19 had the unintended impact of sacrificing social health, draining mental health, and diminishing physical health of marginalized older adults living alone, or previously strained caregivers on lower incomes, often living with several chronic conditions. The stories of our participants demonstrated that known negative impacts of social isolation and loneliness in older adults were ultimately exacerbated by the swift withdrawal of opportunities for social participation . Meanwhile, family and formal caregivers experienced a drain on their mental health and put preventive self-care practices on hold to continue to care for older adults without adequate financial, physical, and human resources. Valuing principles underpinning practices, policies, and programs means facilitating dialogue with decision makers to negotiate feasible recommendations to facilitate social connection amongst community-dwelling older adults and their family and formal caregivers, based on their context. First, stories from our participants reinforce the need for health and social services to scale up outreach and promote access to technology support, and technology-based interventions, as well as physical and social activity programs for older adults and caregivers. Consider, No Isolation, the Norwegian company, who developed Komp, a onebutton computer simplifying digital connections across generations . As of 2021, Komp has nearly 7,000 older adult users in Norway, the Netherlands, the U.K., Sweden, and Denmark, enabling older adults to connect with their families without requiring extensive training or advanced knowledge of technology. Similar technology supports and interventions could be utilized in Canada with particular emphasis on connecting older adults and caregivers in rural and remote locations. Another key consideration, particularly for the federal government in Canada, in the provision of technology and Internet-based supports for social isolation, is addressing the digital divide by extending broadband to rural and remote communities to underserved individuals and communities, including older adults and caregivers, to mitigate feelings of isolation and loneliness . Second, family caregivers continue to be undervalued and overlooked along the continuum of care for vulnerable older adults. Family caregivers should not be sacrificing their social health or draining their emotional health or diminishing their physical health. The family caregivers in this study explained that, before and during the pandemic, they were pushed into "new careers" with little support from formal organizations and governments, resulting in negative personal outcomes. For this reason, it is critical that caregivers, especially family caregivers, receive appropriate educational and financial resources to maintain their own health as well as the health of those whom they care for. Newman et al. reported that e-Health technological tools enable caregivers to access supports virtually, such as psychoeducation, which aims to build caregiver knowledge about the illnesses their family members are faced with and associated caregiving issues. Equipping older adults' caregivers with much-needed supports, such as fiscal and communication infrastructure, to overcome common barriers will ensure that all members of the circle of care do not experience social isolation and do not compromise their own health to meet the demands of family caregiving. Workplaces need flexible family care policies to mitigate the negative impacts of caregiving. Next, participants defined a type of support neededa meaningful collaboration across intersecting organizations that serve older adults and their caregivers for the purpose of enhancing communication with each other to improve timely supports appropriate to their context. It seemed that communities with multiple networks, many affiliated with AFP, developed "workarounds" such as the colour-coded paper posted on a door acting as a communication system in the First Nation community indicating the type of need . Additionally, at-risk older adults living alone were identified by the geriatric emergency medicine nurse to a community organization to facilitate in-person outdoor visits during grocery drop-offs, supporting those most negatively impacted in our region. In Germany, the German Foundation for Civic Engagement and Volunteer Work has focused their efforts on funding a program that supports 500 local alliances for people with dementia, where multigenerational houses and networks of individuals in rural communities come together to support people with dementia and their families . With supportive policies and adequate funding, the support networks for older adults with dementia and their family caregivers can be restructured to ensure they are still connected to their communities, preventing social isolation beyond the pandemic. Further, enabling older adults to age in their homes, if they choose, should be at the centre of all policies, programs, and practices intended to meet the needs of this population, while considering needs to connect socially as part of meeting basic needs. To ensure that older adults' perspectives are considered in all policies, all levels of government and health and social service agencies need to engage with older adults' advisory groups and ensure older adults are members of these advisory groups, valuing principles of inclusion and equity that may influence choices of investments and service provision. Notably, investment in infrastructure like older adults' co-housing communities, which originated in Denmark, has increased quality of life, belonging to a community, and social interaction for residents of these communities . Likewise, innovative transportation options geared toward older adults and their caregivers must be embedded within public transportation systems. Leistner and Steiner's evaluation of the integration of a dynamic ridesharing system into existing government transportation supports the use of this type of transportation to enhance older adults' physical mobility and social participation at a small cost to local governments, while appropriate to local contexts. Moreover, for older adults to age at home, more resources must be dedicated to the home care sector, often comprising marginalized racialized women providing the hands-on care. It also requires attention to policies related to privacy and financial structures to support older adults to age in place, removing barriers, particularly for marginalized older adults. The findings and recommendations for supports align with the World Health Organization's promising practices to address social isolation and lonelinessspecifically, noting the need for rigorous research and collaboration among researchers and decision makers to advance the evidence, develop supportive policies, scale up effective interventions, identifying contexts and people most at risk who would benefit from interventions that affect modifiable factors. Solutions to prevent social isolation exist around the world, and it is imperative that we bring these solutions to our communities to prevent older adults and caregivers from being pushed into social isolation both in and outside of an international public health crisis. As demonstrated in our findings, inclusion is a key principle identified in the study documents, guiding provincial action plans and social isolation research. However, what is lacking are PAR studies, particularly including caregivers to be co-learners to develop contextually appropriate interventions . Building on the reflexive lens of this study, PAR lends itself to opening the opportunity for research with a more critical lens, creating dialogue with decision makers and marginalized older adults and caregivers who confront discrimination based on gender, race, income, rurality to stay socially connected, especially during a crisis such as a pandemic. Age-friendly principles and many regional health and social service planning bodies espouse the principle of inclusion. Why is it that the recommendations proposed by the participants echo those in plans and charters from over 3 decades ago? We support the call for interdisciplinary and multisystem approaches to social isolation that consider the individual, their families, the local context, and the broader society . The next steps are to leverage individual and community resilience, not focusing on the deficits of older adults but further engaging decision makers to reduce barriers to social connection and engagement. Madsen, Ambrens, and Ohl noted that when older adults' capacities are drawn upon rather than agerelated deficits and challenges highlighted when developing policies, community resilience increases. With ongoing input from older adults and caregivers in designing supports needed, a more just, evidence-informed approach to addressing social isolation among our older adults and their caregivers can be addressed, being more prepared for the negative impacts of future crises. While PAR has the benefits of being a nimble, timely research approach to create meaningful dialogue between more vulnerable people, their service providers and relevant decision makers in a particular context, a limitation of this study, is that the specific findings may be less relevant in other settings or cultural contexts. PAR was possible in such a short time frame for this study because of long-standing authentic relationships, strong networks with older adults in leadership positions, and a history of effective collaboration among service providers in a relatively small geographically mixed urban and rural areas with similar goals easily agreed upon. As with our Indigenous collaborators, their approach to addressing social isolation was unique. Unique cultural approaches are not likely applicable to other large urban contexts or where relationships may be more transient. Without these conditions, extra time is needed to establish trusting relationships and an understanding of similarities and differences. We can also not ignore those with no connections or supports who often "fall through the cracks." While older adults were included in all phases of this research project, including making recommendations to decision makers, like other PAR studies, rarely do marginalized older adults hold significant power as potent agents of sustainable change . Particularly, smaller, community-based health and social service providers were extremely strained during the pandemic, so often organizational adaptations to service practices and application of guidance documents were unavailable in writing. Many TF members used the phrase that they did not have documents because "they were building the plane as they were flying it." Communities will always lack resources for social isolation interventions recommended, especially during a pandemic, but older adult volunteers, service providers, and decision makers can adapt our study process for future research and building on this study to enable contextually appropriate responses to mitigate social isolation with a more critical lens that is hopefully sustainable in non-pandemic times. --- Conclusion In conclusion, this paper has reinforced the known negative health outcomes of social isolation, in the context of the COVID-19 pandemic, with older adults and their family and formal caregivers emphasizing mental strain and altered social relationships needing attention. To prevent and mitigate social isolation among these populations, our findings call for organizations, communities, and governments to engage in genuine collaborative communication among the circle of care, enabling older adults to stay connected and supported, caregivers to practise self-care, and outreach serving marginalized older adults living alone and their caregivers. Finally, engaging with age-friendly networks or similar groups to conduct PAR about mitigating social isolation serves to inform researchers, health and social service care providers, regional planning bodies, and government policy analysts regarding meaningful pandemic responses and future planning. Furthermore, these key decision makers need to collaborate in the development and evaluation of the practices, programs, and policies. To create meaningful and sustainable social change to support older adults and caregivers, parties from individuals to governments must acknowledge that social connection is a fundamental social determinant of health and should be addressed with the same primacy as other social determinants of health, including physical health, poverty, education, and employment. Ultimately, lessons learned during the pandemic, therefore, must not be forgotten when COVID-19 no longer shines a light on the needs of the most vulnerable populations.

Cette recherche-action visait à comprendre les répercussions sanitaires des directives qui ont influé sur l'isolement social parmi les personnes âgées vulnérables vivant en milieu communautaire, ainsi que leurs familles et leurs soignants professionnels, pendant la pandémie de COVID-19. L'analyse thématique réflexive de données puisées dans dix documents de politiques et procédures a révélé quatre thèmes : valoriser les principes; déceler les problèmes; établir les priorités et formuler des recommandations. Les entrevues avec 31 participants de Peterborough, en Ontario, ont également révélé quatre thèmes : sacrifier la santé sociale; affaiblir la santé physique; miner la santé mentale et définir les mesures de soutien. Les recommandations destinées aux décideurs ont été mises au point lors d'un événement d'échange de connaissances auquel ont collaboré les participants à la recherche et les membres de l'organisme Age-friendly Peterborough. Les principales conclusions démontrent la nécessité pour les gouvernements et les agences de la santé et des services sociaux du Canada d'améliorer l'accès à des interventions basées sur la technologie, ainsi qu'à des ressources éducatives et financières pour les soignants. Une communication et une collaboration profondes entre les personnes âgées, leurs soignants et les décideurs est également nécessaire pour combler l'écart entre les politiques et les pratiques visant à remédier à l'isolement social.

Background News and social media can play an important role in providing information to publics in cases of new or emerging diseases. It is vital to reach people quickly and share details in a situation in which medical and public health authorities are grappling with how the disease spreads and can best be controlled [1][2][3][4][5]. Major infectious disease outbreaks are highly newsworthy, typically attracting dramatic statements concerning risk [6][7][8]. News media coverage is often a starting point at which people start to reflect on the seriousness of pandemic risk and its implications for their own lives [2,9]. However, news reports can also become hyperbolic or convey misinformation, leading to scepticism and lack of trust in official sources, unfounded complacency or alternatively, generating heightened feelings of fear, uncertainty, depression and anxiety [2,10,11]. The COVID-19 outbreak began to receive public attention in early January 2020, following reports by Wuhan health officials of a cluster of viral pneumonia cases of unknown cause affecting people in that large Chinese city in Hubei province [12]. The news and public health communication environment responding to the COVID-19 crisis has been fraught, frequently characterised by conflicting or rapidly changing information as health authorities and governments struggled to make sense of this new outbreak and identify the best way to control its spread [13][14][15]. COVID-19 news reporters and creators and sharers of social media content have been subjected to continual criticism for disseminating misleading or false information. The term 'infodemic' has been used in some popular media outlets and academic analyses to describe the wealth of 'fake news' and conspiracy theories circulating, particularly in online news sites and social media platforms [16][17][18][19][20]. Analyses of how news reporting and social media content have framed COVID-19 issues across different countries have identified marked variances. Ageism in social media content has been identified, with older people positioned as both more vulnerable and more expendable than other age groups [21]. Racism has also featured in some countries' news reporting, particularly against Chinese people, who have been frequently positioned as to blame for the emergence of the novel coronavirus in Wuhan [22]. Politicisation and polarisation of opinion has characterised COVID-19 news in US television networks and newspapers [23], whereas strong support for government measures was evident in South Korean newspaper reporting [24]. A focus on prevention and control measures, medical treatment and research, and global or local socioeconomic influences was identified in an analysis of Chinese news articles [25]. Information provided in news coverage of COVID-19 in Australia has included reporting of the first cases and deaths and the subsequent rapid spread of the novel coronavirus around the world, accounts of statements, decisions and press conferences held by political leaders and health authorities, strategies to avoid infection, medical controversies and debates and progress towards treatments for COVID-19 and vaccines [5,26,27]. In Australia, very early news reporting focused on the 'mystery Chinese virus' and made continual comparisons to SARS [26]. A study of two major Australian newspapers' COVID-19 coverage found that they were quite slow to begin covering the emerging outbreak, suggesting an initial lack of awareness that it might pose a threat to Australians. Subsequent news reporting largely focused on the social and economic impacts of the crisis. There was little blame or judgement directed at any social or national groups, although panic buyers did receive some criticism [27]. Quantitative surveys to determine how publics were responding to coverage of the COVID-19 crisis in news and social media have identified an association between COVID-19-related news and social media consumption and heightened anxiety and distress in Russia [28], China [29,30] and USA [31]. Malaysian research found that respondents mainly used television and internet news portals to access COVID-19 information. Those who preferred government sources of information were more confident about the control of COVID-19 and believed their government was handling the crisis well [32]. Research in the UK found that people's news consumption surged in the early months of the crisis but gradually returned to pre-crisis levels, with evidence of a growing avoidance of news. A decline in respondents' trust in key sources of COVID-19-related news and information was also noted: particularly in relation to social media sites and government sources [33]. A comparative international online survey included nationally representative samples from Australia, New Zealand, UK, USA, Italy and South Korea [34]. For respondents in most of these countries, government and friends and family were the most trusted sources of COVID-19 information, ranked above the news media and social media. Together with New Zealanders , Australians evidenced the highest levels of trust in their government to give clear and accurate advice on COVID-19. While 58% of Australians said they trusted the news media in general for COVID-19 advice, only 30% trusted information found or shared on social media more specifically. Another online survey conducted early in the Australian nation-wide lockdown [35] found that Australians were consuming news media more than usual due to their interest in and concern about the pandemic. More than two-thirds said that they had been accessing news more than once a day since the outbreak. Half of the respondents were using television reporting as their main COVD news source, while 22% were accessing online news coverage and 18% news on social media. The respondents reported high levels of satisfaction with news coverage of COVID-19 , but it was contributing to people's feelings of anxiety, particularly for women and younger people. Most respondents said that they trusted health experts and scientists and to a lesser degree, government to provide information about COVID-19. Just over half said that they trusted news organisations but less than a quarter of respondents reported that they had encountered high levels of misinformation in the news or social media about COVID-19. Australians agreed that the federal government had done a good job of informing them about the pandemic and how they should respond . The surveys reviewed above are valuable in identifying trends across large populations in attitudes and practices related to COVID-19 information sources. To complement and extend such findings, qualitative social research methods provide a way of investigating people's engagements with personal sources of information about emerging health risks such as family members and healthcare providers as well as with government sources and news and social media reporting in greater depth. This approach provides for explorations of lived experiences in sociocultural contexts [1,2,11,36]. Thus far, few qualitative analyses of Australians' responses to news and social media coverage of COVID-19 have been published. Among other issues, the 'Australians' Experiences of COVID-19′ study investigated people's use and appraisal of information about COVID-19. We wanted to surface the full range of information sources upon which participants relied and those they most trusted: including but beyond media or government sources. --- Methods The study took place during the first 6 months of the COVID-19 crisis in Australia, following the identification of the first Australian COVID-19 cases on 25 January 2020 and the implementation of a nation-wide lockdown from mid-March 2020 [37]. Forty indepth, semi-structured interviews with adults living in Australia were conducted by the second author between late May and late July 2020. In addition to the closing of international borders and some national state borders, the national lockdown included directives for people to work at home where possible, limits on household visitors, bans on public gatherings, the closing of non-essential services and schools, and physical distancing rules. The spread of COVID-19 began to be slowed by April 2020. Restrictions were progressively eased from mid-May 2020 onwards but fluctuated in response to the incidence of COVID-19 community case numbers. Restrictions were re-introduced in the state of Victoria from July until November 2020, following a significant second wave of infection in that state [37]. Due to physical distancing restrictions, the interviews for this study were conducted by telephone. This method also ensured that people living across the nation, including in regional and remote areas, had the opportunity to participate in the study. Interested potential participants responded to an advertisement about the study on Facebook. Participants were offered a gift card to compensate them for their time. Sub-quotas were set and achieved in recruitment to ensure a heterogeneous interviewee group with a spread of participants across gender, age group, and place of residence . Facebook was chosen to advertise for recruitment because of its popularity among Australian adults. At the time this study was carried out, figures on Australian Facebook use show that 60% of all Australians were regular Facebook users, with 50% of the Australian population logging on at least once a day [38]. Using this method of recruitment therefore proved to be fast and effective, and we easily met our sub-quotas. Table 1 shows the sociodemographic characteristics of the participants. The study adopted a qualitative approach that was focused on a wide-ranging interview about the participants' experiences of the novel coronavirus/COVID-19 during the 6 month period following identification of the first Australian cases. All interviews were audiorecorded and professionally transcribed in full. A narrative case study approach was adopted in compiling and analysing the interview materials. This approach sees the indepth interview as a form of shared storytelling, in which participants recount narratives in response to interview questions and researchers formulate their accounts into narratives [39,40]. The second author wrote fieldnotes for each participant soon after she conducted each interview. These fieldnotes were presented in narrative form, drawing on the author's impressions and recollections of how the participants responded to the questions. Once each interview was transcribed by a professional service and returned to the authors, both authors then used the transcripts to augment these notes, inserting illustrative direct quotations from them to configure a detailed narrative case study for each interviewee. These case studies, together with the full transcripts, comprise our research materials for analysis. Some of our findings are reported thematically across the case studies, while in other analyses we present case studies to provide a detailed biographical narrative. For the purposes of the present paper, the set of detailed case studies formed the basis of a topical thematic analysis which focused on the participants' responses to the questions about sources of information about COVID-19. These themes were derived as an iterative analytical process involving both authors working with the research materials of case studies we had developed together with the interview transcripts. This approach to social inquiry is directed at identifying 'making the mundane, taken-for-granted, everyday world visible' through interpretative and narrative practices . As Denzin puts it, human experience 'is a process. It is messy, openended, inconclusive, tangled up'. Hence our focus on interpretation and narrative as modes of analysis: the interviewees interpreted our questions in formulating their responses, and we in turn interpreted their responses in configuring the case studies, identifying themes across the cases and presenting our findings. The first question in the interview prompted participants to think back to how and when they had first heard about COVID-19 and to provide narratives of how they felt about it at that time. This question was followed up by asking participants 'Since that first time of hearing about the coronavirus/COVID-19, what has been the most helpful or useful sources of information for you to learn about the virus?' and 'What has made these sources so helpful or useful for you?'. It is on their responses to these three opening questions that we focus in this article. --- Results --- Initial reactions to COVID-19-related information Given that initial news reports in Australian outlets focused on China and SARS [26], it is not surprising that most participants had first heard about the new infectious disease outbreak through news media sources reporting on the 'mystery SARS-like' cluster of cases in China. The location of the outbreak in first news reports and the comparison with SARS in this early news coverage led people to think that it was a faraway problem that would not directly affect them. Several people drew on their memories of previous outbreaks of novel infectious diseases such as SARS, MERS and Ebola in their responses. For example, Michael initially heard about the virus in Wuhan through television news reporting. He recalls hearing about coronavirus as it was being compared with SARS. He remembered that the SARS epidemic had not affected Australia, so did not think COVID-19 would either. SARS seemed to affect other countries around the world but not Australia. So, I didn't think that it would be as severe as what it ended up, so widespread across the world. So no, I didn't really worry at the time. Greg was even less concerned about the threat of COVID-19 at first, as his initial exposure to news about the coronavirus was via jokes that circulated on Facebook. He remembered that Facebook friends at first tended to make light of the threat of the outbreak: 'I wasn't too sure what to make of it, and enjoyed a couple of jokes when people said "I'm having a corona attack!" and put a photo of a [Corona brand] beer up on Facebook'. It was when Greg heard projections of the number of people that might be hospitalised with the virus in Australia on television news reporting in early March that he began to realise that it was a serious problem. The initial joking on Facebook was countered by the dramatic television news reports of the growing threat posed by COVID-19 to Australians. I had listened to all the news broadcasts: listened with some trepidation to the forecasts of 'the hospitals are going to need thousands of beds', and concerned about that … By early March, I started to pay attention. The initial sheer volume of news reporting and other public messaging about the spread of the coronavirus and measures needed to contain it could be overwhelming for some people. Because of the novel nature of the COVID-19 pandemic and the fast-changing news about it and its potential impacts on Australia, it was common for the participants to observe that they found themselves not being able to look away from news reporting about the crisis once the serious nature and rapid spread of the pandemic worldwide began to be reported in Australian news outlets. Participants commented on the importance of judicious consumption of news and information about the virus to avoid becoming overly obsessed and anxious after realising the risks to Australians of COVID-19. Several participants commented that they began to feel that there was saturation of 'bad news' and fear-inducing announcements from government officials in press conferences and health communication campaigns. Some participants noted the tendency for sensationalism in news reporting and social media activity and the deleterious effects on their feelings of wellbeing. They often talked about 'switching off' from or limiting their exposure to news about the virus as time went on as a way of managing their distress and supporting their mental wellbeing. As Joe commented: when I have looked at the international news and looking at what's happening in America and that sort of stuff, it gets me really worked up and I get very upset about it. I find that quite challenging, and at the same time, I find it very difficult not to look. So, I found it really hard, particularly in the early days, in terms of just not constantly having the news on and constantly hearing about what was going on. It's only probably been in the last two weeks that I've managed to sort of cut that down to maybe two or three times a week, whereas it was two or three times a day. It was just, I had to know what was going on all the time. Several others reported difficulty in keeping up with all the new information being issued from these sources: some of which could be contradictory. For example, Emma described the government-provided information in press conferences or public health campaigns concerning restrictions as often 'confusing'. She noted that some of the restrictions imposed by the government were 'arbitrary' or hard to make sense of. Emma gave the example of the number of people allowed at a wedding or a funeral, a rule which she remembered was constantly changed during the early months of the pandemic: 'It's, like, bizarre and kind of hard to understand'. --- Blame, misinformation and conspiracy theories Many participants were highly aware of the potential for news reporting or social media content to be misleading or inaccurate: problems which themselves have received attention from the news media itself as well as public health authorities in Australia [5,15]. Some people expressed feelings such as frustration, distress or anger around the kinds of information circulating social media platforms. One example is Sarah , whose husband is an essential worker in health services. She was concerned that the misinformation about COVID-19 in the news media and social media could reinforce or sanction careless or negligent behaviours that would place her husband and other frontline healthcare workers at increased risk of infection. She knew from her husband's first-hand experience that the threat of COVID-19 was not exaggerated. My husband was dealing with those patients who are highly contagious, and he was told not to wear full protection and that was frightening. So when people were saying 'It's a hoax, don't worry about it,' I'm like, 'Well, you're putting my husband's life in danger'. So it was really distressing. Other participants demonstrated a high level of scepticism towards the accuracy of information they encountered in the news media. They said that they were careful to try to evaluate the level of risk as it was reported in news outlets, given the news media's tendency towards hyperbole to attract viewers. As James commented: [The media] are going to catastrophise everything and anything, all the information. So, whatever I'm reading, I've got to make sure that I don't just believe it straight away and look into it a bit more and ask some more questions, rather than just saying, yep, okay, I believe that. For Greg , the main source of information to learn about the novel coronavirus and COVID-19 has been television news reporting. He talked about the importance of 'reading between the lines' and being mindful of the polarisation and 'bias of the media' in reporting about the coronavirus. Greg was also concerned about some of the conspiracy theories that were circulating initially, and that the outbreak was not being taken seriously by a section of the community. He was even more vigilant in appraising the validity of news and commentary on Facebook: his second main source of information. I must admit I've become quite careful about reading conspiracy-type theories on Facebook. Yeah, it's a platform for everyone to have their say, but I've discovered that in my own opinion, some theories are quite farfetched. People can be sincerely wrong. Joe said he is surprised by how many people he knows have 'bought into' the conspiracy theories, including his own elderly mother. He perceives these theories as expressed by people who need someone to blame. Joe said that he does his best to counter these claims where he has seen them expressed: which includes in face-to-face interactions or telephone conversations with close family members as well as in social media outlets such as Facebook. There has been some discussion that I've had with family that has been just ridiculous. My mother, who's a bit older and just, I don't know, a bit susceptible to bad information, says all sorts of conspiracy-type things to me, which I've just told her is ridiculous. At one point, I think she was saying that China's done this deliberately, and this was to break the world economy. I mean, the worst one I've seen and heard of, which was from an associate on Facebook, was the 5G theory, which -I just think -I don't know … it seems toa lot of people were buying into that, which surprised me. I think they were desperate to have a cause of something they could point at. --- Most trustworthy information sources Given their caution about news reporting and social media content, many participants talked about being judicious around what sources of information they used to learn about the COVID-19 crisis. They placed an emphasis on trusted, unbiased, reliable sources of information that they assumed were founded on expert medical and scientific advice and research, or on personal experience of the pandemic. Natalia was born overseas and keeps in close contact with friends and family there: including viewing content about COVID-19 they have shared on Facebook. She said that she is careful to check that any news items she sees her friends or family members sharing comes from 'a well-known news source' such as the ABC or the Washington Post news outlet or quoting a scientific study: 'I try to do that, because well, I know how fake news creates fear or hopes for nothing'. Ruth also referred to the ABC as well as the BBC as trustworthy, noting that she uses her smartphone to access their news reporting. I just keep reading on my phone and some articles I discount because I think they're crap, and other articles I think, well, yeah, this seems to make sense … I take the ABC and the BBC as being okay. For most participants, government sources such as the federal government health minister and state premiers and health authorities such as Chief Medical Officers were also viewed as credible. Greg said that he finds information from these sources to be the most helpful, mainly because he believes that ultimately, they have the country's and its citizens' best interests at heart. He positioned himself and other Australians as responsible for following government advice for the collective good of the community and as a way of demonstrating good citizenship. Well that was pretty much the bottom line for me. That, okay, if the government says you've got to socially isolate, well that's what I'll do. I'll take precautions, I'll wear a mask, I'll wear gloves when I go shopping. I did all of that in the early stages. Max spoke about the value he placed on the federal government response communicated in regular news conferences that were closely covered by the news media. He liked keeping up to date with reporting of these news conferences because he thought that they provided the most current and local information about the pandemic and the current restrictions in a situation in which these details could change from day to day. Max found it reassuring and informative that these news conferences and announcements were predictable and appreciated being able to readily access these details using digital news outlets. Even though those news conferences became a bit tedious and repetitive, it was good to know that they were regular conferences … and that you knew that a couple of times a day we were being updated as to what's going on. High value and trust were also placed on the information provided by people known personally to the participants who were considered to have expert knowledge or personal experience of the pandemic. Such sources included friends or family members who work in healthcare, government or science domains. They were viewed as unbiased and therefore more credible than some of the news media reporting. For example, Ruth said that she trusts both her doctor and her brother, who is a scientist, to give her authoritative and factbased advice about COVID-19. I actually discussed it with my doctor, probably three or four weeks ago, because I see him frequently … He said in our particular district there hasn't been any coronavirus cases for three or four weeks and he thought it was quite safe. So, I talked to him about it … I talk to my brother about ithe's a scientist. I think it's factual information and people with scientific backgrounds that provide the information. Sarah noted that with her husband working in a hospital, their family had received a proliferation of COVID-19-related information from his workplace even before the national lockdown took place. She knew from her husband's work experiences that hospitals were engaged in rushed preparations for a predicted surge of patients needing care for COVID-19: 'Yeah that was the word of mouth we were getting. So that was, yeah, it was good in some respects and terrifying in others'. Participants who had family members or friends living overseas also often nominated these people as important sources of details about what life was like in countries such as the USA, UK, Spain and Italy where the COVID-19 crisis was much further advanced than in Australia. Riley , who was born in the USA and still has family and friends living there, observed that: 'certainly once it hit New York, then I was getting inundated with messages from my parents, because it was affecting them very directly obviously'. A small number of participants mentioned faith-based communities or teachings as contributing to their sensemaking around COVID-19. For Greg , it was his fundamentalist Christian teachings that contributed to his growing awareness that the COVID-19 outbreak in China could be serious globally, resulting in the 'end times' he believes is forecast in the Bible. For Riley , the personal risk of infection was really brought home by new measures introduced into the synagogue that Riley regularly attends. I was involved in a lot of stuff in the synagogue and about early to mid-March, early March, they were starting to say we can't shake hands anymore and we can't come close to each other anymore. When they started talking about that in the synagogue, I was starting to really pay attention, I was like 'They're telling me this for a reason!'. I started to take it a bit more seriously, so I'm glad that the people in my religious community were taking it seriously before I started to. --- Bringing sources together As is evident from the participants' accounts outlined above, many used a range of information sources about COVID-19. The relative influence of these sources in some cases changed as the pandemic gathered momentum or as key details about COVID-19 changed over the first 6 months when medical and public health experts were still learning about the ways the novel coronavirus spread, the effects of COVID-19 and how best to contain the pandemic and governments and health officials were struggling to find the most effective and least harmful policy settings. Several people explained the complex processes by which they appraised and made sense of COVID-19 information through a range of sources. For example, Georgia commented that she likes the immediacy of sources of information like Twitter and television news reports but considers them not always trustworthy or reliable. Typically, she will supplement this information through her own online research using government websites and through word of mouth from friends who live overseas and have been more seriously affected or exposed to the COVID-19 crisis. Georgia explained that the government-sourced information is the most helpful for her because it is 'verifiable'. She knows that the government draws on health expertise in formulating its COVID-19 advice and policy. In particular, she finds localised information most useful: for instance when and where it is safe to go outside in her local area, and what actions she should be taking to reduce her own risk of COVID-19 as well as risks to others. It is less important for her to learn about the 'bigger picture' of the pandemic. These practices also help Georgia deal with the plethora of information available about COVID-19: 'Anything where the information is bite sized and verifiable, I appreciate, so I guess in that sense, Twitter is good as long as I then go fact check'. Emma also receives a lot of news through Twitter, preferring to read a range of different sources on that platform so that she is then able to formulate her own views about the issue. She also recounted hearing in the news and social media about people's real-life experiences of becoming ill with COVID-19 and how that was particularly powerful for her. Emma described herself as already living with anxiety pre-COVID-19. She noted that accessing more information and gaining knowledge about COVID-19 made her feel less worried: in part, because it gave her the knowledge to take precautions to avoid contracting the novel coronavirus. Emma was also keen to be aware of what the government was doing to handle the crisis, including how she as an individual could help the collective response. Personally, I find it really helpful to have as much information as possible on things. I think that helps me relax a bit more. I know certain people, it's the opposite, where the more you know about something, reading about something a lot, will make you more agitated. But it was the opposite for mewhere I was like, I would like to know as much as possible about this so I can avoid it and knowing what the governments are doing and knowing what you can do personally to help and so on and so forth. Another example of bringing different information sources together is provided by Darren . He said that he has relied on government-related information in finding out and learning more about the coronavirus and COVID-19. He accesses this information via online government health websites. Darren commented that he finds this kind of information more truthful than the news media 'spin' that is imposed on governmentbased information. I saw the media reports where health ministers and health advisors were giving information out, but to be quite honest I didn't pay too much attention to it, because attached to all that was the media spin afterwards. So I left it alone to a great extent and just relied on the government website and blogs that were from medical personnel. Darren noted that he is cautious about the circulation of 'false information' and 'fake news' on social media. However, he is willing to use social media to access websites and 'serious' bloggers which he accesses as more truthful and trustworthy: 'They are either scientific or they are reliable blogs, if you know what I mean. They are ones that I have read for many, many years'. The description of his evaluation of COVID-19 information sources provided by Mark highlights the importance of the advice offered both by international bodies that can provide general advice and local sources of information, as well as demonstrating that social media sites can be vital platforms for disseminating these details. Mark said that he has 'never trusted the media for reporting anything'. He preferred the World Health Organization's regular media briefings hosted on social media outlets as his chief source of information about the novel coronavirus and COVID-19. Mark said that he used Australian government sources of information as a secondary source to the WHO, to provide more localised information and advice: for instance about guidelines and directives for daily living and how to prevent against contracting or spreading the coronavirus. --- Discussion Similar to previous qualitative research on publics' responses to information sources about new disease outbreaks [1,2,11,36], our findings show that participants were active users of information sources rather than passively accepting news accounts, government spokespeople or social media content as authoritative. The participants demonstrated awareness that misinformation was rife in news outletsand especially social media sitesand that they were taking steps to determine the accuracy of information. Their accounts also highlight the interactions of different forms of information sources, and the sophistication with which participants engage with these different kinds of information. Diverse sources of COVID-19-related information, both international and local, were actively accessed, appraised and engaged with by participants. As was found in survey findings in Australia [35] and other countries [32], traditional media were important sources for participants, as were government sources such as press conferences, health campaigns and websites [34,35] and friends and family [34]. Despite contentions that Australian publics have lost confidence in the advice of public health authorities and governments due to conflicting and rapidly changing information provided [13], our participants demonstrated willingness to trust these sources for information and advice about how to respond to the crisis. Indeed, other research conducted around the same time as our study showed that Australians' trust in government had increased dramatically since the outbreak of COVID-19: largely because they assess government interventions to manage COVID-19 as appropriate and effective [42]. Regular press conferences with government and health officials were important in gaining people's trust and reassuring them that the federal and state governments were working hard to control the crisis. People wanted both very localised information that was directly relevant to them and general information from trusted global health organisations such as the WHO. Healthcare professionals personally known to people, such as their regular general practitioner, were also trusted sources of information. Illustrative of the importance placed on experiential knowledge, the participants referred to the value of having informal discussions with friends and family members who were experts or working in relevant fields, such as healthcare or science, as well as engaging in-person in interactions with groups such as faith-based communities and simply hearing from friends and family who lived overseas about what COVID-19 conditions were like there. Our findings support and extend other research that has highlighted the affective dimensions of engaging with information sources in relation to major health crisis such as outbreaks of new infectious diseases [2,[9][10][11]. Similar to survey-based research in Australia [35] and internationally [28][29][30][31]33], our study's participants reported a high interest in COVID-19 news reports in the initial stages of the pandemic. Some people described feelings of anxiety or distress in response to the plethora of information continually published in news reports and on social media. Others were angry and frustrated about the extent of misinformation that was circulating in the community and online and the potential for it to contribute to the spread of the coronavirus and pose a risk to others. However for many people, keeping up to date with changes in information and news in the rapidly changing environment of the COVID-19 crisis was a form of reassurance and helpful in ensuring they were conforming to best-practice risk avoidance and management. The findings also show how certain key events or experiences made people become aware that the outbreak was threatening Australia and potentially themselves. For some people, this was hearing in the news media about the growing number of cases in their region, drastic government interventions imposed to contain the spread or the identification of infected people in their immediate locale. For others, it was face-to-face encounters or telephone conversations with trusted people or viewing content from friends and family members overseas on social media about how they were experiencing the pandemic in their countries that really brought home the dire threats posed by COVID-19 and what could happen to Australia if the outbreak were not contained. A limitation of our study is that it did not involve a representative sample of Australian adults and therefore the findings are not generalisable to the population as a whole. However, a diverse group of participants was included, and the findings support and provide further detail about the trends identified in large-scale surveys of Australians' news consumption and trust in information sources during the initial months of the COVID-19 crisis [34,35]. --- Conclusions Our findings provide further contextual insights into the complexities and social contexts of these practices and sense-making responses, including how people bring together information from different sources in understanding the threat of COVID-19 and the interactions of digital with non-digital sources. A constantly changing news environment, as was the case during the first 6 months of the COVID-19 crisis, raises challenges for effective communication of risk and containment advice. People can become confused, distressed and overwhelmed by the plethora of information sources and fast-changing news environment. On the other hand, seeking out information can provide reassurance and comfort in response to anxiety and uncertainty. Clarity and consistency in risk messaging is important, as is responding quickly to changes in information and misinformation. Our interview study took place at a certain point in the Australian experience of the COVID-19 crisis . Given the rapidly changing nature of the spread of COVID-19 in Australia since then, including a major outbreak in the state of Victoria and an extended second lockdown in that state, continuing and follow-up research is recommended to better understand how Australians have made sense of and protected themselves against the COVID-19 crisis and which sources have been most helpful for them in doing so. --- --- Authors' contributions DL conceived the project, conducted the literature review, contributed to the analysis, interpretation and discussion of the research materials and wrote the manuscript. SL conducted the interviews, wrote the case studies, contributed to the analysis, interpretation and discussion of the materials and reviewed and commented on the manuscript draft. Both authors read and approved the final manuscript. --- --- --- Competing interests None to declare. ---

Background: A multitude of information sources are available to publics when novel infectious diseases first emerge. In this paper, we adopt a qualitative approach to investigate how Australians learnt about the novel coronavirus and COVID-19 and what sources of information they had found most useful and valuable during the early months of the pandemic. Methods: In-depth semi-structured telephone interviews were conducted with a diverse group of 40 Australian adults in mid-2020 about their experiences of the COVID-19 crisis. Participants were recruited through Facebook advertising. Detailed case studies were created for each participant, providing the basis of a thematic analysis which focused on the participants' responses to the questions about COVID-19-related information sources. Results: Diverse sources of COVID-19-related information, including traditional media, online media and in-person interactions, were actively accessed, appraised and engaged with by participants. There was a high level of interest in COVID-19 information as people grappled with uncertainty, anxiety and feeling overwhelmed. Certain key events or experiences made people become aware that the outbreak was threatening Australia and potentially themselves. Most people demonstrated keen awareness that misinformation was rife in news outlets and social media sites and that they were taking steps to determine the accuracy of information. High trust was placed in health experts, scientists and government sources to provide reliable information. Also important to participants were informal discussions with friends and family members who were experts or working in relevant fields, as well as engaging in-person in interactions and hearing from friends and family who lived overseas about what COVID-19 conditions were like there. Conclusion: A constantly changing news environment raises challenges for effective communication of risk and containment advice. People can become confused, distressed and overwhelmed by the plethora of information sources and fast-changing news environment. On the other hand, seeking out information can provide reassurance and comfort in response to anxiety and uncertainty. Clarity and consistency in risk messaging is important, as is responding quickly to changes in information and misinformation. Further research should seek to identify any changes in use of and trust in information sources as time goes by.

INTRODUCTION Education is usually referred to as a social process or social system, which plays a decisive role in promoting social integration . In addition to viewing education as the dissemination of knowledge and experience, some studies also view education as the transmission of values and a close relationship with culture . Therefore, in a broad sense, education can be seen as a process of changing people's concepts through the transmission of values. In family education, in addition to the task of educating social values , the education of personal values is particularly important, especially for children's current educational background. Intellectual education through educational channels helps families demonstrate the role of positioning, setting educational goals , and supporting children in achieving their goals. The Mekong Delta is widely regarded as an underdeveloped region of education in Vietnam. According to recent statistics, the proportions of the population who have never attended school and who have not completed primary school . These numbers are higher than the average range of the whole country . At the same time, the proportion of the population who graduate from secondary school, high school and university is much lower than the national average point [1]. The author believes that the above indicators are not only related to school education, but also the result of the direct or indirect influence of family education. Considering that Long An is located in the Mekong Delta region and is adjacent to Ho Chi Minh City, my research explores families' perceptions of educational value and educational orientation under the influence of objective and subjective factors. Many studies have shown that not only schools but also families play an important role in education. Feinstein, Duckworth, Sabates showed in their research that it is the family that drives differences in educational outcomes. Families have varying levels of support for their children's goals based on their willingness and ability to take advantage of educational opportunities . When Crawford, Dearden, Micklewright and Vignoles studied the family background and university success of British college students, they also discovered the relationship between family background and their children's academic success. Normally, children from wealthy families go to college more often than children from poor families . Brighouse and Swift also highlighted the fact that children born in different families face unequal prospects. In China, Fengshu Liu's research explained the phenomenon of student dropouts in two rural communities, Chengzi and Bailou, Weichang district, Hebei province, PRC. The research results show that the opinions and decisions on whether their children to continue to go to secondary school are based on the calculation of the costs and benefits of students and families, rather than legal obligations or sense of responsibility . In Vietnam, Lê Thi , Behrman and Knowles , Đặng Thị Hoa , Nguyễn Hồng Hà , Hà Thúc Dũng và Nguyễn Ngọc Anh , Trần Thị Kim Xuyến , Nguyễn Khánh Trung , Nguyễn Văn Tiệp ... and other authors' research also paid attention to the influence of parents' economic factors and educational level determine whether to send their children to further studies or not. Among them, Nguyễn Văn Tiệp used rational choice theory to explain the choice of families in the Mekong Delta region for their children's future education. Nguyễn Văn Tiệp believes that family financial ability has the greatest impact on children's education . This research shows that the important role of the economy affects the education of Vietnamese families. The author believes that, taking the rural areas of Long An province as an example, families not only invest in their children's education according to their inherent economic capabilities but they also have a rational choice to invest in their children's education to obtain long-term benefits. This view may explain why low-income families can still send their children to higher education. In addition, the author is also interested in how children's abilities and efforts can play a role in improving their educational achievement. Therefore, in addition to the influence of economic factors, I pay attention to the parents' educational level and personal efforts, and each person's rational choice explains the necessity of children to go to school or drop out. Halfway through the family, especially those with lower cultural and economic capital. Based on Pierre Bourdieu's view of cultural capital , each social class has its own cultural meaning or structure and is attracted by the socialisation of the family from the beginning, since this habit hides perceptions, thoughts, tastes, evaluations, and behaviors . According to Bourdieu, the habitus characterizes each social group. Habitus do not naturally arise, but are formed in the process of socialization in different environments. These environments include: home environment, school environment, and social environment. This difference can be explained from the accumulation process of the types of capital that individuals learn and receive in each social group through education . I assume that every social group will have views and behaviors about children's education. Especially, the families with higher academic qualifications tend to invest more in their children. Furthermore, based on Barth's rational choice theory, that each individual derives from rational motivation to choose the action that brings the greatest benefit , the authors found differences in the choice of educational actions. Families consider more when choosing to invest in their children's education as a profitable investment. Among them, more educated families understand the benefits of long-term education and accept long-term investments in their children's education than less educated families. It seems that pure farmers are more inclined to rational choices and are more susceptible to external factors, while more educated families have a clearer investment strategy for their children's education. However, because they are all affected by natural and social environmental conditions, the cognition of educational values partially reflects the local cultural characteristics. The author analyses the problem through two themes, namely the cognition of educational value and the factors affecting education; the desire and direction of family education in Hòa Phú Commune. --- METHODOLOGY Combining the scope of the topic, I start from the parents' perception of the value of education and the positioning of the child's educational path, taking the Hòa Phú commune in the Châu Thành district of Long An province as an example. The author uses qualitative data as the main research source, and further uses in-depth interviews, retrospective interviews, narrative research and other methods to understand parents' cognition of children's educational value. Based on results of the analysis, the author understands the parents' desire and direction for their children's education. The author conducted a total of 29 in-depth interviews. Correspondents were selected according to different criteria such as family type, education level, occupation, age, economy, religion and gender. In terms of quantitative research, I used systematic random sampling to survey 150 households in all 5 villages of Hòa Phú commune. These quantitative studies aim to compare the views of children in different social groups on the value and direction of education [2]. --- DISCUSSION --- Family's cognition of educational value and the factors that affect educational level Hòa Phú commune is a newly-built rural commune located in Châu Thành district, which is also known as the "dragon fruit capital" of Long An province. According to statistics in 2020, Hòa Phú Commune now has 1,797 households in 5 hamlets, of which 95% of the population is engaged in agricultural production. In terms of education, Hòa Phú commune has 1 kindergarten and 1 primary school , but no secondary and high schools. Lack of public secondary schools creates difficulties for students due to long travel times and difficult parental management. At present, in order to continue secondary schools, Hòa Phú students fmust enroll in schools adjacent to the communes in the district, such as Vĩnh Công secondary school , Nguyễn Văn Thăng secondary school and the secondary schools in Tân An City. According to the statistics of commune education in 2019, the proportion of commune students completing primary education is 100%, and secondary education is 95%. The proportion of high school graduates entering grade 10 , supplementary education school, vocational high school, and vocational training institutions reached 95.89% [4]. The graduation rate of high school and vocational high school for teenagers aged 18-21 is about 87.22% [3]. Hòa Phú commune has implemented popular primary education for the appropriate age, and the universality of junior high school education reached level 3 [5] in October 2019. From the statistics of commune student enrollment, dropout and graduation rates, it can be seen that the high enrollment, graduation and dropout rates of secondary and high schools in 2019 are relatively low. Sadly, there are currently no statistics on the commune's college students and graduation rates. When considering the level of education, it is obvious not only from the role of the school, but also from the role of the family in guiding children's educational orientation. Families are places where trust and encouragement are built to facilitate children's learning paths. According to Douglas , parents' interest in education is a key factor that affects children's pursuit of academic opportunities . Linking the actual results of improving children's education with family roles, I assume that the stronger parents' perception of the value of education, the better their children's chances of receiving higher education. According to Douglas' analysis, the more parents care about and the more aware of the value of education, the more motivated they are to promote their children's learning. However, the cognitive role is not unique, it is everything. From perception to result, there is also a process that is affected by internal and external factors. It can be seen that in order to improve the level of education, in addition to subjective conditions , objective conditions must also be met. Previously, according to Thạch Phương, in pre-modern Vietnamese society, only a small number of people attended school due to economic downturn and political restrictions. At that time, teachers were extremely respected by everyone , 1989: 574-575). Especially the Long An people respect not only teachers but also educated people, . However, in reality the Vietnamese in the Mekong Delta have not received higher education . There are many reasons for the general lack of education in the local community, such as the abandonment of school due to war , "low economy" , 1989: 574), favorable natural conditions and the imperial government's open policy have prompted the local people to develop the concept of "self-reliance" and 'it doesn't matter if you don't study to become an official' . Many families in the old time "didn't accept children leaving home for school" . Most of these reasons are external factors rather than cognition that does not value learning. Yet, with so much dominance, people put learning behind other priorities. Through our retrospective interviews, the author can learn that natural conditions , the economy , transportation , and consciousness issues are key factors that affect the children's further education. Among them, we can first see that under the direct influence of the natural environment, people consciously choose what they need. If life is easy and they feel that what they have is enough, then they don't consciously work hard. Many people think that even if their children don't go to higher education, they still have a "good" life. An old man in Hòa Phú said: "I can say that they have available land, fields and gardens, and if they don't study, they also have food to eat. In the old time, if people went to school and became literate, they could become an officer. But now, it is not always possible to go to school to be an official, and not easy to find a job" . Favored nature and liberal/open-minded personality , 2001: 1011), ) as well as the easy-going mindset of local people in Southern Vietnam also cause the lack of higher education. Another old man in Hoa Phú said: "In the past, it was easy to to make a living due to the blessed natural environment, so people's temperament was generous and free and easy. They can't stand the shackles of going to school every day. They are unable to follow the school's routines such as attending school, listening lectures, doing homework, so school dropout has become a common phenomenon" . Not all families have the financial means to invest in their children's education despite living in their favorite nature. The factor related to education investment earlier mentioned by previous researchers is economic impact. Đặng Thị Hoa , Phan Thị Yến Tuyết , and Nguyễn Văn Tiệp used survey data from the Mekong Delta provinces to study the education of school-age children, and they all believed that Economic factors play an important role in influencing parents' education and thus their choice of children's education. "Parents' decision to invest in their children's education still largely depends on the financial situation of the family" . "Economic conditions are also the reason why many students drop out of school" , "The decision to ensure that children receive education is a family factor, especially economic factors" . In my observation, the economy does not dominate all aspects of family education, but it has an important influence on education, especially in the previous periods. For a long time, for families without financial strength, the choice of education has been placed after food and clothing. Faced with difficulties, people are forced to choose whether to go to school or go to work early to ensure life. An old woman said: "At that time, it was really difficult in many ways. Everyone must work very hard to take care of food and clothing, so the education of children becomes secondary. My husband and I are both teachers. We still have 5 hectares but still feel difficult. Children must also help their parents make a living so that they can learn as much as possible. At that time, the local people did not have a clear understanding of the benefits of studying" . Therefore, if family members do not have a high determination to study, it is easy to give up. A local man said: "Now, there are fewer dropouts this time around and longer periods of time between dropouts. At that time, the local family conditions were poor and the transportation was inconvenient. This road is the edge of the rice field, and it is easy to fall, but there is not even an ox cart. There are no bicycles at home, but now every household has at least one motorcycle, which was the hardest around 1985" . It can be seen from the analysis that the difficulties and advantages of nature and society affect the educational needs of families. A good natural environment makes it easy for people to earn a living. No need to study to find a job. In addition, the family is mainly engaged in agriculture and does not require a high degree of education. Both are reasons why families do not consciously strive for higher education. In addition, inconvenient road traffic in the past, coupled with financial difficulties, made it a reasonable choice for families to drop out of school. However, things are different now. In terms of universal education, economic and mobility difficulties have been overcome. The advantage disappears when the difficulty is overcome. In the current situation, people are uneducated, unemployed, or engaged in strenuous manual labor without high incomes. Therefore, in the current context, the family no longer thinks and calculates as before, but has to choose another way to meet the needs of society. In the current perception of wealthy families, the results of in-depth interviews show that higher education is not only beneficial for material wellbeing, but also for high-paid mental workers, but higher education also has spiritual benefits, that is, good metality, good behavior, higher status in the community, and better social relationships, and better promotion in career. Therefore, it is a rational choice for the Hòa Phú family to send their children to higher education and invest in career education. Therefore, families must weigh the pros and cons to choose whether to continue their children's studies. Some poor families believe that if they have no money or property for their children, they will try their best to let their children master knowledge and give them a stable career in the future. It can be seen that in families where parents attach great importance to education, both rich and poor are trying to raise their children to adults. The survey results showed that 82% of respondents said they would send their children to complete college. Moreover, lowincome families , the choice of children to complete college is still the highest choice in comparison with the choice of children to complete high school , complete secondary school . Therefore, for low-income families, education is a "mental asset" that parents can give their children. An old woman said: "At that time, I taught my children that our family was very poor, so you must have studied hard. But at that time, I was working from early morning until night, not teaching him at home. At that time, it was very hard here.[…] I did not go to high school, I can only tell my children: Mom works hard, and the children stay at home to take care of their studies. There is a future by studying hard. Obviously, you see that people who study have a happy future, and you see some children who don't go to school, who are now bricklayers and doing heavy work" . In addition, people's knowledge has been improved, and everyone is watching each other and fighting for their children to be "equality with friends". Therefore, in addition to natural conditions, social conditions, economic conditions and other factors that affect parents' educational awareness, we also clearly see the role of surrounding communities. Mr. T.A.T said: "At present, due to the large number of schools and convenient transportation, people in the village also visit each other, learn from each other, and send their children to school" . Most Hòa Phú households realise the value of education in a specific way: education offers the knowledge needed for future life and work. This shows that the value of learning has brought a very real result in people's cognition, rather than they want to be "admired by everyone" aloof. This view is proved by survey results of people's perception of the value of higher education in my research site. According to our survey results, on the general value scale, most people think that the value of higher education is to have a stable job . This is followed by high-income jobs , good knowledge, lifestyle and behavior , and the lowest percentage of social status . Family education investment is related to parents' awareness of children's education. Through in-depth interviews, the author found that the educational level of families mainly engaged in agricultural production is low; however, compared with the past, their views on children's education have changed a lot. In their view, learning is gradually occupying an important and highly practical position, and learning is no longer about literacy, but about acquiring knowledge and a stable job. In addition, to go further, learning is to meet the higher needs of everyone. From the perspective of educational values, families with low levels of education have focused on stable job , while the high education ones have focused on the knowledge, morality and social status that benefits of education than having a stable job. Therefore, the influence of the natural and social environment on the family's perception of children's education can be confirmed. Social changes have led to gradual changes in people's ideas. In the past, family education attached great importance to moral education. At present, the content of education in the family is richer and more diverse, depending on the specific goals of each family... In particular, the value of education has been realised by people in a practical way: through learning, people gain knowledge, acquire skills, and especially gain a stable career in the future. --- The aspirations and educational orientation of families in Hòa Phú The 2019 commune student enrollment and graduation rate statistics prove that education levels have improved. The survey data also reflects the growing desire for education of children from commune families. The survey results show that 54.7% of parents want their children to finish college, 18% to graduate from high school, 6% to complete college education, and 2.7% secondary school graduation. In addition, 18.7% of informants hesitated to answer. More parents than ever want their children to go to college. The survey shows that 100% of parents with postgraduate degrees want their children to have at least a college/university degree, 86.6% of parents with a college or university degree want their children to have a college/university degree. The rate is 71.1% for parents with high school education, 58.5% for parents with secondary school education, and 22.5% for parents with elementary school education. This shows that parents' desire for their children's education is directly proportional to their parents' educational level. Further observing the correlation between parents' educational level and their children's educational willingness, the author found that many parents with lower educational level are hesitant due to lack of educational experience . This shows that most parents with a lower level of education do not have a clear direction and goal for sending their children to higher education. Although they realise the value of learning and consciously strive for higher education; however, they value stable careers more than advanced degrees, so they will not pursue higher education for their children. As a result, social group perceptions and expectations of education are not similar. Most parents who do manual labor only want their children to have a stable life and work, regardless of social status. Ms. T.T.L. stated: "I like my children to go to school and work when they grow up. Just settle down and live a normal and comfortable life. As long as he can learn, I will try to let him learn. Learn to have a stable, leisurely job. It's very hard to do business like me" . Similarly, Mrs. N.T.N. said: "Our grandson has a plan to go to college. He will study for about two and a half years. After graduation, he has a job to support himself. I don't need higher education. I also let him take the university entrance examination, but if he fails, we will send him to a vocational college so that he can have a stable career. Later, he takes care of his future, but parents cannot raise children for life. His father cannot live to feed him forever" . In addition, through our survey, the majority of parents want to send their children to college; however, the proportion of parents who hope their children complete high school is not small. For families who want their children to finish high school, they think that their children will finish high school and go to factories or work in occupations that require the minimum level of high school education. Therefore, if a child graduates from high school, going to college is not the only choice for all families. Although the higher the awareness of learning, the better, but from the perspective of the relationship between 294 needs and abilities, reality shows that if a child cannot afford to go to school, the family can choose to let the child finish secondary school, then let your children learn a craft or complete high school, but don't force them to receive higher education at all costs:. Mr. T.V.T. said: Therefore, in addition to parents creating conditions for children's learning, children also play an important role. In fact, many families need to send their children to higher education, but they cannot realise their wishes due to their limited abilities. Phan Thị Yến Tuyết found in a long-term study of 600 households in coastal Vietnam that there are many reasons why rural students drop out of school. "The learning ability factor is relatively significant " . Of course, in addition to children's abilities, economic factors are important. In explaining the decisive role of the material environment, Halsey, Heath, and Rich compared the educational influence of the family environment with the influence of the family material environment in their analysis, showing the decisive role of the family environment. In addition, the influence of family environment and material environment on family education is also prominent . As for the situation of Hoà Phú Commune, as the author analysed, if the economy used to be a factor hindering the commune's people from receiving higher education in the past, the situation has now changed. The income of Hòa Phú family has increased due to changes in crop structure . Data on 2019 show that the structure of poor households has dropped significantly. Currently, the commune has only 2 social security households and 12 near-poor households . Through my survey, apart from children who cannot afford to study, there are very few cases where parents do not want to send their children to higher education due to financial or family reasons. Mrs. N.T.N.N continued: "My child is in college but he wants to study higher, I create conditions for him, because that is his dream and his future" . Many families, although the economic conditions are not very good, are still willing to send their children to school if they can afford it. As for tuition fees, generally speaking, education fees are now higher than they used to be. However, through the survey, most families believe that the tuition problem does not bring too much pressure to the family. Especially families with sufficient financial resources, it is easier to set goals for their children's learning positioning. Therefore, in the current case of Hoà Phú Commune, economic factors are only a side effect. The main factor affecting children's continuing education is parents' awareness of meeting their children's needs, desires and abilities. Although compared with the past, the family's desire for children's education level has undergone positive changes; however, the educational orientation of children in Hòa Phú community families has a characteristic, that is, the recognition of education level, which represents a kind of desire. Rather than a realistic expectation. The family believes that education is necessary, and does not expect too much from their child. They will not force their children to obey the wishes of their parents. However, in our view, the fact that parents stop at desire can also be considered an advancement in the cognitive process. Because the process of recognizing and shaping the goals of parents should also be considered from the starting point and the goal gap . Because in Pierre Bourdieu's view, an mission should not be measured by its end point , but by the size of the gap between the starting point and the goal. . On the other hand, when examining the relationship between low-and high-educated parents, parents' wishes and expectations for their children are different. A stable rate of career choice is a goal for low-educated parents. This percentage drops when parents are highly educated. When parents are highly educated, they have higher expectations for their children. They also see value in learning beyond. Learning not only leads to immediate, tangible results , but also to long-term appreciation of long-term results, invisibly bringing value to knowledge, ethics, and social relations. It is worth noting that in the education and family education of Hòa Phú Commune, the active role of children in their own learning path is very obvious. The will of the parents is only one lever to facilitate children's learning. Parents influence their children's cognition, helping them understand the value of education and motivating them to learn more. I realise that the improvement of education level requires a process, and the survey shows that the education level of the last three generations is gradually improving. This process is needed to bridge the gap between reality and desired goals. The findings suggest that Pierre Bourdieu's theory of 'cultural capital' fails to explain why poor and uneducated families still insist on sending their children to school. Therefore, the choice of children to pursue higher education and the investment in children's education are not completely influenced by family culture and intellectual capital, but also depend on everyone's efforts and rational choices. Taking into account the educational culture of Hòa Phú Commune families, the survey shows that the attitude of commune families to educate their children is open, not compulsory. The parents focus on the child's self-determination factors and consider the child's role as an active participant in the educational process. Therefore, in addition to the influence of family behavior and social background of different social groups analysed by Pierre Bourdieu's theory of family education, the efforts of individuals also play an important role. In fact, families with the same cultural capital, economic capital, and social capital...just have different outcomes of education. This reflects the rational choice of parents when positioning and investing in their children's education. Furthermore, the study shows the role of individuals and individual effort in choosing higher education despite lower cultural and economic capital. Therefore, Pierre Bourdieu's theory of behavior does not adequately explain the consequences of family awareness and educational practice in Hòa Phú, --- since educational outcomes also clearly represent the highest value and benefit of parents' investment choices. --- CONCLUSION By investigating the views of families on education, the author found that the main characteristics of families in rural areas of Long An Province are as follows: Families have greater awareness and clarity about the value of education; however, different family groups have different investment options for their children's continuing education. This difference is the result of rational choice by parents; In the past, economic factors had an impact on a child's need to learn, but now it has little effect; and A child's learning ability plays an important role in a child's higher education. The findings of this case study are certainly not representative of the entire Mekong Delta region. In order to delve deeper into the issue, it is necessary to expand the research object . In addition, the project also needs to clarify the differences in educational behavior of different social groups . However, the author's findings suggest that families' perceptions of educational values and respect for children's educational abilities and aspirations have changed. This is the basis for guiding children's future learning. The problem is that orientation comes not only from awareness, but from further enhancing the role of parents in fulfilling wishes. Today's industrialised society has a high degree of specialisation, which is suitable for guiding children to learn higher-level and more in-depth majors. However, the family-oriented process also requires the time and effort of both the educational subject and the educational object. --- Conflict of interest: The author declares the non-existence of conflicts of interest. --- Contribution by author: The author is responsible for all components of this work. Funding or funds: No financial support was provided.

The study is to investigate and analyse the perception of parents in Hòa Phú Commune about education and their guidance toward their children's career education orientation in the current period. The study is based on ethnographical fieldwork methodology with 29 in-depth interviews across all five hamlets (cluster sampling) and several observation journeys. The authors spent 3 years working at the research site. The child's continuing education depends largely on his or her own wishes. In fact, there is an interaction between the internal factors (parents' wishes, expectations and needs) and external factors (economic, cultural and social circumstances) of the educational orientation of children in rural families. In many cases, there is a gap between what parents think and what they act about their children's education. Local families rationally choose to guide and invest in their children's education. The research shows how contemporary Vietnamese rural residents view and position their children's education, especially the future vocational education.

INTRODUCTION The state of the adult social care sector in England has been a prominent topic of political and social debate, amplified by the COVID-19 pandemic exposing and exacerbating long-standing challenges for the social care sector . Particular issues have been identified, including funding gaps , workforce sustainability , and an ageing population . Research can support evidence-based change in the sector but with limited capacity, and therefore requires the identification of priority areas to address the issues most relevant and urgent to local care systems . The pressure on social care research capacity -and hence the particular need to prioritise -is increased by distinct challenges . Social care delivery is highly dispersed and localised, being primarily the responsibility of local government in England, and organised through the 152 local authorities across the country. The adult social care sector includes a wide range of different services, often not co-located, and includes a substantial contribution from unpaid 'informal' carers. Moreover, the majority of services are delivered by private sector organisations. Internationally, long-term care systems differ in their funding, organisation and delivery, but in many cases the systems are decentralised and delivery is often contracted out to private sector providers, or is part of a private transaction between individual and suppliers . These features lead to complex research design requirements and recruitment. To advance the feasibility and impact of social care research, it is important to formally identify the research and implementation priorities for the adult social care sector. There are few in-depth reports on the identification of research priorities available for adult social care research, which is in contrast to the situation in health services research . Those prioritisation studies that have been conducted in social care have used different approaches. A prominent example is the study conducted by the James Lind Alliance using a consensus-based approach with multiple stakeholders in priority setting partnership to produce a top 10 of research priorities. Although informative, this report focussed exclusively on social work rather than covering the whole social care sector. In 2019, the National Institute for Health and Care Research School for Social Care Research published a scoping review on research priorities in adult social care , presenting a comprehensive list of 30 research priorities. Stakeholders were consulted for the development of the final list; however, initial priorities were identified through document review. The most recent report of 5 research priorities was published by Cowan and colleagues in 2021 and focussed on rapid prioritisation of innovations in adults social care and social work . The adult long-term care system covers a range of activities and helps support people with a variety of care needs. We can distinguish six main processes involved in providing care support, and these cover: the commissioning of care and associated systems for providing relevant information about care options; the delivering of that care, often divided between prevention support and models of care to support current needs; organisation and development of the workforce ; and on how the care system works with other public services, particularly health care, recognising the interdependencies and the benefits of integration. Box 1 provides further details. The legislative framework governing these processes for publicly funded care in England is laid out in the Care Act 2014 . --- AIMS This study aims to assess the research and implementation priorities for adult social care in coastal southeast England, covering the counties of Kent, Surrey and Sussex. The specific aims are to: generate a set of research questions within each of the six areas of the care system, which we call 'theme' areas for shorthand; and to rank the priority for research between the theme areas. The current study adds to the literature by using a bottom-up approach to priority identification, consulting with stakeholders from the outset. This ensures the identified research priorities are fully informed by the people that the subsequent research will have an impact on. Interestingly, meaningful engagement with community stakeholders from the starting point of the research is still too often overlooked . Furthermore, previously published reports focus on priorities at the national level, potentially not adequately reflecting regional and local needs. Additionally, the studies were carried out before the pandemic, which may have uncovered novel priorities in adult social care. The current study focusses on the regional research priorities for Kent, Surrey, and Sussex. This study was designed within the Applied Research Collaboration Kent, Surrey, and Sussex . Funded by The National Institute for Health Research, ARC KSS supports applied health and care research that responds to and meets the needs of local populations and care systems. The process of identifying research and implementation priorities can support researchers to design studies that are most likely to benefit local stakeholders, as well as safeguarding the relevance and legitimacy of the research topic . Moreover, involving stakeholders from the outset provides the opportunity for successful and meaningful implementation of subsequent research findings . The paper is structured as follows. In the next section we describe the methods used; the results and discussion sections follow, then we reach our conclusion. --- METHODS There are several commonly used methods for the purpose of research priority setting that each have distinct advantages and disadvantages . For the current study, we opted for a mixedmethod approach to address the above aims, employing a qualitative inductive approach with data generated by focus group; and an online survey to maximise stakeholder input. Stakeholders included social care professionals working in various adult social care settings in Kent, Surrey, and Sussex, as well as public contributors. The focus group discussions aimed to identify research priorities for adult social care, and more specifically, relevant sets of interventions and policies within these priority areas that could benefit from evaluation. The survey enabled anonymous comments after the focus groups had taken place and asked participants to rank areas of discussion in order of importance. The REPRISE guidelines for reporting priority setting in health research will be followed as guidelines for what to include when reporting our priority setting process and findings. A separate paper offers reflections on the usability of the online focus group method in social care research. The focus group data were analysed using content analysis of the focus group discussion , adhering to the Consolidated Criteria for Reporting Qualitative Research and analysis of the post-focus group survey. This mixed-method approach allowed us to identify priorities in multiple ways. The focus groups produced qualitative content from which research questions were identified. Additionally, a frequency count of different themes was conducted. The survey allowed for ranking data, as well as qualitative information about the decision-making process justifying participants' ranking. --- ETHICS --- Care Models Care Models are the frameworks by which care is organised. Care models guide and define the way social care services are delivered. They outline how people approaching social care services are treated and provide guidelines for best practice. --- Information This theme refers to the information and advice that people receive when they are navigating the social care system and are trying to arrange their care. Effective information and advice services enable people needing care and informal carers to make informed decisions about their care and how they fund their care. --- Integration Integration refers to the process of establishing integrated care by which heath, social care and voluntary, community and social enterprise services are joint up and collaborate. Service integration aims to improve the effectiveness and quality of services as well as transform the experience of people supported by care. --- Prevention Prevention includes the services and policies aiming to delay, reduce, and prevent the needs for care and support. Preventive services are designed to aid the promotion of wellbeing and independence. The focus on prevention within the Care Act 2014 seems imperative to meet the future challenges of an ageing population and increasing care needs . --- Workforce The social care workforce comprises all staff employed within the social care sector. This includes both public and private care providers, as well as the voluntary, community, and social enterprise sector. --- Commissioning Commissioning refers to the process by which health and social care services are planned, procured, purchased and monitored . The commissioning concept was introduced in the 1990s aiming to encourage greater efficiency, responsiveness, and innovation by making care providers compete for resources. four local authorities in which recruitment took place: Kent County Council, Surrey County Council, East Sussex County Council, and Brighton and Hove City Council. All participants provided written informed consent before participation. --- FOCUS GROUPS Design and participants Six online focus groups were conducted, each including social care professionals employed in East Sussex, Kent, Brighton and Hove, or Surrey, as well as public advisors. The sample comprised thirty-seven participants . See Tables 1 and2 for an overview of the focus group samples and a description of participants' job roles. An additional online focus group with seven participants was conducted, in which members of the public were consulted. The total sample included 44 participants. On average, focus groups included 6 people , which is in accordance with previous research . Data collection was completed at seven focus groups, when data saturation started to occur, which is in accordance with research on sample sizes for data saturation . Participant recruitment was guided by the dispersed and varied nature of the social care sector. Recruitment was also informed by the aim to maximise the variety of perspectives of stakeholders who had the potential to benefit from the identified research priorities. Participants were therefore selected using a combination of different sampling methods that were employed simultaneously: maximum variation sampling , critical case sampling , and chain sampling . Potential participants were identified through a web search and by building on our existing networks. Participants from the local authorities' social care teams were identified in consultation with the local principal social workers. Participants working in care provision or in the VSCE sector were individually approached. Members of the public were recruited through social media and newsletters. All participants contact happened via email. Interested participants were sent a link to an online information sheet and consent form on Qualtrics , a cloudbased platform for the development and distribution of web-based surveys. --- Discussion guide Based on consultation conversations with principal social workers in the participating local authorities, discussion guide for the focus groups was developed based on theme areas described above . Along with a short definition of the theme area, a number of example innovations/questions were provided. This discussion guide was packaged into a visual form . --- Online platform Microsoft Teams was selected as the teleconferencing application for the current study. Microsoft Teams is a part of Microsoft 365 and Office 365, which delivers advanced security and compliance capabilities. Network communications in Teams are encrypted by default, and therefore provide a safe medium to conduct discussions virtually. The meetings were accessible by unique invite link only. --- Post-focus group survey An online survey was created using Qualtrics and sent to focus group participants after participation. Participants were asked to rank the six discussion themes from most to least important. Participants could elaborate on their ranking in a free-text box. They were asked to 'elaborate on their ranking in 5-10 sentences'. These data were analysed to gain insight into participants' decision-making processes when ranking Table 1 N per Online Focus Group. --- JOB ROLE N Local authority adult social care management team 9 Social worker 8 Voluntary sector 9 --- Care organisation 6 Other 12 Total 37 Table 2 Participants' Job Roles. using content analysis. Additional data on accessibility, online communication, and social participation were collected, which have been reported in Keemink et al. . --- Procedure In advance of the focus groups, participants were sent a prompt email containing the visual that introduced the potential themes areas of the focus groups . Participants were asked to formulate which themes they would deem a priority area for social care research and implementation in their local area. They were also prompted to think about local innovative policies and interventions relating to these themes that could potentially benefit from further evaluation or implementation. Participants accessed the focus groups with a unique invite link and entered the Microsoft Teams meeting through a waiting room to ensure that only invited participants would have access. Each focus group included a maximum of seven participants, with four research staff present. All meetings started with a short introduction by the lead researcher reiterating the aims of the study and explaining the order of events. Subsequently, screen recordings were started. The discussion revolved around the identification of examples of innovative policy and practice developed locally that could benefit from research support with further evaluation and/or implementation. Participants were encouraged to think about which research questions would support their practice. Once the discussion had concluded, the recording was stopped. Participants were reminded about the post-focus group survey and thanked for their time. Focus groups took place between June and October 2021 and did not last longer than 2 hours. Participation of professionals was voluntary. Members of the public each received a £20 Love2Shop® voucher as remuneration. --- Analysis Onwuegbuzie et al. provide a comprehensive framework for the analysis of focus group data, from which a tape-based analysis was deemed most appropriate for the current study. We created an abridged transcript of the video-recordings of the seven focus groups to centre the analysis around data relevant to the research question. Data were analysed at the aggregated level. The transcripts were examined for content that expressed a specific research priority for the region. Each identified research priority was treated as a separate meaning unit and formulated as a research question. Subsequently, each meaning unit was categorised as one of the six themes . Categories were allocated by two researchers, and disagreements were discussed and resolved with a third party if necessary. Examples of the categorisation process are provided in Table 3. Frequency of themes was analysed to establish an order of priority. We used the Consolidated Criteria for Reporting Qualitative Research for presenting the analysis. We followed a top-down approach when analysing the transcripts, based on theme area in correspondence with the focus group discussion guide. When each meaning unit had been categorised, discussions were analysed for frequency of themes discussed, which yielded the following order of frequency . A ranking was created in line with the frequency the corresponding theme area was discussed. The quantitative online survey findings were analysed in Microsoft Excel. The qualitative data deriving from the open-answer questions were analysed for content complementing the quantitative survey data. A ranking of theme areas was made by scoring each response and ordering the theme areas by average score . --- RESULTS --- CONTENT ANALYSIS OF THE FOCUS GROUP DISCUSSIONS Table 4 presents the descriptive statistics of the order of frequency. Research questions related to Care Models and Information were discussed most frequently, followed by Integration and Prevention. The themes Workforce and Commissioning had the lowest frequency of discussion. Subsequently, meaning units extracted from the transcripts were examined for overlapping research questions, resulting in a final list of 46 research questions having priority for local care systems and the public in Kent, Surrey and Sussex, presented in Appendix 1. The coding tree representing the content analysis can be found in Appendix 2. --- ANALYSIS OF SURVEY DATA The post-focus group data consisted of both quantitative data and qualitative data . The ranking data were analysed in Excel and provided an additional measure of participants' research priorities for adult social care. The survey was completed by 28 of the 44 participants . --- Ranking Participants were asked to rank the six themes informed by the Care Act 2014 according to what they thought should be priority themes for research and implementation in adult social care in Kent, Surrey and Sussex. Table 5 shows the average ranking based on the survey results . --- Qualitative survey data Analysis of the content of the free-text fields revealed the following insights: --- • Professional relevance The data revealed that some participants based their ranking on the relevance of the six themes to their professional practice. Another reason for experienced difficulties with ranking the themes was that participants considered them inter-related. Their apparent overlap hindered the ranking process, as demonstrated by the following participant quote: "It's hard to disaggregate these in a complex system." These insights into participants' decision-making processes helped put the data in context, which we will elaborate on in the Discussion section. Additionally, the qualitative survey data delineate participants' reasonings about why each theme is important. Table 6 presents participant reasons reported per theme. --- DISCUSSION The current study extended the literature by taking a bottom-up approach to the identification of regional stakeholder-informed research and implementation priorities for adult social care. Previous priority-setting studies within adult social care are scarce and, although valuable, focussed exclusively on national priorities and did not always include stakeholders from the outset . This study was designed within the Applied Research Collaboration Kent, Surrey, and Sussex , supporting applied health and care research that responds to and meets the needs of local populations and care systems. We aimed to identify regional research priorities that would subsequently inform the development of novel research projects in consultation with stakeholders. The process of identifying research and implementation priorities can support researchers to design studies that are most likely to benefit local stakeholders as well as safeguard the relevance and legitimacy of the research topic . Moreover, involving stakeholders from the outset provides opportunities for successful and meaningful implementation of subsequent research findings . We employed a mixed-method empirical design, using focus groups and online surveys for data collection, involving a wide range of adult social care stakeholders. Corresponding with our main aim, we were able to produce an extensive list of 46 actionable research questions that reflect the priorities of the local adult --- THEME PARTICIPANTS' QUOTES EXPLAINING IMPORTANCE Prevention "Too often, clients arrive at our doorstep in a position of need that has been allowed to escalate to a crisis point due to lack of effective early intervention." "I think more should be done to help people to stay well and get help with lower-level issues to avoid escalations." --- Integration "Better integration of services would lessen the chances of people falling through the gaps." "We need to ensure that planned integration is a genuine benefit and is designed with an evidence base in mind." Commissioning "Commissioning is important -we need to ensure we have the right services for all." "Commissioning the right services in the first place is crucial." --- Care Models "We need to think differently about care models moving away from traditional models of care to think more creatively about how we can better meet people's needs." "I prioritised Care Models as I am interested in what 'person-centred truly means in practice, and how this is evaluated." --- Workforce "Workforce should be a priority as organisations, systems, working practices etc are only as effective as the people within them and applying them." "Workforce is key to it all, without a valued workforce paid at a reasonable rate, the rest won't be able to be addressed." --- Information "A good number of clients are unaware that most services are chargeable until financial assessment." "The system is hugely complex to navigate. Accessing accurate and timely information about what adult care services are available is so important." Table 6 Ranking as Established Through Online Survey and Elaborations. social care sector and the local public . The identified priorities were validated during an online symposium event during which participants of the focus groups could reflect and comment. The research questions identified in this study aligned with findings from previous priority setting exercises to some extent, primarily in relation to integration between health and social care services. For example, the Adult Social Work Top 10 published by the James Lindt Alliance includes the question 'Does partnership working between adult social workers and other health and social care professionals result in better outcomes for people using services?', and similarly, Cyhlarova & Clark's list of 30 priority research questions includes 'Focussing on the perspectives of adult social services, their users and carers, what approaches to integrate working with a range of public services are effective and cost-effective?'. As an extension, we present additional research questions on the topic of integration, highlighting specific settings and stakeholder groups. Overlap with Cowan et al. was less apparent, potentially due to their specific focus on rapid evaluation. There was more substantial correspondence between the findings from the current study and the priorities identified by Cyhlarova & Clark . Both studies include a longer list of priorities and focussed on the wider social care context, rather than exclusively on social work. However, the current study offers a novel bottom-up approach, including stakeholders from the outset to ensure reflection of the needs of the social care system. Furthermore, by gathering data in a regional context, we were able to reflect region-specific priorities. For example, we identified several questions relating to strengths-based care practice. Interestingly, Cyhlarova and Clark highlighted that they expected more specific research priorities on strengths-based working in the years following their publication. --- RANKING PRIORITY AREAS The study's mixed-method design generated an interesting pattern of results by gathering data on the ranking of research priorities in two distinct ways. Bryant et al. note that a mixed-method approach allows for the ability to capture the complexity of the studied phenomena and offers stakeholders multiple avenues for input. The ranking based on focus group discussions differed from the ranking based on survey responses . There were two notable differences. The first one relates to the placement of 'Prevention' and 'Care Models'. Survey respondents ranked 'Prevention' as their first priority, whereas this was ordered fourth based on the focus group discussions. Similarly, 'Care models' was ranked first following the focus group discussions, and fifth based on survey responses. The second difference between two rankings comprises the placement of 'Information' and 'Commissioning'. Focus group analysis identified 'Information' as second priority, whereas 'Information' was placed last based on the survey analysis. 'Commissioning' was ranked last in the focus group analysis and ranked third based on the survey responses. There are several explanations for these discrepancies. Firstly, participants' elaboration of ranking decisions indicated that they experienced difficulties ranking the priority areas, as all areas were viewed as essential for further research. A sector in crisis in combination with underdeveloped research capacity may explain why stakeholders express difficulties establishing an order. Secondly, different ways of measuring priority may have led to slightly different results. In the survey, participants were explicitly asked to rank the themes, whereas in the focus group discussions, we used the implicit measurement of frequency of discussion. Lastly, the survey ranking was based on participants' understanding of the different themes, whereas the ranking derived from the focus groups was based on researchers' deductions of discussion topics. It is possible that there was a difference in understanding of the themes between participants and researchers, despite the explanatory visual . The higherorder themes were of a more abstract nature, which can potentially be experienced as confusing. Moreover, the themes cannot be seen as completely unrelated, which further complicated the process. Indeed, participants commented on the inter-relatedness of the themes. Ultimately, these discrepancies do not devalue the findings, but rather underline that it is insightful to use a mixed-method design and offer stakeholders various ways of providing input. Despite the identified differences in ranking, other themes showed a similar pattern. 'Workforce' consistently occurred in the bottom half of the ranking order. The theme 'Integration' was consistently top ranked across the methods, indicating that this is a significant priority for research. The content analysis revealed that a lack of integration within social care and between health and social care is a notable issue and knowledge on facilitation and effectiveness of integration is required. These findings will be considered in the selection of the research questions for primary funding proposals. However, it is important to reiterate that from both the focus group discussions and the survey responses, it became evident that all themes have priority for research. The research questions identified reflect the issues most relevant and urgent to local care systems within Kent, Surrey and Sussex, and provide an important foundation for social care researchers. Regional prioritisation is informative, and it allows for the identification of context-specific stakeholders, but it is likely that the identified priorities carry relevance for other areas across England. Bryant et al. highlight that none of the priority setting studies included in their review assessed whether identified priorities had the potential to impact policy or practice, whereas the principal aim of priority setting exercises in research is to inform the design of relevant studies to support evidence-based practice . --- IMPACT Prioritisation studies can identify areas of activity that can most benefit from research, but to realise that potential, account needs to be made of the capacity to undertake research in that area , reflecting that the feasibility and cost of research will vary between topics. Prioritisation is the start of a process in this regard, leading to a range of possible consequences, including: 1) development of research funding applications, 2) setting-up stakeholder engagement activities to support the research , 3) securing meaningful contribution to the research from people with lived experience, helping to co-produce any research, and 4) ongoing networking with parties interested in outcome of the research. --- STRENGTHS AND LIMITATIONS This study builds on previous studies by exploring regional as opposed to national research priorities. It allowed us to uncover more specific research questions that truly reflected the needs of the local and regional care systems and care recipients. Furthermore, the project was strongly rooted in meaningful stakeholder engagement and involvement. Input from stakeholders was the starting point for each step in the priority setting process, which is still too often overlooked . This did not only support later impact and implementation, but also ensures the subsequent research and findings will be relevant to practice and public. The process of stakeholder-informed priority setting is not comprehensive. Naturally, qualitative preference studies will reflect the views of the people participating in the study. There are limits to which the results can be generalised. Nevertheless, we have endeavoured to be thorough by organising online focus groups in several localities within the region, as well as optimising our recruitment strategy to target a wide variety of stakeholders. This study focused on the Kent, Surrey, and Sussex region, for which we were able to recruit a good cross-section of participants. Although context is important , it is not difficult to speculate that similar priorities exist in other parts of England. During focus group discussions, participants often spoke in general terms rather than discussing particular context-specific issues, suggesting generalisability to some extent. Future studies might want to focus on smaller geographical areas to bring out local needs even further. There are also limitations to the methods we used. We approached the ranking of priorities only at a higherorder theme level, and not at the level of the specific questions. Moreover, there are more nuanced techniques for preference elicitation and ranking. Time and capacity constraints resulted in the selection of a straightforward approach. Nevertheless, we fulfilled our main aim of identifying priority research questions, and these were validated during an online symposium with participating stakeholders. Furthermore, we did not explore differences between the participating localities, or differences between social care professionals and members of the public. These are potential questions to explore for future research. --- CONCLUSION In this multi-method study, we identified stakeholderinformed priorities for adult social care research in Kent, Surrey, and Sussex with the aim of gaining information on the regional research agenda. Overall, there was consistency between the findings of the qualitative and quantitative methods, with some highlighted exceptions. Research into optimising integration within social care and between health and social care was identified as a key priority for the area. The extensive list of research questions produced in this study can support social care researchers to conduct relevant studies that address pressing issues for care systems and the public. --- DATA ACCESSIBILITY STATEMENTS The data that support the findings of this study are available from the corresponding author, JK, upon reasonable request. --- ADDITIONAL FILE The additional file for this article can be found as follows: • Appendixes. Appendix 1 to 2. DOI: https://doi. org/10.31389/jltc.208.s1 --- FUNDING INFORMATION The --- COMPETING INTERESTS The authors have no competing interests to declare. ---

There is an urgent need for sustainable change in the social care sector, and research plays an essential role in the identification of priority areas. Thus far, there have been few priority setting exercises within adult social care research. The current study explores regional priorities for adult social care research in Kent, Surrey, and Sussex. Stakeholders were consulted from the starting point of the project, ensuring that the identified research priorities were fully informed by the people that the subsequent research will have an impact on. Objectives: Our main aim was to identify research priorities for adult social care within the region, and more specifically, relevant activities within these priority areas that could benefit from evaluation.We employed a mixed-method design using online focus groups with social care professionals (N = 37) and members of the public (N = 7), and an online survey following the focus groups (N = 28). Focus group discussions were informed by themes based on The Care Act 2014. Findings: Content analysis was used to analyse discussions, which yielded a list of 46 actionable research questions. Rankings of discussion themes were produced to establish order of importance.We approached the ranking of priorities only at a higher-order theme level, and not at the level of the specific questions.The extensive list of research questions produced in this study supports social care researchers to conduct studies that address pressing issues for care systems and the public.

Introduction On March 11th, 2020, the World Health Organization declared that the coronavirus disease 2019 outbreak had become a pandemic [1]. In December 2019, reports emerged from China regarding the initial detection of SARS-COV-2 as the source of the pneumonia outbreak of COVID-19 [2]. On January 26th,2020 the Jordanian National Epidemic Committee and the Jordanian Ministry of Health had met to put a plan in place to manage the pandemic. The recommendations had included the designation of several hospitals as treatment centers for prospective patients with COVID-19 and established protocols to prevent the spread of the country's infectious disease even before the first case of COVID-19 was reported [3]. The Jordanian Ministry of Health had followed the Epidemic committee's recommendations and opened five hospitals located in different areas around the country designated for treating patients with COVID-19. Ministry of Health had equipped these hospitals with ventilators, personal protective equipment , including disposable gowns, masks, gloves, and face-shields, and trained infectious disease medical staff [3]. Besides, the Jordanian Ministry of health advised all health-care workers from different sectors to wear PPE and implemented quarantine policies [3]. The first case of COVID-19 in Jordan was reported on March 2nd [3], and on March 15th, the government closed the country's borders, suspended schools, banned public gatherings, and issued a stay-at-home order [3]. On March 17th, after a case of COVID-19 was traced to a wedding in north Jordan, the government implemented a curfew [4]. On March 25th, the government lifted the curfew partially and allowed essential services and schools to remain closed. Ministry of Health mandated social distancing, masks in public, and the self-quarantining of asymptomatic positive persons. By the end of April, there were 451 registered cases and 8 deaths [4]. By mid-August, the COVID-19 situation in Jordan began to worsen, with the recording of 20-30 cases per day and toward the end of August, the daily cases were 30-40 [4]. This quickly escalated to several hundred and then to several thousand cases a day, most likely due to a lack of public compliance with recommendations; consequently, the government imposed stricter safety measures and penalties for non-adherence [4]. In Jordan, during the month of August, there were 2,034 confirmed cases , 456 people receiving treatment, 1,508 recovered cases, and 15 deaths [4]. On the global level, by August 15 th , 300,000 health-care workers worldwide had contracted COVID-19, and 2,500 had died [5]. Further, over 1.8 million new COVID-19 cases and 38,000 new deaths were reported worldwide during August; this meant a cumulative total of 25 million cases and 800,000 deaths since the beginning of the outbreak [6]. During the month of November, a total of 817 cases had been recorded among nurses, representing 5.5% of healthcare workers, and 26 COVID-19-related deaths had been recorded among physicians [7]. It should be noted that the figures above regarding case numbers among health-care workers almost certainly do not reflect the actual number of cases among health-care workers, as some infected people exhibit mild symptoms or no symptoms, meaning they are unlikely to be tested [8]. The COVID-19 has significant negative impacts on health-care workers' psychological health, fostering issues such as anxiety, depression, and sleep disturbance [9]. This indicates the necessity of providing psychological support for health-care workers, such as by implementing occupational health surveillance programs that train and educate health-care workers in terms of their ability to address the infectious disease and associated psychological distress [9]. Moral injuries are a form of psychological distress that result from performing an action that contradicts one's own moral and ethical code; such incidents can produce emotional guilt, shame, and anger [10]. These symptoms can contribute to mental-health difficulties, which can lead to either psychological injury or psychological growth [10]. Whether an individual experiences the former or latter consequence is likely to be influenced by how he/she is supported before, during, and after the incident [11]. Health-care workers have been found to experience moral injuries, as well as isolation, and at risk for occupational injuries, and lifethreatening situations [9]. Occupational defines as injuries relate to any disease caused by any biological agent that can be experienced while working or while commuting to work [12]. As a result of the pandemic, rapid spread and the associated increased mortality rate, the pandemic has caused public-health issues worldwide; further, the stress people experience in response to this situation has also had a severe negative effect [13]. Regarding health-care workers, COVID-19 has caused issues such as high health-care demands, increased patient mortality, emotional and physical stress, and rationing of health-care supplies [14]. Further, rapid increases in the number of suspected and confirmed positive cases, low supplies of PPE, overwhelming work-loads, widespread media coverage of the pandemic, perceived inadequate organizational support, and an increased risk of contracting the disease and transmitting it to one's own family have also caused psychological distress among health-care workers [14][15][16]. It is essential to consider both the psychological and physiological influence of the pandemic on health-care workers. Failure to assess and address psychological responses to pandemicassociated stressors can negatively impact health-care workers' physiological and psychological functioning [13]. Notably, during pandemics, health-care workers who provide care to patients are among the populations most likely to experience psychological distress, including depression and anxiety [14][15][16][17]. Previous studies of COVID-19 pandemics have revealed that the psychological effects of infectious disease outbreaks can last long after the event, negatively impacting psychological well-being [18] and causing post-traumatic stress disorder, depression, and stress among health-care workers [19,20]. In the context of the pandemic crisis, health-care workers are expected to deal with patients' traumatic experiences and the unexpected loss of friends, family, and colleagues. As a result, health-care workers are affected by psychological distress, including depression, anxiety, and stress [21]. Batra et al. [22] conducted a meta-analysis to provide new evidence related to COVID-19 impact on health-care workers' psychological well-being. Among the main factors identified as causal in psychological distress are anxiety, depression, stress, post-traumatic stress syndrome, insomnia, psychological distress, and burnout. Higher anxiety and depression levels were more prevalent among females than males and nurses compared to doctors and front-line workers compared to second-line health-care workers [22]. There are four categories of social support: "emotional," "appraisal," "informational," and "instrumental" [23]. Social networks include an individual's family, friends, neighbors, and other close significant persons [23]. For health-care workers, social support reduces occupational stress and prevents common psychological distress and psychiatric symptoms; however, coworker support is also significant for health-care workers, as it impacts self-efficacy and professional efficacy [24]. Notably, negative social support is associated with stress and anxiety among medical staff [15]. COVID-19 is an infectious disease that has affected virtually every nation in the world. Research has currently focused on addressing the general population's well-being with little attention being directed toward health-care workers' psychological distress. Therefore, the present study aimed to assess the fear, depression, anxiety, stress, social support, and the associated factors among Jordanian health-care workers during the COVID-19 pandemic. Also, we aimed to investigate the impact of sociodemographic characteristics on these variables. Through this analysis, we determined that health-care workers in Jordan have high levels of depression, anxiety, stress, and fear of COVID-19, but that they also perceive high levels of social support. --- Methods --- Study design and participants This quantitative study featured a cross-sectional, descriptive, and correlational design. The participants were 365 health-care workers from Amman, Jordan, who completed an online questionnaire distributed through Google Forms between August 16th and August 23rd, 2020 when COVID-19 situation in Jordan began to worsen, with the recording of 20-30 cases per day and this quickly escalated to several hundred and then to several thousand cases a day. However, the number of cases that required hospitalization was low. Individuals were approached for participation through social-media applications, text messaging, and emails. The online Raosoft sample size calculation methodology was used in our study [25]. According to this method a minimum of 378 participants is needed; given that the margin of error alpha = 0.05, the confidence level is = 95%, total population = 21,033 [26], and the response of distribution = 50%. The sample size was also calculated using Krejcie and Morgan method, which provides a similar sample size [27]. Our study was able to recruit a close number of 365 participants. --- Participant recruitment For initial recruitment, the present researchers contacted 24 health-care workers , who were known to the researchers. Through individual phone calls, the researchers informed these coworkers of the purpose and procedure of the study. The researchers then asked the group if they knew of any other health-care workers who met the inclusion criteria , and if they could invite them to participate in this study. An informational document that provided details regarding the survey was distributed to prospective participants. The group forwarded the informational document to other health-care workers through email, text message, or social media. Health-care workers who agreed to participate were contacted by a member of the research team through email or text message. Any questions these prospective participants had regarding the study were answered. A URL linking to the consent form was sent to each individual who agreed to participate, and consent to participate was confirmed through electronic signature . After consent was received, a URL for the Google Forms questionnaire was sent to the participants by text message or email. The researchers emailed the URL to 510 health-care workers, returned 365 responses. --- Inclusion and exclusion criteria Inclusion criteria for participation were: 1) being a health-care worker, 2) residing in Amman/ Jordan, and 3) providing care for patients at the time of the survey. The exclusion criterion was not working the week prior to the data-collection period. --- The e-survey The survey was administered online, and the Checklist for Reporting Results of Internet E-Surveys [28] was used to report the results. The online questionnaire was developed using Google Forms. Google Forms represents a method of quickly gathering participants' responses online. The survey answers were automatically collected in an EXCEL spreadsheet that was imported into SPSS for data analysis. To determine the practicability of the questionnaire, the constituent instruments were pilot-tested beforehand on a group of 30 health-care workers; these individuals were excluded from the main study. --- The research instruments Sociodemographic characteristics and health-related variables. Participants' sociodemographic characteristics, including gender, age, education level, marital status, profession, work type, and clinical experience, were collected. The fear of COVID-19 scale. The participants were asked to report their level of fear regarding the COVID-19 Pandemic. The Fear of COVID-19 Scale is a seven-item scale designed to measure fear of COVID-19 among the general population [29]. Answers are given using a five-point scale . The scores for all seven items are summed to obtain the total score; thus, the range for the total score is 7-35. Higher scores indicate greater fear of COVID-19. The scale has acceptable concurrent validity when compared with the Hospital Anxiety and Depression Scale and the Perceived Vulnerability to Disease Scale; further, the developers determined that the Cronbach's alpha value for the FCV-19S is 0.82, and that its test-retest reliability is 0.72 [29]. For this study, the Cronbach's alpha value was 0.91. Depression, anxiety, stress scale. The Depression, Anxiety, Stress Scale is designed to measure respondents' depression, anxiety, and stress, respectively, over the past seven days [30]. The scale comprises three self-reported subscales, and has a total of 42 items. Each subscale comprises 14 items. Items are rated using a four-point Likert scale ranging from 0 to 3 . The respective scores for depression, anxiety, and stress were calculated by totaling the scores for the respective associated items, and the severity rating index was used to determine the respondent's status in each regard. The severity rating index for each DASS subscale as follow , mild , moderate , severe , and extremely severe . Anxiety scoring comprising normal , mild , moderate , severe , extremely severe 20+. Stress scoring comprising normal , mild , moderate , sever , extremely severe . In the original study, the Cronbach's alpha values for depression, anxiety, and stress were 0.91, 0.84, and 90, respectively [30]. For this study, the Cronbach's alpha values for depression, anxiety, and stress were 0.95, 0.94, and 0.96, respectively. Multidimensional scale of perceived social support. The Multidimensional Scale of Perceived Social Support is designed to determine respondents' perceptions regarding the adequacy of the support they receive from family, friends, and significant others. The MSPSS [31] is a 12-item self-administered scale, and responses are given using a seven-point Likert scale . The scale comprises three subscales: family, friends, and significant others. For each subscale, the mean score is determined by summing the scores for each associated item and dividing the result by 4. The total score is determined by summing the scores for each of the 12 items. For the original study, the Cronbach's alpha values were 0.91, 0.87, and 0.85 for the significant others, family, and friends subscales, respectively. The reliability of the total scale was 0.88. Further, the testretest reliability after 2-3 months was 0.91, 0.85, and 0.75 for the significant others, family, and friends subscales, respectively, and 0.85 for the overall scale [31]. For the present study, the Cronbach's alpha values were 0.89, 0.86, and 0.87 for the significant other, family, and friends subscales, respectively. --- Scale administration The validity of three questionnaires was established using a panel of six experts to ensure the validity of the questionnaires. The validity checked in terms of the survey questions measures what they were intended to measure , the survey contains questions that covered all aspects of the construct being measured , and the extent to which a constructed measure may relate to or predict any outcome for another measure [32]. The six experts are faculty members of PhD holders with a specialty in mental health, medical-surgical, and community. All experts agreed that the questionnaires were valid. The three scales were administered in English. The instruments were pilot-tested on 30 health-care workers who were known to the researchers; these individuals were excluded from the main study. The Cronbach's alpha values obtained through the pilot test were as follows: FCV-19S = 0.86, DASS = 0.90, and MSPSS = 0.84. The test-retest reliability for the same group was as follows: FCV-19S = 0.88, DASS = 0.82, and MSPSS = 0.80. --- Ethical considerations This study was performed in accordance with the Declaration of Helsinki, and approval was obtained from the Human Subjects Review Board of Al Ahliyya Amman University prior to the data collection. Written informed consent was obtained from all participants. The data were stored on a personal computer to which only the main author had access. --- Statistical analyses Data were entered and analyzed using SPSS software . Initially, the data were checked for missing data and outliers. There was no missing data because, on e-survey, we had a star on each question that participants could not move to the next question without answering the previous question. The outliers were screened through visual assessment for scattered plot diagrams, which revealed no outliers. Box Plot and histogram were used to check the normality, as well as the linearity was checked by Pearson correlation, and homogeneity was checked by The Levene's test. Descriptive statistics were used, including frequencies , percentages , means, standard deviations , medians, and interquartile ranges . Variations between sub-categories of demographic variables were checked using chi-square tests. Inferential statistics approaches were used to identify differences in demographic variables; these approaches included independent samples t-tests and variations across demographic sub-groups. Further, Pearson's correlation coefficient was used to determine the relationships between variables and to establish the inter-correlation matrix. To lower the risk of type I errors, the statistical significance level was set at p < 0.05. --- Results The participants' sociodemographic characteristics are presented in Table 1. Participants were distributed over a range of demographic subgroups. Approximately 55% of the participants were women, and most were aged below 50 years and were married . The median family size was three members. Most participants were registered nurses , held 1. Regular communication. 2. Felt that staff well-being was being prioritized . 3. Felt that staff were monitored for symptoms of mental distress, burnout, fatigue, and unrest, and that home-related responsibilities were considered. 4. Felt that staff were kept informed and that efforts were made to raise awareness of the pandemic. 5. The organization established an employee health center. a baccalaureate degree , and provided direct care to patients . Over 65.0% of the participants had over 10 years of clinical experience. Questions regarding the COVID-19 Pandemic revealed that most of the participants had never provided direct care for patients who had tested positive for COVID-19. However, most of the participants reported receiving support from work administrators during the COVID-19 Pandemic, and 58.4% reported high adherence to the stay-at-home regulations. --- Assessment of fear of COVID-19 Table 2 presents the results for the FCV-19S, which reflected the participants' fear of COVID-19. For each item, the mean score exceeded the midpoint of 2.5, indicating a moderate level of fear. The total mean score for the FCV-19S was 23.64, which again exceeded the mid-point for the total score range --- Assessment of depression, anxiety, and stress The mean scores for each subscale of the DASS, are presented in Table 2. Participants displayed extremely severe depression , extremely severe anxiety , and moderate stress . Fig 2 illustrates, for depression, anxiety, and stress, the distribution of the participants across the five levels of severity. Based on the data, approximately 35% of the participants had extremely severe depression, over 40% had moderate to severe depression, and approximately 20% had normal to mild depression . For anxiety, approximately 60% of the participants, reported extremely severe anxiety. Regarding stress, the figure shows an uneven distribution over the severity levels, indicating inconsistent patterns of stress severity. However, approximately 35% was severely distressed. --- Perceived social support The results regarding the social support received by the health-care workers from significant others, family members, and friends, respectively, are presented in Table 2. For significant others, the results indicated that the participants perceived high levels of support from all � For each participant, mean scores for each of these items were calculated. If, when the mean scores for each item were summed, the overall score was below 2.5, the participant was considered to have received insufficient organizational support ; if the score was above 2.5, the participant was considered to have received satisfactory organizational support . IQR: inter-quartile range; PPE: personal protective equipment. https://doi.org/10.1371/journal.pone.0247679.t001 associated sources; the scores for all items exceeded 5 out of 7. These high scores were reflected in the mean score for the subscale , which exceeded the midpoint. Regarding the family subscale, for all associated items the mean scores were above the midpoint of 4, indicating adequate support from family members. The mean score for the subscale was also above the midpoint, indicating high recognition of family support. Similarly, for the friends subscale, for all items the mean scores were above the midpoint, and the mean score for the subscale indicated high recognition of support from friends. The total mean score for the MSPSS was 5.09 out of 7, indicating high perceived social support . Fig 3 shows the distribution of the scores for the three dimensions over three levels of support . The figure shows that all three dimensions are consistently distributed across the three levels. The highest frequency was reported for "high support," followed by "moderate support," and "low support," respectively. This pattern was consistent across all three dimensions . --- Variations across demographic sub-groups The main differences between the demographic sub-groups in terms of the study variables are listed below: • Male participants returned statistically higher scores for fear, depression, anxiety, and stress, respectively, when compared to female participants . However, no statistical difference was found between males and females regarding social support. • Married participants returned significantly higher scores for fear, depression, anxiety, and stress, respectively, when compared to single participants . In addition, married participants demonstrated higher social support when compared to single participants . • Participants aged over 40 years showed statistically higher levels of fear, depression, anxiety, and stress, respectively, when compared with participants aged < 40 years . Moreover, older participants showed higher perceived social support than younger participants . The result of ANOVA revealed that significant relationship between psychological distress and social support and age p � 0.05. • Similarly, participants with more clinical experience showed statistically higher levels of fear, depression, anxiety, and stress, respectively, when compared to participants with clinical experience of less than 20 years . Further, participants with more clinical experience reported more social support when compared to participants with shorter clinical experience . The result of ANOVA revealed that significant relationship between psychological distress and social support and clinical experience, p � 0.05. • Participants who provided care for patients who had tested positive for COVID-19 reported higher levels of fear, depression, anxiety, and stress, respectively, when compared to those who did not provide care for patients who were COVID-19-positive . • Participants who took vacation days during the pandemic reported lower levels of fear, depression, anxiety, and stress, respectively, than did those who did not take any vacation during that period . However, in relation to social support, there was no significant difference between the participants who took vacation days and those who did not take vacation days . • No significant differences were observed between professions regarding any of the study variables . --- Factors influencing social support during the COVID-19 pandemic According to the correlation matrix presented in Table 6, both clinical experience and social support have a weak significant positive correlation with fear, depression, anxiety, and stress, with correlation values being approximately 0.20 and below. However, fear, depression, anxiety, and stress were positively correlated, with correlation values ranging between 0.60 and 0.90; this indicated strong relationships. --- Discussion The findings of the present study provide insights into health-care workers' psychological status during the COVID-19 Pandemic. This study analyzed a mixed group of health-care workers in Jordan five months after COVID-19 was declared a pandemic. Factors associated with health-care workers' psychological distress were determined to include being male, married, aged 40 years and older, having more clinical experience, and working directly with patients who have been diagnosed with COVID-19. Fear, depression, anxiety, and stress were positively correlated. All participants reported psychological distress; however, those who were 40 years of age and older showed a statistically higher level of psychological distress. The health-care workers' concerns were due to several factors. A possible reason for the high level of distress among older workers is that the risk of severe respiratory distress as a result of COVID-19 increases with age, meaning older adults are at higher risk [33]. People at increased risk and those who live with or visit such people need to take precautions to protect themselves from getting COVID-19. Thus, older health-care workers may have reported higher psychological distress because older people can have health issues that make them more prone to complications, and they could also live with young children and/or have older people in their extended family, which could cause them to worry about bringing the virus home to their family members. The current study's findings also indicated that health-care workers who took vacation days reported lower levels of depression, fear, anxiety, and stress, respectively. These results support those of Luceno-Moreno et al. [34], who established that long working hours contribute to psychological problems, and those of Barello et al. [35], who observed work-related psychological pressure, emotional burnout, and somatic symptoms among health-care workers in Italy. The impact of working long shifts, 12 hours and more, on nurses and health-care assistance found 24% of nurses and health-care assistance were more likely to miss days of work due to sickness [36]. Thus, health-care workers are encouraged to take vacations from work for helping health-care workers relax, which contributes to preventing stress. Therefore, during pandemic situations vacations from work are necessary for reducing psychological distress among health-care workers, leading to lower levels of depression, fear, anxiety, and stress. Of course, the effectiveness of this can depend on the local quarantine policy and burden experienced by health-care workers. The study results indicated weak correlations between years of clinical experience and fear, anxiety, and depression, respectively. The challenge that the pandemic brining to health-care workers such increase acuity of care and increased patients' volume and, uncertainty healthcare professional safety, as a result of reusing of personal protective equipment which was not part of health-care professional practice [37]. Health-care workers are hearing about potential surge, which was expected to hit harder, dealing with severe ill patients and death. With experience, health-care workers may adjust to stressful working environment but research, however, stressors may accumulate and cause psychological distress [38]. Pappa et al. [39] conducted a systematic review and meta-analysis of the prevalence of anxiety, depression, and insomnia, respectively, among health-care workers during the COVID-19 Pandemic. Anxiety was assessed across 12 studies, and a prevalence of 23.2% was returned; meanwhile, depression was assessed across 10 studies, and a prevalence of 22.8% was returned. The findings of Pappa et al. [39] support the results of the current study, as they indicate that health-care workers experience anxiety and depression during COVID-19; however, Pappa et al.'s findings also contradict the results of the present research, as the systematic review and meta-analysis showed a higher prevalence of anxiety than depression. Our study found higher depression than anxiety. Finally, Labrague and De Los Santos [40] found that 123 of 325 nurses examined had dysfunctional anxiety levels. Labrague and De Los Santos [40] also indicated that COVID-19 anxiety is associated with social support, organizational support, and personal resilience. These findings support the current study results by showing that front-line nurses are affected by anxiety during the COVID-19 Pandemic. To help health-care workers provide care under extremely difficult clinical circumstances such as COVID-19 pandemic, the emotional and behavioral reactions vary among health-care workers should be acknowledge and empowered through education and training to overcome fear and empathetic distress [37]. The results of our examination of social support during the COVID-19 Pandemic indicated that health-care workers perceive themselves as receiving high levels of social support. Our findings showed that health-care workers gain social support when providing care to patients. The health-care workers we examined perceived high levels of all types of social support, with mean item scores exceeding 5 out of 7 for all. The overall mean score was also above the midpoint. Thus, the results showed that, during the COVID-19 Pandemic, health-care workers gain support when providing care for patients. These results support the findings of a narrative review by Heath et al. [13], which showed that support offered before and during an incident influences whether health-care professionals experience injury or psychological growth. Heath et al. [13] also indicated that clinicians who have healthy, meaningful personal and professional relationships are contented and have a lower risk of burnout. Heath et al. also showed that health-care professionals who have work responsibilities that interfere with their home lives are more likely to experience burnout, leading to stress when providing care to patients [13]. Also, feeling the guilt of transmitting the infection to family members at home, health-care workers experience stigmatization. Self-stigma, mostly, if health-care workers were in direct contact with infected patients, they preferred to stay away from them [41]. Moral injuries have been widely reported among health-care workers on duty during the COVID-19 pandemic [10]. The clinical and ethical challenges that these workers face can foster psychological distress, and health-care workers with poor psychological health affect the quality of care provided at their institutions, as well as their coworkers capability to work [42]. Anticipating the problem may help lessen its impact, and early identification of psychological distress and health-care support is essential. Indeed, COVID-19 infection becomes an occupational injury when health-care workers contract the virus through work or while commuting to work [11]. To support health-care workers during future health emergences and protect them from such injuries, health-care leaders should, in accordance with other regulatory agencies around the world, rapidly implement policy changes at institutional levels and at the local level to facilitate a shift in culture towards improved well-being and workplace environments. Social support is necessary as coping mechanism to decrease health-care workers' psychological distress and promote positive feelings. Spinale et al. [43] reported that social support is correlated with spirituality. Spirituality is associated with transcendental values that are generally influenced by personal experiences and grounded in religious traditions; however, a comparable sentiment can be achieved in a non-religious context. Spirituality can foster positive feelings and promote physical and mental health [43]. People with greater spirituality have also been reported to experience higher levels of well-being [44]. Thus, improving spirituality among health-care workers during pandemics may help them relieve their physical and psychological distress, and also support coworkers, patients, and patients' family members. This is especially important during pandemics, as these are times when spiritual specialists or religious leaders are unable to closely contact patients and health-care workers. In summary, the present findings show that health-care workers feel depressed, anxious, stressed, and fearful of the pandemic. This means that health-care workers are making critical decisions in the course of their work while experiencing notable distress. Direct support from management can help staff develop positive perceptions about work, and can help them manage stress. However, inadequate protection, perceived stigma, and negative feedback from patients can exacerbate COVID-19-related psychological distress [45,46]. Also, health-care workers who perceive high level of psychological distress, need psychological support [47]. Que et al. [45] suggested that psychosocial interventions should be provided in the early stages of pandemics for health-care workers who are at risk of experiencing psychological distress. According to our findings, adequate social support is essential for addressing stress, anxiety, and depression. However, additional research is required to explore the long-term effects of the COVID-19 Pandemic on psychological distress among health-care workers. --- Strengths and limitations The strength of this study is that it measured psychological distress and social support among health-care workers five months after the WHO declared the COVID-19 outbreak a pandemic, and after public services in Jordan were reopened after the lockdown. The study also considered health-care workers' psychological concerns after the pandemic was declared. This is a strength because psychological distress among health-care workers during the pandemic has been somewhat understudied. On the other hand, this study also contains limitations. One of the principal limitations is the cross-sectional nature of the study. Psychological distress was only evaluated cross-sectionally; consequently, we could not obtain information regarding existing causal relationships. Further, the data did not represent the entire population of health-care workers and the services in which they worked , also, did not include other variables such as whether the participants had had any personal experience of loss or illness due to COVID in their family or friends, and, as a result, the findings cannot be used to make useful generalizations regarding health-care workers as a whole, or to determine specific variables' correlations with specific groups of health-care workers. A larger sample of health-care workers recruited from various areas in Jordan is needed to verify the results. Moreover, further research is needed to explore the long-term effects of the COVID-19 Pandemic on health-care workers. --- Implications for health-care workers The results of this study showed that the COVID-19 Pandemic has fostered psychological distress among health-care workers in Jordan, and health-care workers have become acutely conscious of the threat of the virus' spread. Thus, safeguarding the psychological well-being of health-care workers is crucial during pandemic situations. Employers should endeavor to identify approaches that can improve psychological distress among such workers. Most health-care workers have direct contact with patients, and this can cause high levels of anxiety. Managers and leaders should increase the support available for health-care workers in their organizations and in health-care workers' own social networks. Early identification of psychological stress is important. Being male, older, and having more clinical experience increase the risk of stress during pandemics. Thus, during such situations psychological support is essential for this group. However, older health-care workers should also proactively seek psychological support. Further, efforts should be made to develop coworker support; health-care workers could aim to help others implement effective decision-making in response to pandemic-related challenges. --- Practical implications The results of this study suggest that measures should be implemented to protect the mental well-being of health-care workers during the COVID-19 Pandemic. Leaders in health-care facilities should realize the importance of close relationships with health-care workers during the extraordinary times they are facing in this pandemic. In addition to ensuring that the physiological needs of health-care workers, such as availability of PPE and safe working environments, are met, leaders should reassure health-care workers that they and their families will be adequately supported should they become infected with COVID-19. This support should include medical, financial, and psychosocial assistance for both the health-care workers and their families. Moreover, leaders and managers of health-care facilities should make efforts to identify sources of anxiety and fear among health-care workers, and should schedule rigorous assessments by professional psychologists and mental-health professionals. At the primary and secondary levels, regular meetings should be held with health-care workers to promote the development of healthy patterns of coping with the stressors of working with patients with or suspected of having COVID-19. At the secondary level of prevention, individual counseling for mental well-being concerns and early treatment is essential. Teams of professional psychologists should be available at each institution for health-care workers to contact at any time, and prompt treatment should be provided, and as follow-ups. In addition, peer support and group discussions should be encouraged. The major issues for health-care workers are fear, depression, anxiety, and stress. The participants in this study felt that they received high social support, but they also showed higher psychological distress. These characteristics should be considered when developing strategies to address this. It is not clear whether health-care workers physically distance themselves from their families as a result of lockdowns, social-distancing recommendations, and their close contact with patients. If so, social support in the workplace could give health-care workers a sense of being a member of a social network; consequently, health-care workers should be provided with opportunities to establish and strengthen such professional relationships. Further, healing moral distress and occupational injuries are important. This requires collaboration between health-care workers, administrators, and representatives of the community; in particular, an ethically admissible code for pandemic contexts should be established that can strengthen health-care workers' morals. --- Conclusion Our study demonstrated the presence of fear, depression, anxiety, and stress among healthcare workers during the COVID-19 Pandemic. The health-care workers examined considered social support from families and friends to be important during the pandemic, and demonstrated a need for increased social support to adjust to psychological distress. Factors determined to be associated with psychological distress were being male, married, aged 40 years or older, and having more clinical experience. The influence of these factors may be related to the environment in which health-care workers practice. Thus, this study suggests that health-care organizations pay attention to health-care workers' well-being and promote early assessment and identification of psychological distress. It is also necessary to address social support through policy since, as a result of social distancing, there are fewer opportunities for social interaction and to attend events. Social support systems play an important role in protecting health-care workers and reducing the prevalence of psychological distress. --- All relevant data are within the paper and its Supporting Information files. --- Supporting information S1

The emergence of Coronavirus disease 2019 (COVID-19) has affected health-care workers' psychological and mental health. Few studies have been conducted examining the psychological effect of COVID-19 on health-care worker psychological health in Jordan. Therefore, the present study aims to assess the respective levels of fear, anxiety, depression, stress, social support, and the associated factors, experienced by Jordanian health-care workers during the COVID-19 Pandemic. This study adopted a cross-sectional, correlational design to collect data from 365 health-care workers in Amman, Jordan, from August 16th to 23rd, 2020. Along with collecting sociodemographic characteristics, the Fear of COVID-19 Scale, the Depression, Anxiety, Stress Scale, and the Multidimensional Scale of Perceived Social Support electronically administered to participants. The majority of the participants (69.3%) were registered nurses. The mean overall score for the Fear of COVID-19 scale was 23.64 (SD + 6.85) which again exceeded the mid-point for the total score range (21), indicating elevated level fear of the COVID-19 pandemic. Participants had displayed extremely severe depression 40%, extremely severe anxiety 60%, and 35% severely distressed. Scores for depression (21.30 ± 10.86), anxiety (20.37 ± 10.80), stress (23.33 ± 10.87) were also high. Factors determined to be associated with psychological distress were being male, married, aged 40 years and older, and having more clinical experience. Assessment of social support indicated moderate-to-high levels of perceived support for all dimensions (significant other: 5.17 ± 1.28, family: 5.03 ± 1.30, friends: 5.05 ± 1.30). Weak significant correlations were found between social support and the other study variables (r < 0.22), indicating a weak association with fear, depression, anxiety, and stress, respectively. Overall, Jordanian health-care workers sample reported fear, depression, anxiety, and stress. The associated factors were being male, married, aged 40 years and older, and having more clinical experience. Regarding social support, participants primarily relied on support from their families, followed by support from friends.

Introduction Surgical site infections are associated with significant morbidity, mortality, and costs after surgical procedures. 1,2 Complicating up to 5% of surgical procedures nationally, SSIs are common and often preventable. Consequently, reducing SSIs nationally is a priority for patient safety efforts led by The Joint Commission, the Centers for Medicare & Medicaid Services, and consumer organizations such as the Leapfrog Group. One strategy these organizations have used to encourage reductions in SSI is public reporting of hospital performance. For example, Hospital Compare, Medicare's public reporting website, and the Leapfrog Group publish hospitals' SSI rates for consumers to view. Another strategy to reduce SSI incidence is through the use of financial incentives. For example, Medicare's Hospital-Acquired Conditions Reduction Program , created under the 2010 Patient Protection and Affordable Care Act, is a pay-for-performance program focused on infections and other adverse safety events. 3 Under this program, 85% of hospitals' performance scores are determined by infection metrics from the Centers for Disease Control and Prevention's National Healthcare Safety Network , including complex SSI after colectomy and abdominal hysterectomy. Under the HACRP, hospitals in the worst-performing quartile are penalized 1% of their total Medicare payments, amounting to more than 1 billion dollars in its first 4 years. 3 Consequently, the risk-adjustment methods used for SSI have significant consequences, both in terms of public reputation and financial stability for hospitals. Prior work 4,5 suggests that teaching hospitals and safety-net hospitals are disproportionately penalized under the HACRP, leading to concerns that performance scores may be driven by patient characteristics outside hospitals' control, rather than the quality of care that hospitals deliver. Although the NHSN has recently incorporated diabetes, sex, age, and obesity into their SSI risk adjustment models, no adjustment is made for social risk factors, despite prior studies [6][7][8] showing that social risk factors are associated with an increased risk of infection for other surgical procedures. Understanding whether social risk factors are associated with SSI for colectomy and abdominal hysterectomy, and whether accounting for these factors would meaningfully change hospitals' performance on these measures, has implications for public reporting and value-based payment models, but has not previously been described, to our knowledge. Therefore, in this study, we aimed to determine whether SSI rates after colectomy or abdominal hysterectomy differ by patient race/ ethnicity, neighborhood income, or insurance type. We also modeled the potential outcomes of accounting for social risk factors on relative performance for safety-net hospitals and teaching hospitals compared with their peers. --- Methods --- Data Patients undergoing colectomy or abdominal hysterectomy during the years 2013 and 2014 were identified using data from the State Inpatient Databases 9 for Arizona, Florida, Iowa, Massachusetts, Maryland, New York, and Vermont. These states were selected because they allow linkage of individual patients across time, allowing for the ascertainment of postoperative events that happen either during the index hospitalization or after discharge. We identified procedures using International Classification of Diseases, Ninth Revision, Clinical Modification procedure --- JAMA Network Open | Surgery --- Social Risk Factors Our primary risk factors were 3 sets of social risk variables: race/ethnicity , insurance status , and median income for patient zip code in quartiles, all as defined in the Healthcare Cost and Utilization Project data. 9 Individuals missing data for key risk factors were excluded. Additional variables used for risk adjustment included age, sex , obesity, diabetes, and whether the hospital was designated as an oncology specialty hospital, as established by the NHSN for risk adjustment of complex 30-day SSIs for colon and abdominal hysterectomy procedures. 12 Obesity and diabetes were defined as per the Elixhauser Comorbidity Index. 13 We also adjusted for overall severity of illness by estimating each patient's risk of an American Society of Anesthesiologists score of 3 or greater, on the basis of the probability derived from a logistic regression model developed in a population undergoing colectomy . Oncology specialty hospitals were identified using the American Hospital Association Annual Survey. 11 --- Outcomes Our primary outcome was the occurrence of a complex SSI. We defined complex SSI as an infection coded during an inpatient hospitalization or requiring operative treatment in an ambulatory surgery facility within 30 days of the index procedure. All other SSIs were categorized as noncomplex and were not outcomes for the purpose of this study. Complex SSIs were identified using ICD-9-CM diagnosis and procedure codes within 30 days postoperatively, as described elsewhere. 14,15 Briefly, SSIs recorded during the first 30 days after surgical procedures were identified using ICD-9-CM diagnosis codes from encounters in the State Inpatient Databases and State Ambulatory Surgery Databases. 16 Prior work 17,18 in this area has suggested that use of these codes has sensitivity similar to that of routine clinical surveillance for detecting complex SSI, but a lower positive predictive value. We censored the observation period to avoid misclassification of SSIs after a subsequent surgery using the NHSN procedure list, as described elsewhere. 15 --- Statistical Analysis Patient characteristics were described using standard statistical tests as appropriate, and unadjusted complex SSI rates were calculated. We then ran generalized linear models including NHSN risk adjustment elements , 12 accounting for clustering by hospital, to examine associations between key risk factors and complex SSIs. We did this in a stepwise manner, such that models first only included each social risk factor alone and subsequently included the NHSN risk adjustment elements and the remainder of the social risk factors. To determine whether adjusting for social risk factors would affect the relative performance of safety-net or teaching hospitals on these measures of complex SSI, we ranked all hospitals performing at least 20 colectomies on their complex SSI rates and assigned percentiles to each. We defined safety-net hospitals as the quintile of hospitals with the greatest proportion of patients with Medicaid health insurance and teaching hospitals as those reporting a medical school affiliation in the American Hospital Association Annual Survey. 11 We ranked all hospitals in order of performance on each complex SSI type and calculated the mean percentile score for safety-net and non-safety-net hospitals, as well as teaching and nonteaching hospitals, using no risk adjustment , using current NHSN risk adjustment, and after adding social risk factors to the models. Because higher infection rates are worse, a hospital with a percentile score of 0 would be the best performer and a hospital with a percentile score of 100 would be the worst performer. We also examined the proportion that would be in the worst quartile of performance under each risk adjustment scenario, because the current HACRP program assigns penalties to hospitals in the worst quartile of performance on a broader set of infection measures. --- Results --- Patient Population In total, 149 741 patients met the inclusion criteria, including 90 210 patients undergoing colectomies and 59 531 patients undergoing abdominal hysterectomies . The colectomy cohort had a or unknown insurance or were uninsured; 24% were from the lowest quartile of median zip code income. Within 30 days of surgery, the complex SSI rate was 2.55%. The hysterectomy cohort was younger, with a mean age of 49.8 years, and all were female. Fourteen percent had an estimated ASA score of 3 or higher, and 18% were coded for obesity. Fifty-two percent of the sample were white, 26% were black, 14% were Hispanic, and 8% were other or unknown race/ethnicity. The predominant insurer was private insurance , 16% had Medicare, 19% had Medicaid, and 3% had other or unknown insurance or were uninsured; 27% were from the lowest-income zip codes. Within 30 days of surgery, the complex SSI rate was 0.61%. --- Association Between Social Risk Factors and Rates and Odds of Complex SSI For the colectomy cohort, in unadjusted analyses, black race was associated with lower raw rates and lower --- Outcomes of Adjusting for Social Risk Factors on Safety-Net and Teaching Hospitals' Performance on SSI For colectomy, safety-net hospitals had similar unadjusted complex SSI rates compared with non-safety-net hospitals . The mean percentile score was 46.1 among safety-net hospitals and 50.2 for non-safety-net hospitals , although these were not statistically different from one another . Similarly, 23.9% of safety-net hospitals were in the worst quintile of infection rates compared with 24.2% of non-safety-net hospitals . Adding current NHSN adjustment or adding social risk adjustment did not change these patterns Comparing teaching with nonteaching hospitals for hysterectomy, complex SSI rates were similar , but teaching hospitals had higher percentile ranks (mean [SD] performance percentile, 55.0 b To estimate American Society of Anesthesiologists score, logistic regression was used to identify people more likely to have a score of 3 or higher. c Elixhauser Comorbidity Index definitions were used for comorbidities. ( [ --- Discussion We found that social risk factors were inconsistently associated with complex SSI rates after colectomy or hysterectomy. For colectomy, Medicaid status and living in a low-income zip code were associated with higher complex SSI rates. For hysterectomy, no social risk factors that we examined in this study had statistically significant associations with SSI. Safety-net hospitals performed similarly to non-safety-net hospitals on measures of complex SSI for both colectomy and hysterectomy, and teaching hospitals performed similarly to nonteaching hospitals. Adding social risk factors to current risk adjustment methods for complex SSI did not change Adjusted odds ratios and 95% CIs for SSI after colectomy and hysterectomy . safety-net hospitals' relative performance under the program, although it widened the gap between teaching and nonteaching hospitals. Patients with either individual or community markers for poverty had higher complex SSI rates for colectomy. Although prior studies 19,20 have demonstrated that several clinical factors are associated with SSI rates in this procedure, including more-complex operations and more-complex patients, open vs laparoscopic procedures, diabetes, and obesity, as well as tobacco and alcohol abuse, there are no prior studies, to our knowledge, examining social risk factors in the context of colectomy. Our findings should be considered exploratory, but if they are confirmed, they may suggest that patients from disadvantaged backgrounds could benefit from additional targeted preoperative care and postoperative monitoring to reduce the risk of complex SSI. Interventions such as surgical bundles, optimization of diabetes control, and close postoperative surveillance could help reduce these disparities. In addition, public reporting and pay-for-performance programs that measure hospital performance according to hospital SSI rate could consider evaluating whether the addition of social risk to risk adjustment models could improve their accuracy. Because patient and neighborhood income levels are outside the control of hospitals, such adjustment could yield fairer performance comparisons, while encouraging all hospitals to focus infection prevention efforts on their highest-risk populations. We did not find consistent associations between social risk and complex SSI for hysterectomy. Similar to colectomy, prior studies 21,22 have identified complex surgery, open vs laparoscopic approach, diabetes, and obesity as risk factors for SSI for this procedure. Rates of complex SSI after hysterectomy are very low-0.61%, or roughly 1 in 200 operations, in our sample-which may make identification of differences associated with any particular traits more difficult. This may also speak to the difficulty in using such rare events as quality measures, because a single infection could change hospitals' measured performance from exemplary to poor. The use of multiple types of infections in NHSN and other programs may mitigate this problem to some degree, however. Some differences in SSI rates by social risk were reduced in magnitude after risk adjustment, suggesting that there were medically mediated associations between social risk and SSI. Although Medicare insurance is not considered a social risk factor , it is a marker of age and medical complexity and was consistently associated with higher SSI risk; this raises the possibility of unmeasured confounding, because the NHSN risk adjustment is minimal compared with other clinical outcome measures used in the Medicare program. Further research should examine whether more-robust risk adjustment might explain some of the risk we found to be associated with local or community poverty and whether better risk adjustment could improve these measures' fairness in measuring hospital performance. --- Limitations This study has limitations. We relied on billing data for event ascertainment and for risk adjustment and had data for a limited number of states. The NHSN SSI data are drawn from a different set of hospitals nationally, and NHSN surveillance is based on clinical data sourced from primary records and active hospital-level surveillance rather than billing data; therefore, although we are examining the same clinical event, our rates and patterns may not be identical to theirs. Our risk adjustment may also be subject to similar limitations in coding sensitivity for comorbidities as our outcome measure. Event rates, especially for hysterectomy, were very low, which may suggest limited sensitivity of our analysis for SSI events and likely limited our power in terms of identifying significant associations between social risk factors and our outcome. Our use of an inpatient and ambulatory surgical database to identify events may also contribute to the low observed incidence, because although we expect most complex SSI events included in the outcome to be managed in this setting, some may be managed exclusively in the outpatient setting and, thus, would be excluded from our analysis. This could introduce bias if lower-income populations were more likely to be managed in the outpatient setting, although we are aware of no evidence that this is the case. We simulated percentile ranking to illustrate the outcomes of additional risk adjustment, but the HACRP and other patient safety programs also include other safety measures, so our findings do not directly reflect performance or payment under these programs more broadly. Finally, we had data on only a limited set of social determinants of health, and our negative findings particularly for hysterectomy should not be taken to suggest that no social risk factors are associated with postsurgical outcomes; further research could address additional social risk factors such as education or employment, as well as associated ones such as health literacy. --- Conclusions We found inconsistent associations between social risk and complex SSI. For colectomy, we identified specific patient populations, such as individuals with Medicaid coverage and those in low-income areas, that could be targeted to reduce infection rates. For colectomy, consideration could also be given to adjusting SSI rates for social risk factors in public reporting programs or pay-for-performance programs like the HACRP. The outcomes of patient safety-focused pay-for-performance programs on health care systems serving socially at-risk populations should be closely monitored to identify disparities and opportunities for improvement. --- Dr Joynt Maddox had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

IMPORTANCE Surgical site infection (SSI) is an important patient safety outcome. Although social risk factors have been linked to many adverse health outcomes, it is unknown whether such factors are associated with higher rates of SSI. OBJECTIVES To determine whether social risk factors, including race/ethnicity, insurance status, and neighborhood income, are associated with higher rates of SSI after colectomy or abdominal hysterectomy, 2 surgical procedures for which SSI rates are publicly reported and included in pay-forperformance programs by Medicare and other groups.This cross-sectional study analyzed adults undergoing colectomy or abdominal hysterectomy, as captured in State Inpatient Databases for Arizona, Florida, Iowa, Massachusetts, Maryland, New York, and Vermont. Operations were performed in 2013 through 2014 at general acute care hospitals in the United States. Data analysis was conducted from October 2018 through June 2019. EXPOSURES Colectomy or hysterectomy. MAIN OUTCOMES AND MEASURES Postoperative complex SSI rates. RESULTS A total of 149 741 patients met the inclusion criteria, including 90 210 patients undergoing colectomies (mean [SD] age, 63.4 [15.6] years; 49 029 [54%] female; 74% white, 11% black, 9% Hispanic, and 5% other or unknown race/ethnicity) and 59 531 patients undergoing abdominal hysterectomies (mean [SD] age, 49.8 [11.8] years; 100% female; 52% white, 26% black, 14% Hispanic, and 8% other or unknown race/ethnicity). In the colectomy cohort, 34% had private insurance, 52% had Medicare, 9% had Medicaid, and 5% had other or unknown insurance or were uninsured; 24% were from the lowest quartile of median zip code income. In the hysterectomy cohort, 57% had private insurance, 16% had Medicare, 19% had Medicaid, and 3% had other or unknown insurance or were uninsured; 27% were from the lowest-income zip codes. Within 30 days of surgery, SSI rates were 2.55% for the colectomy cohort and 0.61% for the hysterectomy cohort. For colectomy, black race (adjusted odds ratio [AOR], 0.71; 95% CI, 0.61-0.82) was associated with lower odds of SSI, whereas Medicare (AOR, 1.25; 95% CI, 1.10-1.41), Medicaid (AOR, 1.23; 95% CI, 1.06-1.44), and low neighborhood income (AOR, 1.14; 95% CI, 1.01-1.29) were associated with higher odds of SSI. For hysterectomy, no social risk factors that were examined in this study had statistically significant associations with SSI after adjustment for clinical risk.were found. For colectomy, infection prevention programs targeting low-income groups may be (continued) Key Points Question Are social risk factors, including race/ethnicity, insurance status, and neighborhood income, associated with higher rates of surgical site infection (SSI) after colectomy or abdominal hysterectomy, 2 surgical procedures for which SSI rates are publicly reported and included in pay-for-performance programs nationally? Findings In this cross-sectional study of 149 741 participants, Medicaid insurance status (a marker for poverty) and living in a low-income zip code were associated with higher SSI rates after colectomy, even after adjusting for clinical risk. For hysterectomy, no social risk factors that were examined in this study had statistically significant associations with SSI after adjustment for clinical risk. Meaning For colectomy, infection prevention programs targeting low-income groups may be important for reducing disparities, and policy makers could consider taking social risk into account when evaluating hospital performance.

Introduction The New Zealand early childhood curriculum, Te Whāriki , was considered one of the first developed in the world-holistic in nature and with bi-cultural foundations . According to Te Whāriki, inclusion is defined as encompassing "gender and ethnicity, diversity of ability and learning needs, family structure and values, socio-economic status and religion" . However, despite the fact that Te Whāriki provides a framework for the early childhood education sector to consider effective inclusion, questions remain about how, at the very practical level, ECE providers are able to work in the interests of equity, social justice and inclusion of young children with disabilities, especially those from low income families. Furthermore, despite policy in support of inclusive education, multiple studies showed that many parents of young children with special needs have difficulties accessing intervention services and may be less satisfied when they do receive services . In addition, research and anecdotal evidence also indicated that some ECE providers are less than welcoming of children with disabilities and actively discourage families from enrolling in their programs . It is not uncommon for poor children with special needs to be excluded or denied the right to participate in ECE settings of their choice, as these children are also largely invisible in discussions about child poverty . --- The Impact of Poverty on Early Childhood At present, poverty is a chronic adversity that affects approximately 14.9 percent of children living in New Zealand . Poverty is often associated with specific events such as losing a job or having a child with disabilities, which create difficulties in families meeting basic necessities such as food, housing and health care, and may adversely affect parenting . The literature also reveals that the timing of poverty is important, with low income during the early childhood and early school years having the most negative effects on school achievement . According to Wordsworth and colleagues , in general, young children experience poverty in three dimensions: deprivation , exclusion , and vulnerability . While childhood poverty is not deterministically associated with poor outcomes, it does make it more difficult for some, especially those with disabilities, to transcend adverse life circumstances . Indeed, the combination of economic hardship and disabilities can become a powerful force that exacerbate crises for young children. The implications of these findings are that intervention during the early years of children's lives to alleviate poverty is most likely to support child and family wellbeing. --- Parents' Experiences of Early Childhood Inclusive Education According to Bronfenbrenner's ecological theoretical perspectives, the immediate setting of the family is the key context for human development. Research has addressed the issues of how some families are able to maintain nurturing and supportive family environments for children in the face of challenge. For example, New Zealand researcher Kalil's review of the empirical studies on family resilience reveals that family connectedness, religious affiliation, family belief systems and parenting beliefs are factors relating to resilience. The organization of the family, the flexibility of members' roles, activities, and relationships in which the family engages can all be protective factors supporting children's development . In summary, family processes mediate between economic and other stress and risk factors such as disabilities and can have great impact on child outcomes. While the New Zealand ECE curriculum has focus on the importance of working in partnership with families, very little research has been directed toward the aspect of building family resilience. Parents' voices and perspectives on early childhood inclusion has not attracted much attention from researchers and ECE service providers in New Zealand . In addition, a rapid increase in the size and share of young-child population with disabilities pose challenges to ECE programs and policymakers in the early childhood arena. These historic demographic changes are converging with efforts of the ECE sector to scale early childhood services and improve their quality. With the increase in the number of young children having special needs living in a low-income family , efforts to understand parents' perspectives of early childhood inclusive education and establish high quality preschool education that support both child and family wellbeing are an urgent priority. Against this backdrop, this study was conducted to explore low-income families' experiences of early childhood inclusive education in New Zealand. This paper draws on findings from a qualitative study on 30 low-income parents' views about early childhood inclusion in New Zealand. It is believed that by investigating parents' views on inclusive education, ECE services in New Zealand can provide better support for families and children from low-income backgrounds. These parents' views, expectations, and needs can help ECE providers develop effective strategies and resources to address these needs. At this point it may be helpful to make a digression to briefly define some key terms used in the paper. The term young children refers to children aged 9 and under. We focused on families with young children due to situating our larger study in the context of early childhood care and education. Special needs refers to a specific category of exceptionality such as learning disabilities, intellectual disabilities, or emotional/behavioral disorders requiring some modifications of standard curriculum, methods and/or equipment, as well as the emotional and social environment of the classroom and school, to obtain optimal benefit from schooling . For the purpose of this study, the terms special needs and disabilities are used interchangeably. The following research questions framed the study: What inclusive practices are put in place in the ECE sector that have a strong focus on supporting families of low socioeconomic status? What are low-income families' perspectives on early childhood inclusion in New Zealand? What are these families' coping strategies and support needs? --- Methodology According to social constructivism, our knowledge of reality, including the domain of human action, is a social construction by human actors and that this applies equally to researchers . In addition, as the social world influences how each individual constructs the truth, understanding social process involves getting inside the world of those generating it . From this perspective, the qualitative study, which is informed by social constructivism, is appropriate for studying context-specific, unique processes . As this research sought to investigate parents' experiences and views on early childhood education in New Zealand, it seemed that by constructing a qualitative study and involving people in specific settings, different perspectives and experiences would emerge. This study is based on 30 semi-structured interviews with low-income families in the north island of New Zealand. Parents were eligible for the study if they were receiving public assistance such as welfare or were Community Service Card holders , and if they had at least one child with special needs aged 9 or under. Families who met the criteria were asked to volunteer to participate. We focused on low income families of young children due to situating our larger study in the context of early childhood care and education. --- --- Instrument The semi-structured interview protocol was developed through multiple discussions with the research team about the items and structure. All questions were open-ended and were first piloted with two parents whose interviews were not included in the analysis. --- Procedure Ethical approval was granted by the relevant university Ethics Committee prior to commencing the participant recruitment process. A verbal and written notification regarding the interviews, analysis and possible uses of the data, and the consent form assuring privacy and anonymity was given to participants prior to the interviews. All interviews were conducted face to face and on average lasted 35 minutes. --- Data Analysis Each interview was audio-recorded and subsequently fully transcribed. The interpretation of the interview data was facilitated by group discussions within the research team, and a credibility check of themes was conducted by the principal investigator to ensure that the emerging themes are internally coherent and grounded in the transcripts . Saldaña's streamlined code-to-theory model was used for the data analysis. In the open-coding process, the transcripts were first scanned and broken into various sections, categorized, and labelled. An initial list of themes relating to the families' experiences was drawn up by grouping comments that clustered together, and a list of relevant verbatim quotes corresponding to these themes was created. At the second stage of the data analysis, which is also known as the axial coding process, the categories under similar labels were recombined and compared, which allowed for further generation of concepts and identification of relationships among components. Finally, the processes of open and axial coding were merged and a master list of the main themes emerged. Data was categorized into three conceptual domains including inclusive practices for families of low socio-economic status, resources and strategies that strengthen family resilience, and low-income families' perspectives on early childhood inclusion in New Zealand. To encapsulate participants' experiences, a list of relevant verbatim quotes corresponding to these themes was also created. These statements, which include rich descriptions of the main themes that emerged, are quoted as supportive evidence of the participants' meaning-making experience. --- Findings The following are the three main themes that emerged from the data accrued from the interviews: inclusive practices for families of low socio-economic status, low-income families' perspectives of early childhood inclusion, and parents' coping strategies and support needs. --- Inclusive Practices for Families of Low Socio-Economic Status Participants were asked what resources and strategies were provided by the EC sector to meet their children's special needs. Responses from the participants showed that in general, the support their children received was mainly from ECE teachers, educational psychologists, and early intervention specialists such as speech therapists. --- EARLY CHILDHOOD INCLUSIVE EDUCATION 10 More specifically, the majority of parents found that assigning a classroom peer buddy or an adult aide to their children helped in providing time for one-onone instruction. Unfortunately, more than half of the parents reported that there were no adult aides assigned to their children's classrooms, making individualizing very difficult. Thirteen parents stated that some of their children's ECE teachers had backgrounds in special education and the knowledge they shared with them was very helpful. Five parents mentioned that their children were provided resources such as sensory toys to use as part of the action plans made by the itinerant early intervention teachers. These parents also expressed appreciation for how some of the itinerant early intervention teachers made an effort to support staff of early childhood classes which their children attended to provide personalized learning and support. Two parents commented that they were given support by the psychologists, and with the behavior interventions their children had shown significant improvement. Approximately half of the parents also mentioned that teachers' home-visits made EC inclusion more effective as the connection those teachers made helped build a sense of trust and understanding for both the children and parents. Furthermore, according to some parents , the EC professionals who worked as a team with their families were the effective ones. It was evident that when the partnership between home and school was strengthened, children's educational experiences were also improved. These findings are consistent with other studies emphasizing the importance of the home-school relationships . Some parents commented that in the past they did not know much about parenting a young child with challenging behaviors, teachers who helped them set reasonable boundaries for children's behavior were highly appreciated. Two parents also reported that they were thankful to the teachers who helped them build a repertoire of effective parenting strategies, as these strategies had helped improve parentchild interaction as well as child behaviors at home. The subtle effect of well-meaning but misjudged attempts to promote inclusion was also mentioned by a number of parents , suggesting how at times those, including ECE professionals who do not know much about disabilities can make their families feel uncomfortable. So what are parents' perspectives on early childhood inclusion? This topic will be discussed next. --- Low-Income Families' Perspectives on Early Childhood Inclusion in New Zealand Respondents were asked to rate their level of impression with New Zealand's early childhood inclusion and satisfaction with support received on a 5-point scale, with 5 being the best rating and 1 being the worst. When asked about their level of impression with different aspects of New Zealand's early childhood inclusive education , participants gave the highest ratings to curriculum design , accessibilities of center facilities , and inclusiveness of center activities . More specifically, about 80.3% of the participants reported that in terms of time and structure, their ECE programs were relatively flexible, and in general children with special needs were allowed to play and to learn at their own pace. --- EARLY CHILDHOOD INCLUSIVE EDUCATION 12 However, when parents were asked how they felt about the "effectiveness" of inclusive education at their centers, the results were less encouraging. Out of the 30 responses, only four parents mentioned that their children's special needs were met at the ECE programs. About 56.7% mentioned the ECE teachers did not seem to have sufficient knowledge and skills needed to teach children with special needs. More specifically, these parents reported that teachers did not purposefully involve their children in everyday activities nor were they able to deliver any disability-specific interventions at the classroom level. These parents commented that teachers' lack of consideration for individualized instruction, modifications, and accommodations make inclusion less effective. Finally, 15 families reported that there was a lack of opportunities to engage in their children's education. According to these parents, the EC centers did not make any individualized family service plan for their children, nor were they asked for input for their children's education plan. As low-income families, they often lacked access to the private intervention services and resources. One mother further commented: The ECE program does organize activities to engage families during school holidays and weekends. But most of the time we could not afford to make those field trips with others, for example, an outing to the zoo or the Butterfly Creek will cost our family about 70 to 100 dollars, which is way beyond what we would pay for a one-day trip. But the ECE center did not really understand that it is for financial reasons we missed out those trips, and they did not seem to bother. These parents also indicated that they would like to learn more about applied behavior analysis and to try to use these strategies at home. Another two parents mentioned that the speech therapy their children received was helpful, but the duration and frequency of the sessions were not enough . However, eleven parents felt that they were not supported by early intervention teachers and administrators at their programs. On a positive note, some parents said that their children's kindergartens were faith-based and staff there were able to provide spiritual support which were helpful. These parents also elaborated that because they were from the same faith traditions, they were able to look at their situations from similar perspectives. The majority of the parents agreed that inclusion should begin with recognizing that children with disabilities have the right to access high quality early childhood services and education, just like their typically developing peers do. Along the same line of thought, some participants shared that this right should not be affected by disability, and that ECE programs should do all they can to find solutions for helping children with additional needs. A couple of parents said that the EC teachers who respected the worth and dignity of every child were also those who demonstrated warmth and affection to their children. --- One participant reported: The ECE program my son attends is inadequately resourced to cope with young children with disabilities. Though the center has educational psychologists and early intervention teachers who visit on a regular basis, the support available is far from enough. Obviously, there was a dichotomy of views about the "effectiveness" of early childhood inclusive education. It also appeared that one of the reasons why parents were not generally confident in the teachers' skills and knowledge was ECE teachers' lack of training in special needs education. This raises the question of what might be a way forward, and what coping strategies are effective in helping children with disabilities reach their potential, enabling them and their families to live in some measure of financial security? This leads to the next theme: parents' coping strategies and support needs. --- Parents' Coping Strategies and Support Needs Reports from the parents showed that their children with disabilities faced a plethora of challenges that require adaptation and resilience for them to develop typically. Economic hardship also posed great challenges for them. The perspectives of the participants on the stresses families experienced and the buffers which enabled them to cope with these are summarized in the sections below. Coping strategies. Participants were asked to identify strategies, resources, and inclusive practices that helped them cope with disabilities of their children and financial stresses. Interestingly, according to the majority of the participants , maintaining a positive outlook is one the most important strategies parents used. Results from the interview data also indicated that among the 21 participants who were affiliated with a religion/faith, 16 of them said that their religion/faith enabled them to have a positive outlook on negative experiences and to display resilient attitudes. This resonated with the literature which suggested that when coping methods such as religions, are used to positively reframe a child's disability or financial strain, they often lead to positive adaptations and enhance one's general health and wellbeing; whereas when self-destructive behaviors, such as blame and punishment for past sins, are used, the coping became maladaptive . These findings also echoed what the literature suggests about the impact of family belief systems on family resilience . A father talked about his experience: We do believe faith plays a crucial role in our daily lives and many aspects of it, life would be harder without faith. Our family has been affected by our son's autism in many ways, and the issues of unemployment and poverty only made things worse. But we were able to become more resilient as a family because of our faith…we developed the ability to develop and grow strengths, and meet life's challenges as a team. We are thankful that our God has enabled us to work through all these difficulties in a positive way, and emerge stronger in the process. One mother shared that faith contributed meaning in hardship: When our daughter who has a handicap was born into our home, we had an opportunity to turn to the Lord in a fresh way. After all these years, both my husband and myself have come to realize that there is value in service to people with disabilities, and this value is primarily found in being drawn closer to our Lord who set the example for us. --- Another parent commented: As God has loved us unconditionally, we learn to love our child unconditionally, regardless of his weaknesses and strengths. In the process of raising our son, we have learned to be patient, grateful, faithful, and kind. It is obvious that according to these families, faith encouraged transcendence and spirituality , and made meaning of adversity and aided in surviving stressful situations. In a similar vein, one father shared that faith affirmed strengths and possibilities : Parenting the child with disabilities teaches us lessons that could never be learned in any other way, and God has blessed us in the process. We know that though God allowed the handicap to occur, He would in some way work it out to the good to all involved. According to some parents , the use of positive reappraisal, seeking social and professional support as well as support from the wider extended families are effective problem-solving strategies that have helped to mediate stress. These parents shared that support from extended families helped them respond to and cope with stress effectively. For example, some grandparents and relatives becoming involved in the family in areas related to school, and spiritual or recreational activities fulfilled a number of supportive functions. In particular, they provided support such as transportation, grocery shopping, babysitting, emotional support, household responsibilities , as well as financial support. A mother gratefully expressed her feelings: The early years in my son's life was a particularly difficult time for our family, as we adjust to the "diagnosis" and all the changes this means to family life. What has been helpful was to talk with our extended family members and friends in the faith community. It's been heartening to see virtually all of our friends and extended family members express overwhelming support for our family. Some participants also reported that effective problem-solving skills, and a sense of mastery and self-confidence enable them to persevere in the face of pressure. Another parent talked about her experience: In our minds, we are just living out our dream. But then our daughter came along, and she was born with Down Syndrome. Both my partner and myself believe that the differences are the very thing that makes ours richer and fuller. It pushes you to think in a new way about how you think, speak, act, and live. Two parents shared that when EC teachers work reciprocally with families, as parents they were able to develop a more complete picture of their children's development. These parents also mentioned that opportunities the EC programs provided to support the parent-child interactions and family bonds were helpful. Support needs. During the interviews, parents identified a number of areas where support is needed, namely the lack of resources and services, economic adversity, and social isolation . There was strong evidence that parents were concerned about the lack of support from both ECE service providers and society. A father of a son with Autism Spectrum Disorder reported: My son received early intervention service once per three months, and behavioral interventions once every two months. We really hope to receive more support from both the Ministry of Education and the ECE program. About 63.3% of the families indicated that there was a lack of early intervention services and health care available. A high percentage of parents were also disappointed that the majority of the ECE teachers they worked with did not know much about special education or disability-specific interventions, and were therefore not able to provide the support their children needed. These difficulties were exacerbated when families also had to battle financial challenges, and this situation caused an increase in distress levels and limited their ability to cope effectively. A great number of parents also expressed being financially stressed when having to pay for specialized equipment for their children that insurance did not cover or when they sought private therapies and alternative interventions. For these parents to appropriately provide for the special needs of their children, they had to pay out of pocket, and these out-of-pocket expenses consumed a large percentage of their family income. Another mother said: The financial burden of caring for a child with a severe disability often leads to high level of stress. It is so difficult to balance the needs of our child who has a disability with the family's needs…The job market in New Zealand has never been promising, and I am not sure when I can get a job and some stable income and provide my family some financial security. Several parents indicated that others' opinions on individuals with special needs also caused a certain degree of social isolation and stress, and they needed social support in order to deal with spectators who looked down on their families for having children with disabilities. The following comments are indicative of the struggles some parents had: We knew our child would be born with Down Syndrome, as the disability was detected in the womb, but we still wanted to keep her. All our children's lives are valued, regardless of their abilities . We believe that our value as human beings does not come from being physically or mentally perfect. Although we are aware that not all people are in favor of pro-life choices, we often have to deal with spectators who don't agree with our life choices when we take our daughter out, which can be stressful . One participant further commented that he understood that even in a developed country such as New Zealand, persons with disabilities are at a disadvantage in social situations, educational attainment, and labor market outcomes, and he was worried about the unmet needs of his son as well as his future. --- Discussions By engaging with issues of low income, disabilities, faith, and inclusive practices, the study aimed to provoke discussion and localized collective action in support of early childhood inclusion. To this end, this study has explored low income families' experiences of early childhood inclusion, in an effort to understand their perceived effectiveness of inclusive practices in New Zealand as well as these families' coping strategies and support needs. --- Limitations As with other studies, this study acknowledges some limitations, including the fact that there were only 30 families involved in the research, and the effects of socioeconomic conditions, religious differences, and cultures were not analyzed in this study. A larger study should be conducted in the future to investigate the specifics of early childhood inclusion in New Zealand. For example, further research into specific early childhood programs and home interventions can build on the factors discovered through this study. --- Conclusions In general, the results from the current study concur with previously published research reports on early childhood inclusion: children with special needs often require a well-coordinated system of care, including individualized educational services, interventions,, specialized equipment, support groups, and other resources . In addition, findings from this study underscore the urgent need to address barriers facing low-income families and parents with young children with special needs. Consistent with previous findings regarding initial teacher education and professional and development in New Zealand, this study also revealed that the lack of special education training for early childhood teachers, along with insufficient specialist and EARLY CHILDHOOD INCLUSIVE EDUCATION 20 paraprofessional support, are essential factors that mitigate against early childhood inclusion in New Zealand . What is important about the new results is that though the majority of the families participating in this study appreciated the flexible time and structures of the ECE programs their children attended, parents were concerned about the lack of intervention services for their children. In addition, since families of lower socioeconomic status often had limited access to appropriate health care and educational services. This situation can cause an increase in distress levels and limit their ability to cope with disabilities and financial adversity effectively. Lastly, findings of the present study highlight the importance of the use of positive coping methods, such as maintaining a positive outlook and seeking social support, and the role faith plays in family life because of the negative effect of poverty and disabilities on child and family wellbeing. In the case of the majority of the participants in the study, such coping strategies led to positive adaptations and enhanced the general wellbeing of the families and their children with disabilities. Faiths, along with other factors, have played a significant role in contributing to inclusive education services. --- Availability of data and material Yes, the original data is available. Code availability Not applicable --- --- Not applicable --- Conflicts of interest/Competing interests No conflicts of interests Authors' contributions All authors whose names appear on the submission 1) made substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data; or the creation of new software used in the work; 2) drafted the work or revised it critically for important intellectual content; 3) approved the version to be published; and 4) agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

New Zealand has a reputation for having one of the most inclusive education systems in the world. However, research and anecdotal evidence show that many parents of young children with disabilities have difficulties accessing intervention and health care services and may be less satisfied when they do receive services. In addition, though a plethora of research has been done on inclusive education, little attention has been given by researchers to lowincome parents' perspectives on early childhood inclusion in New Zealand. This paper draws on findings from a qualitative study on 30 parents' experiences of early childhood inclusive education in New Zealand. Parents participating in this study came from different religious backgrounds, represented diverse ethnicities, all had at least one child who had a diagnosis of disabilities and/or chronical illness, and met the low-income criteria of New Zealand. Results showed that though the majority of the families appreciated the flexible time and structures of the early childhood programs their children attended, parents were concerned about the lack of intervention services for their children. In addition, these low-income families reported that they had limited access to early interventions and resources. The findings also highlight the importance of the use of positive coping methods (e.g., maintaining a positive outlook and seeking social support), and the role faith plays in family life.

Background Mental health problems, such as depression, are increasingly recognized as a major threat to the capacity of older adults to fully participate in daily life [1,2]. China represents an anomalous case in this regard, with lower than expected prevalence of depression in its older population [3][4][5]. This deviation has been explained by cultural factors found in most East Asian nations, most notably strong filial responsibility and high levels of respect and family support for older people deriving from Confucian values of filial piety [6,7]. However, recent studies have found the prevalence of depression among older adults in China has risen precipitously over recent decades [8], while at the same time families have grown smaller, altering the contours of family life of older individuals [9]. Urbanization and societal developments in China have arguably increased uncertainty about care availability among older adults, with commensurate risks to their mental health [10]. Consequently, it is important to explore the relationship between perceived future care support and depression in contemporary China. In this investigation, we used data from a nationally representative data set of older adults in China to examine whether perceived availability of future care predicts depressive symptoms independently, as well as in conjunction with health, social, family, and economic factors. We rely on social support and control theories as explanatory frameworks for understanding how uncertainty about care may adversely influence mental health in the older population of China. --- Social support theory We first rely on the stress buffering and double-jeopardy hypotheses within social support theory [11] which state that resources moderate the impact of unexpected and stressful events on well-being outcomes [12,13]. We argue that economic, health, and community resources may mitigate the negative impact of lacking an expected care provider as a form of stress-buffering, while economic, health, and community deficits may magnify the negative impact of lacking an expected care provider as a form of doublejeopardy [14]. The double-jeopardy hypothesis suggests that financial insufficiency, poor functional health, and living in a rural area elevate the risk impact of care uncertainty on emotional distress by magnifying feelings of precariousness. 1 Stress-buffering and double-jeopardy perspectives predict the same data pattern, making it difficult to adjudicate which is the preferred explanation, as each is consistent with the other in predicting an interaction effect. --- Control theory Control theory provides a complementary vantage point from which to view the implications of anticipated care on psychological well-being from the view of a psychologist. Control theory derives from the assumption that achieving mastery over life's challenges is an important prerequisite for reducing stress caused by adverse contingencies related to old age [11]. An important element of mastery are the resources that one can bring to bear in order to meet those challenges [15,16]. In this context, resources include both cognitive/intrinsic traits and social/extrinsic supports, and sometimes both [17][18][19]. For example, the perception that social support will be available to satisfy needs caused by physical decline in later life represents the cognitive appraisal of one's control over a potential external resource. Lacking a sense of personal control over meeting future needs has been found to have adverse consequences for health and wellbeing of older adults [19]. Two aspects of control theory are relevant to our investigation. First, anticipating that care needs will be met is a desired state and an important aspect of security about future uncertainties. Being able to count on others produces confidence that the social environment can be effectively managed. Second, expectations about the social environment are shaped by cultural and political contexts. China represents a case-in-point by having strong norms of filial duty and a weak social safety net for its vulnerable inhabitants [7,20], both of which raise the stakes of expecting a weak social support system in old age. The dictates of filial piety enhance expectations for support and care from family members-adult children in particular-the violation of which may lead to a sense of loss and increased distress [7]. --- Factors related to care expectation The expectation that a care provider will be available when needed is likely to be associated with deficits that are also related to depression. Examples of such deficits include poor health, financial inadequacy, lack of kin supply, social isolation, and weak public and community supports [21,22]. Consequently, we examine the association between perceived availability of future care and depression controlling for these possible confounding factors. According to the stress buffering and double-jeopardy hypotheses, we specifically focus on economic well-being, functional health and urban/rural residency as important confounders and moderators of the association between perceived availability of future care and mental health. 1 Care uncertainty is a risk factor to emotional distress as according to Chinese traditional culture, care is expected mostly from children when parents become old or disabled. Distress is usually associated with one or series of negative events and will usually go away with change in environment or removal of stressor. Professional intervention and medications are not usually necessary, but actual and perceived social supports such as usual friendship and family networks help. Depression could be the result of persistent and long-term emotional distress. Economic strain tends to be associated with psychological distress and its absence associated with psychological wellbeing [23,24]. This result has been found in older Asian populations, including China [8,21,25] and Vietnam [26]. Health is one of the most robust correlates of late-life mental health, Studies across a variety of nations consistently find that poor health-whether indicated by chronic diseases, functional limitations, or pain-is associated with depression. Studies in the United States have found that the perceived availability of social support can buffer the impact of age-related deficits and vulnerabilities on depression in later life [15,27,28], a finding also demonstrated in China [8,21,29], Taiwan [29], Korea [25] and Vietnam [26]. Rural residency is also positively associated with depression among older adults in China [22] . This finding may be explained by the unique challenges faced by rural elders such as inadequate pension coverage [22], low availability of mental health professionals, and underdevelopment of public programs serving vulnerable elders . In addition, rural elders may be left behind in their natal villages by migrant children [30]. --- Social determinants of depression Psychological wellbeing is an important component of later life quality [2]. The previous studies in China [8,21,22] and Asian countries [25,26] found that psychological wellbeing differs significantly among older population by age, gender, marital status, social and economic status, physical health and social support. In more details, education, financial status , and chronic diseases were significant and important predictors for depression in China, Vietnam and Korea; females and older people with ADLS and pains had a significant higher depression score in China and Vietnam; Emotional support could prevent both the onset and progression of depression among older adults in Viet [26]; multiple factors, including age/cohort, functional problem, marital status, urban/rural residence, lack of contacts with children, perceived future care support, active participation, senior center, community amenities, childhood health were all associated with depression in China [8,21,22]. Literature also documents the mental health benefits of perceiving that a support provider will be available in later life and perceived availability of support-even more than actual support itself-offers comfort and security to older adults, with positive consequences for their emotional well-being [31][32][33][34]. A study of the older population of Taiwan found that perceived availability of a support provider was protective of mental health [35]. Similarly, perceived future care support was shown to be negatively associated with depression among older adults based on a pilot survey data in two Chinese provinces in 2008 [21]. Nevertheless, the perceived availability of informal care has rarely been studied nationally in China and how functional health, financial adequacy, and urban/rural residency will moderate the relationship between expected care support and depressive symptoms has not been fully investigated. In this study, we are filling the research gap by, using data from a nationally representative data set of older adults in China; applying social support and control theories as explanatory frameworks for understanding how uncertainty about care may adversely influence mental health in the older population of China; investigating how functional health, financial adequacy, and urban/ rural residency will moderate the relationship between expected care and depressive symptoms. --- Hypotheses In this investigation, we rely on the social support and control theories as the frameworks to hypothesize that older adults in China who do not expect to have an available care provider will have more depressive symptoms than those who have such an expectation . Further, we hypothesize that this relationship will be at least partially explained or confounded by the presence or absence of health, financial, and social deficits . Finally, we rely on stress-buffering and double-jeopardy paradigms to hypothesize that functional health, financial adequacy, and urban/rural residency will moderate the relationship between expected care and depressive symptoms either because: resources ameliorate the negative impact of having no expected care provider or deficits elevate the negative impact of having no expected care provider . --- Methods --- Sample Data used for our analysis derived from the China Health and Retirement Longitudinal Study , a nationally representative survey of the population 45 years or above living in China. Beginning in 2011 and continuing biannually, CHARLS collects information on a variety of health, social, family, and financial characteristics through face-to-face interviews in respondents' homes [36]. For our analysis, we used the 2013 wave of data which totaled 18,246 respondents, of whom 14,988 also participated in the previous 2011 wave from which the lagged depression measure derived. Sample weights were applied in our analyses to account for design effects and survey non-response. --- Measures The dependent variable of interest, the score of depressive symptoms, was measured with 10 questions from the Center for Epidemiologic Studies Depression Scale [8,22,25]. Respondents were asked how frequently in the last week they: were bothered by things; had trouble concentrating on things; felt depressed; felt everything was an effort; felt hopeful about future; felt fearful; had restless sleep; was happy; felt lonely; and could not get going. After reverse scoring the two positively worded items, we assigned a score from 0 to 3 for each item as follows: 0 for 0 days, 1 for 1-2 days; 2 for 3-4 days; 3 for 5-7 days. Scores of these 10 items were summed to create an additive scale score ranging from 0 to 30, with higher scores indicating more depressive symptoms. The reliability of depression items was tested using Cronbach's alpha and found to be satisfactory at each wave . The key independent variable, perceived availability of future care , was measured by the following question "Suppose that in the future, you needed help with basic daily activities like eating or dressing, do you have relatives or friends who would be willing and able to help you over a long period of time? The response option was yes or no. Coding this variable in such a manner allows us to discuss unavailability of support as a risk factor for depression. Additionally, for respondents answering "yes", a follow-up question was asked about whether the source of expected care support would be children, other relatives, or friends. Examining those with care expectations by source of care revealed the importance of off-spring in care availability: 98% of respondents expected care from their children while only 2% expected care from other relatives or friends. This provides evidence to support the existing of traditional filial piety culture in China and the importance to control for the number of children, the frequency of contacts/visits and living arrangement between parents and children into the final regression models. We chose potential control variables based on previous studies of the determinants of depression in China and other Asian countries [21,22,25,26,29,[37][38][39]. These variables were categorized as demographical, socio-economic/ finance, health and social factors. Demographics included age and gender. Age was divided into four groups: 45-54, 55-64, 65-74 and 75+. Socio-economic status was represented measured by the highest educational level achieved, perceived living standard, and urban/rural residency. Education was measured as the highest level of education achieved based on three categories: Primary schooling or less; secondary schooling; and college or higher degree. Relative living standard was assessed by responses to the question "Compared to the average living standard of people in your city or county, how would you rate your standard of living?"2 Response options were much better, a little better, about the same, a little worse, much worse, collapsed into three categories corresponding to better, same, and worse [40]. Worse than the average relative living standard is treated as having financial strain/uncertainty. Urban and rural residency was determined by the most recently published statistical standard by the Chinese National Bureau of Statistics based on an area's social and economic development [40]. Health factors included limitations in activities of daily living , limitations in instrumental activities of daily living , number of chronic diseases, functional loss, poor memory, level of chronic pain, and childhood health. ADL limitation was indicated if the respondent -reported difficulty performing any of the following basic activities: bathing/showering, eating, dressing, getting into or out of bed, using toilet, or controlling urination and defecation. IADL limitation was indicated if the respondent reported difficulty in any of the following household activities: doing household chores, preparing hot meals, shopping for groceries, managing money, and taking medications. Chronic diseases were assessed as the number of diagnosed health conditions categorized as none, one, two to three, and more than three. Functional loss was indicated by whether respondents reported any of the following disabilities: brain damage/mental retardation, vision problem , hearing problem and speech impediment ." Self-reported memory was assessed with the question: "How would you rate your memory at the present time?". A dichotomous variable was created differentiating excellent/very good/good from fair/poor. Level of pain was ascertained by the question "Yesterday, did you feel any pain?", of pain was assessed as no pain , a little pain , some pain , quite a bit of pain , and a lot of pain . Childhood health was measured by asking "How would you evaluate your health during childhood, up to and including age 15: excellent , very good , good , fair , poor . Both level of pain and childhood health were continuous variables controlled in the multivariate regression models. Social factors included family structure, intergenerational arrangements , and social activities. Family structure was measured by marital status and number of children. Marital status was operationalized as currently married or cohabitating, formally married , and single or never married. Number of living children was categorized into four groups: no children, one child, two children, and three or more children. Intergenerational contact was measured as the most frequent form of contact with non-coresident adult children. Contact frequency was categorically assessed as daily, weekly, monthly, once per year, or less than once per year. Geographic distance from children was based on the location of the closest child: same/adjacent household/ dwelling/courtyard, another household in your village/ neighborhood, another village/neighborhood in your city/ county/district, or outside your city/county/district. Three categories were generated from these responses: Having at least one child living in the same or adjacent household, having a child living in the same village/neighborhood but not co-residing or living adjacent, and having all children living beyond the village/neighborhood. Participation in social activities was assessed by whether or not respondents participated in each of three types of activities in the last month: leisure activities ; helping activities ; and educational activities . The CHARLS interviewed both the husband and the wife in a same household as long as they were both aged over 45, as one of them being the "main respondent". We note that the CHARLS study randomly selected one respondent per household to report family-level information, including number and location of children, as well as contact and exchanges of money and support with children. This method resulted in about half of our sample having no directly reported data about children. We empirically handled this issue by generating a category for "non-response" in order to retain these cases in our analysis. As a robustness check, we also estimated models borrowing values from reporting respondents and found very similar results . --- Analytic approach Previous studies on risk factors of depressive symptoms among older adults in China generally find that depression correlates with being older, female, retired, physically disabled, chronically ill, financially stressed, and low educated, and having weak social and family support systems [6,8,21,22,41]. We account for these factors in our predictive models of depressive symptoms as a function of expected future care. Emotional distress deriving from uncertainty in meeting future care needs may be assuaged by financial resources , good functional health , and living in an urban area . In order to select the most relevant control variables, we used univariate OLS regressions predicting depressive symptoms to identify plausible variables with coefficients significant at <.10 and r-squares higher than 0.01. Variables that did not meet these criteria were excluded from our analysis. Selected variables are presented in Table 1. Multivariate OLS regression was then used to explore the association between perceived future care availability and depressive symptoms, first with only demographical variables controlled, then sequentially adding health, socioeconomic, and social variables until all control variables were entered to the final model. We used this hierarchical estimation approach to assess the unique contribution of each variable grouping toward explaining the association between perceived future care and depressive symptoms. Next we tested stress-buffering/double jeopardy hypotheses by adding interactions between perceived future care availability and financial insecurity, functional disability, and urban/rural residence. We estimate the confounding effect of rural-urban residents, financial conditions, and ADL when they are controlled parallelly with perceived care support, while estimate the moderating effect when they are interacted with perceived care support in the model. Since we were alert to the possibility that perceived availability of future care is endogenous to depressive symptoms--a condition that would exist if depressed individuals were less capable of mobilizing an effective support network-we also estimated regression models controlling for a lagged measure of depressive symptoms taken in 2011. Finally, as mentioned above, we have treated respondents with missing information on family support into a "not reported" group as it is missed by random so that we can run the model with full sample size. However, robustness check has been done by assigning the reported values to other respondents within a same household. Only very small and insignificant differences were found between using the initial data and imputed data . --- Results --- Descriptive analysis Distributions for all study variables are shown in Table 1, as well as differences between sub-groups in average depressive symptoms and in the proportion of those who expect that future care would be available to them. Among 17,000 respondents, a large majority of respondents aged 45 and older expected that care would be available to them in the future. That almost one-third of the sample did not anticipate the availability of a care provider outside of their spouses, signifies the possibility of a care-gap for older adults in Chinese society, even among those with children, particularly when their children were not living in a same city/county. Further, our univariate analysis indicated that respondents without future care expectations had significantly higher depressive scores than those who had such expectations . We restrict our discussion of bivariate results to factors that were associated with both care expectations and depressive symptoms. Importantly, we control for these factors in regression models in order to isolate the unique contribution of care expectations in predicting depressive symptoms. Results in Table 1 show that when compared to the youngest residents aged 45-54, those aged 55-64 were less likely to expect care availability as well as have more depressive symptoms; while those aged 75 and older were more likely to expect care availability, but had fewer depressive symptoms.. Women were more likely than men to expect future care availability but experienced more depressive symptoms. Older Chinese living in urban areas and those with at least college education were less likely to expect care availability as well as experienced less depression when compared to rural and less educated individuals, respectively. Respondents who had two or more chronic diseases, poor/fair memory, more pain, and poor childhood health, were less likely to expect care availability and reported more depressive symptoms than their counterparts. In terms of family factors, those who never married, were childless, and had children who visited infrequently were less likely to expect care availability and experienced more depressive symptoms when compared to their counterparts. With respect to social activities, engaging in leisure activities was associated greater likelihood of care availability and fewer depressive symptoms; in contrast engaging in educational activities was associated with less expected care availability and fewer depressive symptoms, --- Multivariate results Estimated coefficients for perceived ability of future care applying various sets of control variables, both with and without controlling for lagged depressive symptoms, are shown in Table 2 . Both contemporaneous effects of PAFC and lagged effects of PAFC are presented. The baseline model with no controls shows a significant positive relationship between PAFC and depressive symptoms. Recalling that no PAFC is coded as "1" if care was not expected, this result indicates that those who did 2), indicating that the effect of PAFC on depression is robust to the application of a rich set of control variables as well as a lagged indicator of depressive symptoms. We tested interaction models to examine whether financial, health, and urban/rural conditions alter the adverse impact of not expecting care on depressive symptoms. As shown in Table 3, each of the three interaction terms were tested separately in a contemporaneous model and a lagged model . In the contemporaneous model , all interactions tested were statistically significant and in a direction that suggests the impact of lacking an expected care provider on depressive symptoms was more severe for those with functional impairment, worse financial status, and rural residency . In models with lagged depressive symptoms controlled, only functional impairment significantly interacted with PAFC. Based on estimated coefficients from the full model, we calculated and then plot predicted depressive symptom scores formed by the interaction between ADL functional impairment and PAFC, holding all covariates at their mean values. These predicted values are presented in panel A of Fig. 1 without controlling for lagged depression and panel B of Fig. 1 with a control for lagged depression . These figures reveal a pattern characteristic of buffering or double jeopardy, although somewhat weaker in strength with lagged depression controlled. Among those respondents who were functionally healthy, there is little change in depressive symptoms by whether or not future care was expected. However, in the presence of disability, not anticipating a care provider was particularly disadvantageous. Viewed another way, the combination of disability and the absence of a future care provider produced elevated depressive symptoms, a form of double jeopardy in the consequences of expecting unmet need. --- Discussion This investigation examined depressive symptomology among middle-aged and older adults in China as related to their expectation of having a care provider available to them. We found a strikingly large percentage-almost one-third, of the Chinese population aged 45 and olderthat did not expect to have a care provider in the event of future need. Even with the rapidity of social, economic, and family change in China, such a high prevalence of uncertainty is surprising in a nation still guided by the Fig. 1 Predicted depressive score for interaction group by perceived availability of future care and functional impairment and . Notes: Predicted depression score was calculated based on the significant estimated coefficients Tables 6 and7 in Appendix 2 and mean values of all the significant predictors. Data source: CHARLS 2013, weights are used precepts of filial piety, where children are expected to provide for their parents' needs. Most notable among our findings was the persistence of care expectations in predicting depressive symptoms, even after controlling for a large range of economic, health, and social factors as well as a lagged predictor of the outcome. Although about one third of the effect of perceived availability of future care on depressive symptoms was explained by these covariates, the robustness of this findings provides evidence for the utility of social support and control theories as explanatory paradigms. We infer from our results that uncertainty about having a care provider to meet basic needs in later life weakens one's sense that the future is predictable and controllable, thereby inducing distress. That the expectation of a future care deficit is tied to a demonstrable outcome confirms the power of subjective appraisals to influence affective states, harkening back to the observation of Thomas and Thomas [42] that "if define situations as real, they are real in their consequences" . Our findings also indicate that elevated emotional distress among those with uncertainty about having a future care provider is partially explained by the absence of children and, among those with children, having infrequent contact with them and living farther from them. Given declining fertility rates in China, these findings suggest that the anticipation of unmet need may surge in the older population with concomitant consequences for their mental health. Our test of resource-buffering and double-jeopardy hypotheses was most strongly observed with respect to functional health. Functionally healthy individuals have the physical resources to resist or delay consideration of the consequences of lacking a future care provider. Alternatively, functionally impaired individuals are prone to realizing the true implications of being absent a caregiver. Interactions with financial security and urban/rural residence were fully explained by the lagged measure of depressive symptoms. That is, pre-existing depression in the high-risk groups formed by the intersection of care uncertainty with financial stress and rural residency provided an alternative explanation for data patterns, affirming the utility of a longitudinal approach with lagged predictors. Our findings that ADL and worse living standard could increase depression generated by the lack of PFCS is consistent to the findings in Bangerter et al. [12] that bad physical health and low material support could increase depression of children generated by mother's problems. However, the moderating effect of urban /rural residence on depression found in our study is new and unique in literature. Several limitations of our investigation deserve mention. First, our measure of perceived future care availability was limited to one question with a dichotomous response option. Thus, it was not possible to discern the degree of uncertainty or assess the reliability of this indicator as a true measure of uncertainty. Second, we did not assess whether the expectation of a care deficit prospectively results in an actual care deficit, leaving open the possibility that individuals without an expectation for care will eventually recruit a care provider from their informal network. Alternatively, those individuals without such an expectation may have their needs met by formal care services. Further, some individuals who expect to have a care provider may be disappointed and not have that individual available when needed. Finally, future cohorts of older adults in China will meet the needs of old age care under very different conditions than existed for the current cohort, having been exposed to a relatively more prosperous economy, having smaller families, and experiencing new forms of filial piety. These historical exposures may produce new challenges, but also new opportunities for older adults in meeting the needs of later life. Economic growth will deliver financial resources to older adults that may mitigate some of the disadvantages a declining supply of offspring. In addition, the Chinese government is beginning to develop home-and community-based services as well as a residential old age care system that, while still limited, are likely to be further developed to the benefit older adults in the future. --- Conclusions Our major findings suggest that uncertainty about future care pervades a large segment of the older population of China, and is not without consequences for emotional wellbeing. Given the aging of the Chinese population, the attendant growth in the chronic disease burden [43], and attenuation of filial resources, the government will likely develop policies to meet the expected growth in unmet need. Such policy initiatives include financially incentivizing adult children to continue their caregiving roles, developing community-based care, instituting a long-term care insurance program, and providing low-cost service-enriched housing for frail older people who do not have sufficient family support. In addition, our results underscore the importance for China of developing mental health services for middleaged and older individuals, particularly targeting older adults with health difficulties and limited family support. To the degree that uncertainty with regard to future care adversely influences mental health, we suggest that greater attention be devoted to addressing the implications of weak formal and informal safety nets for older adults. Meeting mental health needs in later life may become more challenging as the pace of economic development, cultural change, and urbanization accelerates in China. We suggest that future research explore how expectations for care have shifted in relation to these new contingencies, and their implications for the psychological and physical well-being of older adults in China, as well as in other middle-income and emerging economies undergoing similar demographic and social change. --- --- Appendix 1 --- --- --- Competing interests The authors declare that they have no competing interests. ---

Background: Major concerns have arisen about the challenges facing China in providing sufficient care to its older population in light of rapid population ageing, changing family structure, and considerable rates of internal migration. At the family level, these societal changes may produce care uncertainty which may adversely influence the psychological wellbeing of older individuals. This paper applies social support and control theories to examine the relationship between perceived availability of future care and psychological wellbeing of older adults in China, and how this relationship is moderated by economic insufficiency, health vulnerability, and urban/rural context. Methods: Analyses are based on data from the China Health and Retirement Longitudinal Study, a multi-panel nationally representative household survey of the Chinese population aged 45 years and older. Data are taken from 2013 and 2011 waves of the study, with an initial sample size around 17,000, in which around 11,000-14,000 respondents are used for our final regression model. The score of depressive symptoms was measured in both waves with the Center for Epidemiologic Studies Depression Scale (CES-D10), and perceived availability of future care was measured in 2013 by asking respondents the question "Suppose that in the future, you need help with basic daily activities like eating or dressing, do you have relatives or friends (besides your spouse/partner) who would be willing and able to help you over a long period of time (yes/no)?" Results: Multivariate regression analysis revealed that uncertainty regarding future care support was associated with greater depressive symptoms even after controlling for factors confounded with care uncertainty such as family structure, socio-economic status, and a lagged measure of depression. Further, older adults without an anticipated source of care faced double jeopardy in their depressive symptoms if they also experienced functional limitations. Conclusions: Considering rapid aging of the Chinese population, anticipated increases in chronic disease burden, and possible attenuation of filial care, this analysis suggests that older adults in China may increasingly face health and social conditions detrimental to their mental health. Polices that remedy these concerns should be discussed, developed and implemented.

The use of web pages for personal purposes during working hours, often referred to as cyberslacking or cyberloafing , has to an increasing extent become an issue of concern . Employers typically fear that such use may reduce productivity, and as such cause economic loss . This fear may be rational, as approximately four of five employees who uses computers at work do visit web pages for personal purposes during working hours . As a result, employers have tried to limit such use of the Internet by the implementation of specific usage policies at the workplace . Still, it has been suggested that use of the Internet for personal purposes during working hours may also have some beneficial effects, such as counteracting stress and boredom, and increasing job satisfaction and creativity . In a recent randomized controlled trial 96 undergraduate management students were allocated to three different 10-minute rest conditions during a letter search task; a control condition involving another simple task; a rest condition where the students could do what they wanted except surfing the Web; and a Web-surfing condition. After the break the students in the Web-surfing condition were the most effective ones. The Web-surfers also reported lover levels of exhaustion, boredom and higher levels of engagement, compared to the other conditions. The authors suggest that Web-surfing is restorative as it normally is pleasurable . In a survey consisting of 191 respondents it was found that most respondents felt that some form of cyberloafing at work would be acceptable. Men cyberloafed more than women. The respondents generally experienced that cyberloafing had a positive effect on work, men somewhat more than women. In general, browsing was found to have a positive effect on mood, whereas e-mailing was found to have a negative effect on mood . At present, not much is known about factors that influence attitudes towards and the actual use of the Internet for personal purposes while at work. In one study of 1,024 workers, it was found that male gender, educational level, and work autonomy were all positively associated with cyberslacking, whereas age was negatively related to such behavior . In another study, it was found that the norms of coworkers and supervisors were positively associated with minor cyberloafing, but unrelated to major cyberloafing. In the same study, it was found that cyberloafing was positively associated with an external locus of control . In a study of 310 employees at a university in the Philippines, Restubog et al. found that gender was unrelated to cyberloafing, that age was positively associated with cyberloafing, and self-control was negatively associated with cyberloafing. There was a stronger negative relationship between perceived organizational justice and cyberloafing for employees with high as opposed to low levels of self-control. In the only study we are aware of concerning this topic that used a representative working sample, cyberslacking was positively related to male gender and racial minority, and negatively related to age . Online social network sites have become immensely popular during the last decade . Such sites are defined as web-based services that allow individuals to: 1) construct a public or semipublic profile within a bounded system, 2) articulate a list of other users with whom they share a connection, and 3) view and transverse their list of connections and those made by others within the system . Although it is likely that such sites have over one billion users worldwide, we are still not aware of any study that has investigated factors that might be related to the personal use of such sites at the workplace. Some previous studies have in general linked the use of social network sites to the five-factor model of personality, which implies that human personality can be assessed along five main dimensions: Extraversion , Agreeableness , Conscientiousness , Neuroticism , and Openness to Experience . Some controversy exists regarding the latter dimension, as some prefer to denote this as Intellect/Imagination rather than Openness to Experience . In one study based on this model, it was found that addictive tendencies towards social network sites were positively related to Extroversion and Neuroticism, and negatively related to Conscientiousness . Another study by and large confirmed these findings, as the authors reported that Extraversion, Neuroticism, and Openness to Experience were all positively related to the frequency of social network site usage . In a recent study, it was found that the use of social network sites was positively related to Extraversion, but negatively related to Conscientiousness . In one study the five-factor model of personality was investigated specifically in relation to cyberloafing. After completing a questionnaire concerning demographics and items pertaining to the five-factor model of personality, the participants, 213 students, were instructed to watch a video lecture on a computer. On each computer a program registering websites visited and instant messages exchanged was installed. Cyberloafing were negatively related to Agreeableness and positively related to Extraversion. An Extraversion x Conscientiousness interaction effect was also found, reflecting that for students with high scores on Conscientiousness, Extraversion had little impact on cyberloafing, but for students with low scores on Conscientiousness, Extraversion were strongly and positively related to cyberloafing . In a survey-based study with 147 respondents it was found that Neuroticism and Extraversion both were positively associated with cyberloafing, after adjusting for demographic variables . The use of social network sites at the workplace may also be associated with work-related factors. One such potential work factor is job demand, which is assumed to be positively related to effort , and can accordingly be expected to be negatively associated with loafing. Furthermore, empowering leadership has been found to improve performance in teams , and it can thus be speculated that this also may counteract loafing at work. Another work-related factor which may prevent loafing at the workplace is a social climate characterized by trust and social support, as this has been found to facilitate communication on task-related issues and the coordination of work . Finally, positive work challenges has been found to be positively related to job satisfaction, job involvement, as well as organizational commitment, and might therefore be an additional barrier to work-related loafing. Up until now, most previous studies on cyberloafing have focused on the use of web sites for personal purposes in general. Furthermore, the majority of previous studies related to this topic have relied on a relatively low number of respondents. In addition, assessment of cyberloafing has to a very limited extent been based on scales with well-documented psychometric properties. Further, we are not aware of any study that has investigated whether the five-factor model of personality may be related to cyberloafing. Finally, very few studies have systematically investigated whether central work and organizational variables are related to cyberloafing. Against this backdrop we conducted a survey, in which we constructed an attitude and a behavior scale that reflects the use of social network sites for personal purposes at the workplace. We assessed demographic variables in terms of age, gender, relationship status, education, and management position. Personality was assessed in terms of the five-factor model of personality, whereas work-related factors were assessed in terms of the availability of social network sites at work, policies prohibiting cyberloafing, work demand, empowering leadership, social climate, and positive work challenges. The aim of this study was to examine whether the demographic, personality, and work-related factors could explain the variance of the scores on the attitude and the behavior scale that reflect the use of social network sites at work. --- Methods --- Sample and procedures Respondents were recruited to a survey concerning the use of social network sites at their workplace, through a link available in articles about Facebook addiction published in the Internet edition of seven different Norwegian newspapers. After clicking the link, potential responders obtained access to a web-based survey. The survey was administered by SurveyXact . All responses were stored on a server administered by this company. After one month the data was retrieved, and SurveyXact sent an Excel file containing all the collected data to the research team. Of the respondents, 11,018 respondents were employed at the time they took the survey, and answered all questions. This sample was comprised of 5,656 males and 5,362 females. 7,082 were currently in a relationship and 3,936 were not. In terms of education, 555 had completed only compulsory school, 1,349 had completed high school, 2,713 had completed vocational school/technical college, 3,904 had a bachelor's degree, 2,283 had a master's degree, and 217 had a Ph.D. In terms of professional position, 811 were top-level managers, 1,821 were midlevel managers, 2,764 performed other managerial functions, and 5,622 had no managerial duties at all. The mean age of the sample was 35.4 . Obtaining informed consent was not considered necessary by the Regional Committee for Medical and Health Research Ethics in Western Norway, as no intervention was conducted and because all data was collected anonymously. --- Instruments --- Demographics Questions concerning age, gender, level of education, relationship status, and managerial responsibilities were asked. --- Attitudes Towards and Personal Use of Social Network Sites at the Workplace Although there has been developed scales assessing the content of cyberloafing no scale or inventory assessing general attitudes or general cyberloafing behavior concerning the personal use of social network sites at the workplace has so far been constructed. We therefore generated six items designed to measure attitudes towards the personal use of social network sites at the workplace as well as seven items designed to measure actual personal use of social network sites at the workplace. All items were answered along a 5-point Likert scale . The respondents were informed in the instructions that the items pertained to the use of Facebook/Twitter or other social network sites for personal purposes. In order to assess the factor structure of these 13 items, data from half of the sample was randomly drawn and subjected to an exploratory factor analysis. The Kaiser-Meyer-Oklin measure of sampling adequacy was .92, and the Bartlett's test of sphericity was significant , demonstrating that the data was suitable for factor analyses. Principal component analysis was used as the extraction method, and all factors with Eigenvalues greater than 1.00 were retained and subjected to a direct oblimin rotation. In order to validate the factor structure, a confirmative factor analysis and correlation between the two latent variables) was conducted on the other half of the sample . --- Social network site availability and employer policy regarding use of social network sites at the workplace In order to assess the availability and accessibility of social network sites at the workplace, the respondents were asked to indicate their level of agreement with the following statement: "At my workplace I have the technical opportunity to access Facebook/Twitter or other social network sites via a PC, mobile phone, or similar" . As a measure of employer policy regarding the use of social network sites during working hours, the following three items were constructed: 1) "My employer has imposed restrictions for employees concerning the use of Facebook/Twitter and other social network sites during working hours," 2) "My employer has created technical obstacles which make it difficult to use Facebook/Twitter or other social network sites during working hours," and 3) "My employer has implemented specific rules for the use of Facebook/Twitter or other social network sites during working hours." All three statements had the same response alternatives , and according to the instructions the questions pertained particularly to personal use of social network sites during working hours. The scores on these three items was added in order to make a composite score indicating employer restrictions concerning personal use of social network sites during working hours. --- Mini International Personality Item Pool -Five-Factor Model measure Mini-IPIP was used as a measure of the five-factor model of personality. The Mini-IPIP comprises a total of 20 items, four reflecting each of the following five dimensions: Extraversion, Agreeableness, Conscientiousness, Neuroticism, and Intellect/Imagination. Each item is answered on a 5-point Likert scale . The Cronbach alphas for the five subscales of the Mini-IPIP in the present study were .78, .75, .66, .66, and .67, respectively. --- General Nordic Questionnaire for Psychological and Social Factors at Work The QPS-Nordic comprises a total a 118 work-related items, 80 of which are used to create 26 scales. In the present study, four subscales of the QPS-Nordic were used: positive challenge at work , quantitative demands , empowering leadership , and social climate . Responses to all scales are provided along a 5-point Likert scale. For the first three scales the response alternatives were: 1 = very rarely or never, 2 = quite rarely, 3 = sometimes, 4 = quite often, and 5 = very often or always. For the social climate scale, the response alternatives were: 1 = very little or not at all, 2 = little, 3 = some, 4 = quite a bit, and 5 = very much . The Cronbach alphas for the four scales in the present study were .81, .76, .87, and .70, respectively. --- Statistics Table 1 shows an overview of the distribution of the nominal variables in the present study, and the mean scores and their standard deviations of the interval variables. In order to investigate how demographic, work-related variables as well as personality variables related to attitudes and behavior concerning the use of social network sites for personal purposes during working hours, two hierarchical regression analyses were performed, where the composite scores of the attitude and the behavior use scales comprised the dependent variables, respectively. Basic demographic variables, age, gender, relationship status, educational level, and management position were entered in the first step. As the educational level and management position variables were nominal, these were dummy coded before being entered into the regression analyses. In the second step, accessibility of social network sites at the work place and policies prohibiting the use of social network sites for personal purposes during working hours were entered as these variables were directly pertaining to the dependent variables. In the third step, the personality variables were entered, in terms of the scores on the five subscales of the Mini-IPIP. At the fourth and last step we entered the scores of the four scales of the QPS-Nordic . Preliminary analyses were conducted to ensure no violation of the assumption of normality, linearity, multicollinearity, and homoscedasticity. --- Results Concerning the scale assessing attitudes towards and personal use of social network sites at the workplace two factors were retained in the explorative factor analysis. The unrotated two factors explained a total of 46.5% and 12.3% of the variance, respectively. The rotated solution indicated that the six first items loaded on an attitude factor, whereas the remaining seven items loaded on a behavior factor . The correlation between the two factors was .35. The confirmatory factor analysis showed a close-to-acceptable fit with the data. The modification indices suggested correlations between the error terms for the following pair of items: 8 and 10, 8 and 11, 10 and 11, 3 and 4, 12 and 13, and 1 and 5. A model including these correlated error terms fitted the data well . The correlation coefficient between the two latent factors was .75. The standardized regression weights varied between .26 and .85. The Cronbach alphas for the attitude and the behavioral scale for the sample as a whole were .84 and .88, respectively. Table 3 shows the results for the regression analysis where attitudes towards the use of social network sites during working hours comprised the dependent variable. The demographic variables age, gender, relationship status, educational level, and professional position were entered at step 1, and explained a --- Table 2 The mean and SD for the 13 items assessing attitudes towards and personal use of social network sites at work, their pattern matrix in the explorative factor analyses and the standardized regression analyses in the final confirmatory factor analysis The results for the regression analysis where self-reported use of social network sites for personal purposes during working hours comprised the dependent variable are also shown in Table 3. The demographic variables age, gender, relationship status, educational level, and professional position were entered at step 1, and explained a total of 5.6% of the variance, F 11,10995 = 59.6, p < .01. Accessibility of social network sites at the workplace and policies prohibiting the use of social network sites for personal purposes during working hours were entered at step 2, explaining an additional of 9.1% of the variance, ΔF 2,10993 = 583.6, p < .01. At step 3, the five personality dimensions were entered. These variables explained an additional 3.3% of the variance, ΔF 5,10988 = 89.4, p < .01. Finally, the four work-related variables were entered. These variables explained an additional 0.8% of the variance, ΔF 4,10984 = 27.8, p < .01. The model as a whole explained 18.8% of the variance, F 22,10984 = 115.9, p < .01. The following independent variables were significantly related to self-reported use of social network sites for personal purposes during working hours: age was negatively related; gender was negatively related; relationship status was positively related; a high school diploma, master's, and Ph.D. degrees were all positively related ; top-level management positions were positively related ; access to social network sites at work was positively related, policies prohibiting the use of social network sites at work was negatively related; Extroversion was positively related; Conscientiousness was negatively related; Neuroticism was positively related; positive work challenges were negatively related; and quantitative work demands was negatively related. --- Discussion The overall aim of the present study was to investigate demographic, personality, and work-related factors that might be related to attitudes towards and the actual use of social network sites for personal purposes during working hours. We constructed two scales in order to assess the two latter constructs, and the data suggests that their psychometric properties were acceptable . The standardized regression coefficient for item 8 was however low. The discussion pertaining to how each single independent variable relates to these will be based on the results from the fourth and final step of the regression analyses. Age was negatively related to attitudes towards and actual use of social network sites for personal purposes during working hours. This runs counter to one study on cyberloafing but is in agreement with three previous studies within this field , and also with studies showing that use of social network sites is more prevalent among younger than among older persons . Males had more positive attitudes towards and more use of social network sites during working hours than did females. This finding is in line with previous studies on cyberloafing . People who were not in a relationship were more positive towards and used social network sites more at work than people who currently were in a relationship. Studies have for example shown that some use social networks sites to meet potential partners , and this motive is presumably stronger in single people than in people who are in a relationship. Thus, social network sites probably have a more important social function for singles than for people in a relationship, an assumption that can explain our findings. In general, the findings showed that educational level was positively related to both attitudes toward and actual use of social network sites for personal purposes during working hours. This is in line with previous studies within this field and might reflect that people with a high socioeconomic status are more familiar with computer use and modern technology than are people with a low socioeconomic status . This finding may also reflect that people with high socioeconomic status are less worried than people with low socioeconomic status about losing their jobs or they may put more emphasize on the role of social relationships for career advancement than people with low socioeconomic status. Both mid-and top-level managers expressed less positive attitudes towards the use of social network sites for personal purposes during working hours compared to respondents with no managerial functions. This is in line with studies indicating that employers generally are concerned with such use because it is assumed to influence productivity in a negative way . However, an intriguing finding from the present study was that top-level managers, in contrast to other managerial groups, reported significantly more use of social network sites for personal purposes during working hours than did employees with no managerial functions. We cannot provide any firm explanation for this, but the finding is in line with studies showing that higher-status employees engage in more frequent personal Internet use at work compared to lower-status employees , and that employers are concerned about the overall consequences of cyberloafing . Future studies should investigate more closely if workers in different positions differ in their use of social network sites at work in terms of maintaining personal relationships, and whether this is related to career advancement, bonuses, etc. Not surprisingly, access to social network sites at work was positively related both to attitudes towards these sites and to their use for personal purposes during working hours . Interestingly, policies prohibiting the use of social network sites for personal use during working hours were negatively related both to attitudes towards and to the actual use of social network sites for personal purposes at work. This does suggest that an active policy that prohibits cyberloafing works, and is as such in agreement with previous studies on this issue that show that, for example, sanctions against cyberloafing reduce the occurrence of such behavior . In terms of personality, Extraversion and Neuroticism were positively related to attitudes towards and the use of social network sites for personal purposes during working hours. These findings are in line with previous studies on predictors of the use of social network sites in general . Our findings regarding these traits are partly in line with the study of Krishnan et al. who found that Agreeableness was negatively and that Extroversion was positively associated with cyberloafing, but they are in complete agreement with the findings on cyberloafing by Jia . It has been suggested that extroverts use social network sites as a new and additional way of expressing their social tendencies , whereas people with high scores on Neuroticism are assumed to prefer web over face-to-face interactions, as the former is less anxiety-provoking than the latter . Conscientiousness was, on the other hand, negatively related to attitudes towards and the use of social network sites for personal purposes at work. This is in agreement with studies on personality and social network site use in general , but not with previous studies on personality and cyberloafing . Intellect/Imagination was positively related to attitudes towards the use of social network sites for personal purposes at work, but unrelated to such use in practice. Agreeableness was unrelated to both the attitude and the behavioral variable. The findings related to Intellect/Imagination and Agreeableness in the current study also seem to reflect general findings concerning personality and the use of social network sites , but is not consistent with other studies on the relationship between personality and cyberloafing . Interestingly, it should be noted that meta-analytic reviews have shown Conscientiousness to be positively related to job performance and Neuroticism to be negatively related to job performance . The last block in the regression analyses included work and organizational-related variables . Positive challenges at work were negatively related both to attitudes towards and the use of social network sites at work for personal purposes. This indicates that having something interesting and challenging to do at work can counteract cyberloafing. This conclusion is in tandem with studies showing that employees who have lower levels of job involvement and intrinsic involvement are more likely to engage in cyberloafing behavior . Empowering leadership was however unrelated to both dependent variables. Although this leadership style has been associated with good performance , one study also showed that autonomy was positively related to cyberloafing . Quantitative work demands were, however, negatively related to attitudes towards and the actual use of social network sites for personal purposes. Thus, having a heavy work load might prevent cyberloafing. In spite of this, work overload is not recommended as a strategy to counteract cyberloafing, as such overload is stressful and might have several negative consequences . Instead, it is suggested to focus on employees who do not have enough to do, as they might turn to cyberloafing as a means of passing time . In this perspective cyberloafing may not be counterproductive, but it can on the contrary be seen as a way of counteracting boredom and as a means to provide some stimulation. It can also for some be a way of counteracting stress. Henle and Blanchard found for example that stressors such as role ambiguity and role conflict both were positively related to cyberloafing, whereas role overload, which is similar to work demands, was negatively associated with cyberloafing. Henle and Blanchard argue that cyberloafing is a way of counteracting stress, but that role overload simply does not provide time for counteracting stress by engaging in cyberloafing. The last variable in block 4 was social climate. Although a good social climate might improve task-relevant communication , a good social climate can probably convey both liberal and restrictive norms concerning cyberloafing . This might explain why this variable was unrelated both to attitudes towards and the use of social network sites for personal purposes at work. The study by Krishnan et al. suggests that small and controlled cyberloafing rest-breaks may act as "digital water coolers" -enhancing workers productivity and effectiveness. However, the challenge for employers is to facilitate restrict, optimal and self-regulating cyberloafing among their employees. Hence, cyberloafing-breaks can function as pleasurable refreshment, and need not necessarily be a negative time consumer. The results of the present study suggest that stimulating assignments may prevent heavy cyberloafing. Job rotation or job enrichment in order to minimize narrow, fragmented work tasks, adjusting the work load and pace of work to suit employees capabilities and resources -thus avoiding both over-and underload may also prevent heavy cyberloafing during work hours. Hence, employers may facilitate these actions with a view to reducing the potential negative consequences related to cyberloafing for personal use among their staff. Concerning the strengths and limitations of the present study, a high number of respondents provided a high level of statistical power to the analyses. However, this may also imply that some trivial relationships may turn out significant. In that respect, it can be noted that block 4 in the regression analyses explained less than 1% of the variance in regard to both attitudes towards and the use of social network sites for personal purposes at work. A strength of the present study was the construction of two psychometrically acceptable scales for use as dependent variables. We also believe it is an asset that we focused on one specific use of Internet , and we are as such in line with perspectives concerning problematic Internet use, which argues for a content-specific approach . As far as we are aware, this is one of the first studies that link the five-factor model of personality to cyberloafing; which can also be regarded as a strength. The cross-sectional design employed in this study can however be considered a limitation, due to the fact that it prevents us from drawing conclusions about possible cause-and-effect relationships. This limitation also pertains to other studies of cyberloafing, thus longitudinal studies within this area should be called for . It should be noted that all data were based exclusively on self-reports; thus, the findings may be affected by the common method bias . In terms of the five-factor model of personality it should be noted that we assessed only the higher level factors, thus future studies should also investigate the relationship between the different sub-facets of the five-factor model of personality and cyberloafing. In the present study we did not specifically investigate potential interaction effects between the different five factors in terms their ability to predict cyberloafing, although one such interaction was found to be significant in a previous study . Finally, we would like to point to the convenience sample of the present study, which all analyses were based on. Due to the methodology, especially in terms of sampling and the lack of information about nonrespondents, it is not known whether the current sample is representative, and whether the data can be generalized to other populations. The fact that the respondents in the present study were recruited via a link about Facebook addiction published in the Internet edition of different Norwegian newspapers may have influenced the sample. In addition, the participants completed the survey via the Internet, which also may have biased the sample. In line with this it has been found that web-based recruitment may predispose the study to certain groups compared to more traditional recruitment procedures . This implies that the sample is not appropriate for estimation of one specific population parameter. However, the sample is assumed to be suitable for estimations of the relationship between parameters .

A total of 11,018 employees participated in a survey investigating whether demographic, personality, and work-related variables could explain variance in attitudes towards and actual use of social network sites for personal purposes during working hours. Age was negatively related to both dependent variables. Male gender, single status, and education were positively associated with both dependent variables. Managers had negative attitudes to use, but top-level managers reported more use than other respondents. Access to social network sites at the workplace was positively related to both dependent variables, whereas policies prohibiting showed the opposite relationship. Extraversion and Neuroticism were positively related to both dependent variables. Conscientiousness, positive challenge at work, and quantitative demands were all negatively related to both dependent variables.

Introduction Social and physical environments have been recognized as important determinants of health [1,2]. In Africa, rapid urbanization has changed disease patterns. Abrahams et al [3], and the references therein, describe three transition processes: the demographic transition ; the epidemiological transition ; and the nutritional transition . Official statistics show that developing countries undergoing socioeconomic transition face an increase of the chronic diseases and a decline of infectious diseases. Abraharms et al [3] note that a country or a region within a country at any given time may be at different stages of nutritional transition. In particular, the urban characteristics vary widely between cities and within the same city coexist formal and informal settlements , resulting in socio-economic-cultural inequalities that may affect the health of their residents [4]. Understanding the genesis and the shape of the city makes it possible to recognize the diversity of its neighbourhoods and the characteristics of the dwelling and their inhabitants. Expansion of the cities in developing countries, and in particularly Africa, as rural migrants seek access to work and other opportunities has resulted in the growth of informal areas without infrastructures. On the other hand, the process of reconversion of informal areas in order to provide basic services is expensive and time consuming resulting in the existence of different conditions and opportunities to their population [1]. In the Sub-Saharan African region, it is estimated about 62% of population lives in informal settlements [5]. Worldwide, it is estimated that about one billion people live in informal settlements and slums [1,6]. Accurate health statistics are almost non-existent in informal settlements and the lack of data has masked health disparities within cities [1,7,8]. It is essential that slum dwellers are also captured in health statistics to make it possible to identify and address intra-urban disparities [7]. Because different countries use different specific and dynamic terminology to identify "urban" [4] and "slum" [8], official statistics on urbanization are not sufficient to accurately describe inter-urban and intra-urban variations, and more research studies are needed. Diez-Roux [9] pointed out that there has been an explosion of interest in neighbourhood health effects within public health and epidemiology because it is clearly inadequate to consider only individual-level characteristics; it is necessary also to consider characteristics of the groups, or contexts, where the individuals belong, as well as understand how social inequalities impact health. The role of housing policy and urban planning policy on health, taking into account their impacts on the context in which individuals live and work, has been explored by several authors [6,9,10]. At least in developed countries, the role of urban planners and developers in planning of and/or modifying of the built environment to improve health populations and to reduce harmful exposures is well established. Neighbourhood deprivation and inequity in the built environment have been linked to physical inactivity, obesity and cardiovascular diseases, among others, in developed countries [11][12][13]. In Africa, there is a lack of integration of urban planners to help solve specific health problems despite recent developments that demonstrate the willingness and awareness of planners to deal with health problems in urban settings [14]. According to Ziraba et al [15], overweight/obesity might take epidemic proportions in Africa. Malhotra et al [16] reported that regional and national studies have shown differences in prevalence of overweight and obesity by age and gender, but there are few studies evaluating the association of obesity with socio-demographic factors. Moreover, these studies are based on Demographic and Health Surveys designed to collect nationally data on demographic and health indicators, usually at regular intervals of five years [15]. Abrahams et al [3] described 40 countries in Sub-Saharan Africa and Cape Verde is displayed as having relatively high levels of overweight/obesity, and low levels of underweight in women, as well as high intakes of energy and fat. In South Africa, studies for identifying priority groups for public health obesity control programs have been implemented. In other African countries, these studies are rare. According to the World Bank, Cape Verde is a lower middle income country. In Cape Verde, to the best of our knowledge, there has been no research on identifying urban health patterns. An epidemiological transition is currently under way in this archipelagic country, where the non-communicable diseases are already the leading causes of death [17]. This work describes a part of a research project, Urban Planning and Health Inequities-moving from macro to micro statistics , which aims to study three urban units/areasformal, informal and transition-in the city of Praia, using a mixed of quantitative and qualitative research methods to provide critical information for understanding how socio-economiccultural inequalities and the urban morphology affect the health of their residents. According to its national census [18], in 2010, Cape Verde had 491.875 inhabitants and the municipality of Praia presented a strong growth of its population during the prior two decades: 71.276 in 1990, 106.348 in 2000, and 131.719 in 2010, respectively, with 86.5%, 88.5% and 97.0% living in urban environment. Thus, the city of Praia had around 127.767 inhabitants, being the capital and largest city in Cape Verde. Compared to Nairobi, for example, it is a small African city, where disparities between rich and poor urban areas are not so marked. The informal unit is not a typical slum and the formal unit is not an upscale residential area. Nonetheless, it is important to investigate differences in the health profile of urban units of this small city. Involvement of residents is pointed out by Unger and Riley [8] as an important step to address social and health disparities . The concept of social capital is related with norms, trust, and networks that can facilitate collective action for mutual benefit which can be considered important determinants of health. Tomita and Burns [19] assessed neighbourhood-specific social capital in a South African study, using four variables in the SA-NIDS Household questionnaire: support network and reciprocity, association activity, collective norm and values, and safety. The authors included a question about individual preference to remain in the neighbourhood, which is considered a moderator of the relationship between neighbourhood social capital and health outcomes [19,20]. With the exception of South Africa and some studies in Kenya [14,21], to the best of our knowledge, there are few studies about this topic in African countries, compared with other parts of the world. In particular, the bidirectional relationship between neighbourhood design and social capital was explored in Galway, a small city of Republic of Ireland [22]. The potential of social capital of informal units for action in health is also described by Unger and Lee [8], among other, giving a particular attention to slums of São Salvador da Bahia and Rio de Janeiro in Brazil. Cape Verde is characterized by a historical grounding of transnational family life contributing to a central role of the women in the family structure , frequently labelled 'matrifocal' or 'female-headed' [24]. The female position in the family was fostered in a context of flexibility of households and instability of conjugal relations, described by several works [23,24]. In fact, most women in Cape Verde do not live with the father of their children and receive almost no financial or emotional support from them, a reason why most women organize their lives counting on other members of the household and other women living in the neighbourhood. This scenario is the basis for the description of Cape Verde by Drotbohm [24] as a country where female relatives still carry more responsibilities than males. The local notion of family as not necessarily based on biological kinship or bound to a specific locality but having its centre in a certain household was shaped by the historical development of Cape Verdean culture based on migration [24]. The Commission on Social Determinants of Health [25], and other authors , also highlighted the role of the collaboration between civil society and local communities on the one hand, and policy makers and researchers on the other to minimize systematic differences in health of different groups living in urban contexts. As pointed out by Ompad et al [26], identifying and addressing disparities in terms of social determinants of health is an important step to achieve the Millennium Development Goals, namely the first three goals related with poverty, education and gender. In urban poor settings, Kjellstrom and Mercado [1] pointed out gender as the major determinant of disadvantage in health. In the scope of the UPHI-STAT project, the social and physical environments of each unit/ area or subarea were described and their residents were inquired, giving attention to biological and socio-demographic variables, some household possessions and amenities, movements inside and outside areas related with work and access to services and food. The individual data were focused on variables with importance to chronic diseases , diet, physical activity). The main objectives of the UPHI-STAT project are: to characterize the morphology of the city in terms of social and physical environments , in the city and in the three urban units; to characterize each unit in terms of socio-demographic characteristics, physical activity, diet, and biological markers to cardiovascular diseases; to explore associations and correlations between different types of variables at individual, subarea and area levels; to identify the health profile of each urban unit; and to provide critical information for understanding how socio-economic-cultural inequalities and the urban morphology affect the health of the individuals of the areas and subareas. The lack of high-quality data to support evidencebased policies continues to be a concern in many African countries, and the UPHI-STAT project can be used as a baseline for future surveillance in Praia. To our knowledge, this intraurban study is pioneer in Cape Verde. The project was implemented by a multidisciplinary team that combines different areas of expertise, including international public health, nutrition, urban planning, social epidemiology and statistics. This manuscript will present some findings of the UPHI-STAT project with focus on differences among inhabitants of the three urban units; how the inhabitants perceive their neighbourhoods; what were the characteristics of the participants in the different stages of our study; overweight/obesity and abdominal adiposity indicators with focus on women of the three urban units; physical activity by gender and urban units, using self-reported information and pedometers in working and non-working days; the relationship between physical activity in leisure time, perception of security and lack of infrastructures. The manuscript also describes the data collection process and explores benefits and drawbacks in terms of methodological issues and the possible involvement of civil society and local communities in future actions. We will give a particular attention to residents of the informal unit and women, because we anticipated they are the most disadvantaged groups, in terms of some health outcomes and some modifiable factors . --- Materials and Methods --- Studied areas This is an intra-urban study [26] that includes three urban units of the city of Praia, Cape Verde , corresponding to the following neighbourhoods: Plateau -the formal unit; a part of Palmarejo-the transition unit, and a part of Vila Nova-the informal unit. The three neighbourhood units have different levels of infrastructure and population density. Understanding the urban morphology of the city is based on different analyses: quantitative , organizational and functional and qualitative . The formal urban unit reflects a more regulated and consolidated urban form . The informal unit is characterized by the absence of basic infrastructures and services with sidewalk and roads in poor conditions, because this area has grown without an effective planning model. This unit has reduced economic activity and the majority of the population has rural origins. The transition unit combines both formal and informal characteristics. In 2014, Praia's formal and informal areas comprised about 43% and 21% of the total of the city, respectively [27]. We use "urban unit" or "urban area", however, this terminology varies across countries. Based on Geographic Information Systems , the delimitation of each urban unit was performed using the Geographical Urban Units Delimitation model [28] adapted to the context of this city and to the objectives of this research. The GUUD model operates on city macro scale by defining simplest set of urban units, called by "cellular units", that can be considered homogenous and representative in accordance with the local reality. The model makes it possible to consider a variety of determinants as: urban compactness and morphology, street patterns, zoning, population density, socio-economic characteristics and environmental performance. Statistical subsections joined to the information on geographical data base provided by the INE CV [18] have been adopted as starting point for the urban unit . The statistical subsection has also been adopted as a geo-referenced grid in order to implement this approach to the city of Praia. The use of this geo-referenced grid allows to work with several levels of quantitative information in the research process. From within this geo-referenced grid, the GUUD model is calibrated to collect a selection of data inputs to focus the delimitation criteria: construction timeline; population density; urban morphology and land-use patterns. These frameworks enable one to shift from the macro scale of city to the micro scale of urban unit, filtering the global database with selected parameters. In each urban unit, the geographical coordinates of some infrastructures, services and households, among other, were identified. The geographical coordinates of the private households were explored to provide the sampling frame used in the sampling strategy. --- Sample size, sampling strategies and data collection Sample size determination, to implement a research project in the field, needs to address statistical requirements and time, cost, human resources, and ethical and technical issues. The sample size required to estimate a binomial proportion , using a confidence interval with a specified level of confidence and a desired precision, for both an infinite population and for a finite population, is a common problem with several answers, according to desired values, methods and initial estimated proportion [29,30]. According to WHO [31], in 2008 the estimated prevalence of elevated blood pressure in Cape Verde was 44.1% and the prevalence of overweight was 34.4%. As the first estimate results in a larger sample, in each area, around 592 inhabitants should be selected to ensure a 95% confidence level and a desired precision of 4%, using the Wald method. When the research protocol was designed we expected to select at least 1776 individuals -one adult randomly selected in each household-in the three areas. However, when the project was approved, new data were obtained about the prevalence of some events of interest and about the population in each area, thus we increased the sample size and considered a proportional sample allocation to each urban unit. At the end a total sample with n = 1912 was obtained. To select a random sample we needed the sampling frame, i.e., a complete list of all residents at least 18 years old who lived in each unit for at least 6 months. Given the lack of this type of sampling frame, we developed an alternative sampling frame based on the geographical coordinates of private households in each urban unit, combining GIS and statistical software. Nonclassical households and homeless were not included in this study. The urban planning team identified the geographical coordinates corresponding to households, providing the centroid of the polygons which is supposed to represent a building or a detached house. However, the spatial visualization shows roofs which may represent a household or a set of households, for example, a building with 7 floors with 2 households per floor. In the last case, we repeated the corresponding geographical coordinate 14 times. Field workers were needed to complete this exhaustive field work in order to provide a more realistic list of households in each area. This list was exported to SPSS statistical software and a random sample was generated for each area. The geographical coordinates of these samples were again exported to the ArcGIS program to mark them on a map to facilitate the work of the 14 interviewers trained to apply the UPHI-STAT questionnaire . The interviewers visited each random geographical coordinate, identifying how many households were eligible in each building and all adults living there. Each interviewer carried a set of cards to represent the number of households of a building and the number of adults of each household to randomly select one or more floor and one adult by household . If the selected adult was located at the moment and gave his/her written informed consent, the UPHI-STAT questionnaire was applied. Otherwise one or two more contacts were made to locate the selected adult before that household was considered as unreachable. --- UPHI-STAT Questionnaire-First Stage In the scope of the project the UPHI-STAT questionnaire was developed and pre-tested by 14 local interviewers trained full-time for one week in November 2013 and three additional days at the end of January 2014. The survey was conducted in the three areas form January through end of March 2014. Several issues were explored in group discussions in the classroom and after field work with local communities in order to standardize procedures. As natives of Santiago island, the interviewers brought important contributions to the questionnaire and the data collection process. The questionnaire was written in Portuguese languagethe official language in Cape Verde-but "crioulo" was also used by interviewers. Topics covered in this questionnaire include: 1) socio-demographic characterization; 2) perceptions and relationship with the living space; 3) selected household possessions and amenities; 4) self-reported health status and access to health services; 5) alcohol and tobacco use; 6) eating habits; 7) acquisition of and access to food; 8) physical activity; and 9) self-reported anthropometric data-weight, height and desired weight. The final sample size was constituted by 145 participants in the formal unit, 1144 in the transition unit and 623 participants in the informal unit, in a total of 1912 respondents to UHPI-STAT Questionnaire. --- Anthropometric Measurement and Dietary Consumption-Second Stage After completing the UPHI-STAT questionnaire, the respondent was invited to visit the team of 4 local nutritionists, in a near place within each unit, to collect anthropometric measures. Anthropometric measurements and body composition by bioelectric impedance were collected using a TANITA SC330 S 1 , accurate to 100g, minimum 2kg and maximum 270kg, calibration up to 300.000 uses. The measures obtained were: body weight, body fat, muscle mass, bone mass and body water. Participants' height was assessed using a stadiometer Seca 1 , accurate to 1mm, where individuals were barefooted and with the weight evenly distributed on both feet, arms along the body, heels together, standing up straight and looking straight ahead. Waist and wrist circumferences were measured, according to standard procedures. Based on selfreported data and real measurements for weight and height, BMI was calculated according to Quételet equation and recoded into: underweight , normal , overweight , and obesity , according to WHO recommendations [33]. The body fat was categorized according to gender and age as described by previous works . Waist circumference was collected because it is used as a surrogate marker to define abdominal obesity. WC was measured midway between the lower rib margin and the iliac crest in the horizontal plane and it was recoded into categories to describe cardiometabolic risk by gender [36,37]. The waist-to-height ratio was also obtained by dividing the WC value to the height of the individual. WHtR values >0.5 suggest cardiometabolic risk [37,38]. Dietary consumption was assessed using a 24-hour diet recall applied through interview by the 4 local nutritionists. Through 24-hour diet recall, the nutritionist asked individuals to recall foods and beverages they consumed in the previous day to the interview from the moment they woke up in the morning until they went to sleep at night. The second stage was completed by 599 of the 1912 participants -22 in the formal unit, 283 in the transition unit, and 294 in the informal unit. Ahead, some characteristics of the participants of the first and second stages will be explored. --- Pedometers-Third Stage If the participant agreed, a pedometer was delivered with a registration form to collect the number of counted steps in a day , during a week. This stage was scheduled for a sample of about 200 inhabitants, according to the availability and durability of this type of equipment. The functioning of the pedometer was explained to the participants by the nutritionists. This stage was completed by a total of 118 participants: 1 in the formal unit, 80 in the transition unit and 37 in the informal area. In the statistical analysis the formal unit was not included. --- Qualitative Study-Fourth Stage According to study design, based on a preliminary statistical data analysis, a qualitative study was developed to complement the quantitative approach, intending to add data and information for understanding how socio-economic-cultural inequalities and the urban morphology affect the health of residents. Ten focus groups with a total of 48 participants in the three units and two semi-structured interviews, with the directors of two non-governmental organizations in informal and transition units, were performed during November 2014. The focus groups of each unit were defined according to nutritional status, age group and gender of the participants. All of them had participated in the first two stages and some of them also participated in the third stage. --- Ethical Approval The UPHI-STAT project was approved by two Ethic Committees-in Cape Verde and in Portugal , taking into account the field where the research takes place and the host and funding institutions of Portugal. Two copies of informed consent forms were presented to each participant, one to sign, that was given to the interviewer, and the other one to be kept by the participant. --- Statistical Analysis Exploratory data analysis and classic descriptive statistics were used to describe and summarise the main variables of our datasets. Tables herein present n for qualitative variables and median accompanied by interquartile interval , denoted by IQR, for non-normal quantitative variables. An assessment of the normality of the quantitative variables, using Kolmogorov-Smirnov and Shapiro-Wilk tests, was performed to ensure this prerequisite for many parametric tests. Another underlying assumption in parametric tests is the homogeneity of variances assessed by Levenne test. If assumptions of parametric test were not valid, we used non-parametric tests . Associations between qualitative variables and comparisons of proportions were explored through Chi-Square test or the alternative Fisher exact test. Measures of association for dichotomous variables were calculated when justified. Confidence intervals for proportions were obtained by Wilson and Agresti-Coull methods, preferable to the traditional Wald method [39]. To estimate the prevalence of overweight/obesity based on self-reported weight and height, we used the real measures to determine the sensitivity and specificity of the classification , considering the measures performed by nutritionists as a gold standard. These concepts of diagnostic tests are introduced in the corresponding confidence intervals through the Blaker's, Sterne, Clopper-Pearson and Wilson methods as described by Lang, Reiczigel, and co-authors [40,41]. The concordance between nutritional status obtained by the categorization of BMI, based on selfreported measures and real measures, was assessed by Kappa coefficient. After these initial approaches, multivariate analyses were performed using generalized linear models, among others, for analysing multiple variables in an integrated way, adjusting for potential confounders. Some simple and multiple binary logistic regression models to obtain odds ratio and adjusted odds ratio are also presented to identify possible factors associated to the adherence to nutritional status evaluation and also to explore the adherence to pedometers. The Hosmer and Lemeshow goodness of fit test and residual analysis were performed. SPSS version 22.0 , R [42] and EpiTools [43] programs were used to explore our datasets. --- Results --- Inhabitants of the urban units Table 1 describes some characteristics of the residents within the three urban units, corresponding to the following neighbourhoods: Plateau -formal unit; a part of Palmarejo -transition unit and a part of Vila Nova-informal unit . These units present differences in terms of urban planning and also exhibit marked disparities in terms of some sociodemographics variables of their inhabitants. There are no significant differences among urban units in terms of gender distribution of the respondents. In all units, there was a higher participation of women than men in the UPHI-STAT questionnaire. Significant differences at 5% level were found for all other variables presented in Table 1. The formal unit presents an older age structure and the transition unit a younger age structure . Academic qualifications reveal a gradient in terms of the percentage of respondents not attending school-formal 3.5%; transition 5.4%; informal 14.2%-and attending secondary and high school, with similar values between formal and transition units and a lower percentage in the informal unit. Based on UPHI-STAT questionnaire, the unemployment rate is higher in the informal unit , while in the other units is 6.2% and 17.5% . The high percentage of retired people in the formal unit is expected, taking into account the age structure of this urban unit. Despite this, the percentage of students in this area is also higher than in the informal unit . In terms of marital or family status and children, married and unmarried partners represent 31.1% and 34.0% in the formal and transition units, respectively, and about one quarter in the informal unit. However, the majority of the respondents have children, varying from 72.0% in the transition unit to 82.3% in informal unit. The median number of the children is higher in the informal unit, with three children, compared to two children in other units. Briefly, some health-related variables were also analysed in each unit, revealing an excess of chronic conditions reported by inhabitants of the informal unit, compared with other units, despite the older age profile in the formal unit. Regarding chronic diseases, the most commonly reported was hypertension with 15.7%. Self-reported hypertension varied significantly among urban units , with 19.3% in the formal unit, 11.4% in the transition unit and 22.5% in the informal unit. In the informal unit 11.2% of inhabitants reported that they were on a diet prescribed by a health professional, which was significantly different among urban units at 5% significance level . Almost half of the residents of the formal and transition units reported the self-monitoring of their weight, compared to 35.4% in the informal unit. --- How the inhabitants see its neighbourhood-urban unit Table 2 summarises participants' opinions and perceptions regarding their neighbourhoods and/or city. In general, the inhabitants reported enjoy living in these three units, using a 5-point scale. In the informal unit, a slightly lower percentage of respondents reporting that the degree to which they liked living in its neighbourhood was "somewhat high" or "very" , compared to respondents in the other two units . The most common reason given to live in the unit was tranquillity for respondents from Plateau and Palmarejo , whereas it was "housing and support from family" for respondents from the informal unit . Employment was reported as a reason to live in the neighbourhood, in an expected way, decreasing from formal to informal . In the last 5 years, the evolution of the three urban units was favorably classified. Very few respondents pointed out a worse situation in the formal and transition units. However, in the informal unit, 20.8% reported a negative evolution of the unit. When we asked what is necessary to improve the neighbourhood, insecurity appears as the main concern in the three units, notwithstanding with significant differences among units, with more than three-quarters of the respondents of informal unit referring the need for more security. Focus group discussions also reinforce this pattern in all units. Need for improvement in the global environment was reported by respondents of the units significantly differently and with an increasing trend from formal to informal units. The lack of some facilities was also reported, namely the need for a health centre, centres for the elderly and sport facilities, with significant differences among the urban units . The need for a health centre was reported by 62.7% and 69.0% in transition and informal units, respectively, but only by 20.7% in the formal unit, where there is a hospital. Regarding the need for centres for the elderly, perhaps the age structure and family explain the "V" pattern, with a lower value in the transition unit and higher values in other units -formal 46.2% and informal 52.3%. This type of pattern appears also for improvement of accessibility, although the issue was reported less frequently. On the other hand, an inverted "V" pattern with a higher value in the transition unit was observed in relation to the need for sport facilities, gardens and green space and public spaces. There are not significant differences among urban units regarding the need for entertainment and cultural activities , garbage removal in public spaces , public transportation and more stores and shops . Although the city does not have public transportation, this fact was not among the most mentioned issues, perhaps, due to the network of taxis and private buses easily accessible at low prices. Schools and kindergartens were mentioned by 15% of the respondents, again with an increased trend from formal to informal unit. --- Adherence to nutritional status evaluation and the use of pedometers As mentioned before, only 599 of the 1912 participants completed the second stage of our study: going to the places where the nutritionists collected the anthropometric measures and food intake information, through a structured 24 Hour Recall questionnaire. By convenience, within this text, we use a short description to designate this stage: "adherence to nutritional status evaluation". It is important to understand what characteristics of the participants or family-related variables , and environment contexts that may influence adherence to this stage. Considering adherence to nutritional status as a binary dependent variable , several simple logistic regression models were initially performed to identify a set of variables with p 0.20. To avoid confounding, these variables and other variables were analysed simultaneously using multiple logistic regression models to identify possible factors that could explain adherence to nutritional status evaluation. Some results are presented in Table 3 for a sub-sample without missing values of size n = 1833. After adjustment for potential confounders, the urban unit was still one of the significant variables, revealing a higher level of adherence of residents of the transition and informal units compared to formal unit. According to the fitted model, individuals from informal unit were nearly six times more likely to participate in this stage of our study, compared with the formal unit. Women were also revealed to be more likely to participate in this stage than men. Participants who were unemployed, students and other workers , compared to participants who reported to be working, more frequently visited the team of nutritionists. The simple logistic regression model and some multiple regression models suggested a higher participation in nutritional status evaluation by participants who self-reported suffering from chronic conditions. However, this significant association disappears when BMI based on self-reported measures of height and weight was included in the models. When included indirectly self-reported BMI in the model , it was found that a higher BMI was associated with the adherence to nutritional status evaluation. This situation is particularly relevant to the next subsection. Taking into account the lack of published works about the use of pedometers in Cape Verdean communities, it is important to explore the adherence to this type of tool. Only one individual delivered the report sheet of steps obtained with the pedometer in the formal unit. In this stage, the inhabitants of the transition unit accepted and brought back the sheet of steps more frequently than informal unit. Using a similar approach, with logistic regression models, the most important associated variables were urban unit , education, considering as the reference category "none or preschool", the adjusted Odds Ratios were 6.96, p = 0.001 for primary; 7.70, p < 0.001 for secondary and 8.81, p < 0.001 for high school. In some models diet prescribed by health professional was also a significant variable at the 5% level. --- Overweight/obesity classifications and abdominal adiposity indicators with focus on women Considering the two BMI measures , two different values of prevalence of the overweight/obesity, defined by BMI 25kg/m 2 appear. However, there are some issues that we need to address. As the adherence to nutritional status evaluation is conditioned by self-reported BMI, this result leads to the conclusion that anthropometric measures obtained by bioelectric impedance in the subgroup should be higher than the remaining group , who had self-reported measures. Consequently, using the real BMI to quantify the overweight/obesity will result in an overestimation of the corresponding prevalence in the studied population. On the other hand, the estimation of prevalence of overweight/obesity will be underestimated by self-reported measures. For a sub-sample of size n = 395 without missing values in any variable involved, the concordance between these two BMI was obtained, revealing a significant agreement . Despite this agreement, 38.2% of the participants indirectly self-classified as underweight were classified by nutritionists as normal BMI. On the other extreme, 30.6% of the participants indirectly self-classified as normal BMI were classified as overweight, and 31.0% of the respondents self-classified as overweight BMI were obese. Based on the self-reported measures for 1399 individuals, 553 were reported as overweight/obesity with a 95%CI [37.0, 42.1], obtained by Wilson method. In the same way, after excluding four pregnant women, the magnitude of the same event was 338/595 95%CI [52.7, 60.8], using the real BMI. Taking into account the issues associated with both confidence intervals, we use concepts of an imperfect diagnostic test to correct the first estimate. In fact, it is possible to use the sub-sample with both BMI to estimate the sensitivity and the specificity of the indirect binary self-classification as overweight/obesity , considering the measures performed by nutritionists as a gold standard. After that, confidence limits for prevalence of overweight/obesity adjusted for sensitivity and specificity are calculated-95%CI [38.4, 45.7]-using Blaker's, Sterne, Clopper-Pearson and Wilson methods as described by [40,41]. For each urban unit, the differences between women and men were frequently significant, as expected, at least for some measures. In this work, we do not present this type of results. Deliberately, we give a particular attention to results for women, because women present higher prevalence of overweight/obesity, and on the other hand, they show a higher discrepancy between the actual weight and the desired weight. In addition, women participated more than men in all stages of our study, which is a good indication for future health promotions in these urban units. Table 4 summarises results for measures that were self-reported, those measured by nutritionists, and others derived from these for women by urban unit. Before the table analysis, initially analysing in more detail the paired subgroup of women simultaneously with both values, there was a significant difference by Wilcoxon signed rank test . It was noticed that women tend to report a height greater than the real value. In terms of weight, there is no significant difference between self-reported and measured weight . Thus, in this particular context, the problem with indirectly self-assessment BMI is caused by overestimation of height. Table 4 shows the self-reported weight, height, and BMI, where we found significant differences among urban units, indicating an unfavorable situation for the informal unit. In percentage terms, the BMI classification indicates 55.6% of overweight and obesity in the informal unit, despite having a younger age structure than the formal unit. Also due to the reduction in sample sizes, using the measures obtained by nutritionists, particularly in the formal unit, there were no significant differences across urban units. Even taking into account overestimation of the percentage of overweight/obesity it is clear that the three units present high levels. Using fat mass , the Kruskall-Wallis test reveals significant differences among units , with similar percentiles for informal and formal units and a better situation in transition unit. The fat mass classification presented significant differences only at 5% level , and aggregating overfat and obese categories, we found 87.6%, 74.9% and 81.9% in formal, transition and informal units, respectively, for women observed by nutritionists. These percentages are higher than the ones corresponding to overweight and obese categories based on BMI. Abdominal adiposity indicators also show a critical situation, according to the recommendation of the WHO, with the percentage of women with metabolic risk about 74.6%, without significant differences among urban units. According to WHtR, 77.6% presented metabolic risk. In general, Table 4 shows a better situation in the transition unit and a worse situation for women of the informal unit. For some indicators, the informal unit presents values similar to the formal unit which is characterized by an older age structure. --- Physical activity by gender and urban unit Table 5 describes results relating to physical activity in global terms, and in work and in leisure time, for women and men in the three urban units. Concerning physical activity for women, the overall percentage was 67.4% , with significant differences among urban units only at the 5% significance level . In the transition unit, about 70.4% reported global physical activity. Physical activity in men was higher, around 85.2% , without significant differences among urban units . Women in the formal unit reported more physical activity at work, however, the percentage was only 16.3% . Women in the informal unit reported less physical activity in leisure time p = 0.005). Among men, there is a significant difference among units , and it is the informal unit that presents more physical activity at work. In terms of physical activity of men at leisure no significant differences were found between units. In terms of intensity, both men and women tended to report moderate physical activity. Regarding physical activity in leisure time, the magnitude of the practice was very discrepant by gender, with a 95%CIs [22.6, 27.4] for women and [53.2, 60.2] for men. As global physical activity also includes walking, it is important to refer the complementary information of the third stage, based on the report sheets of steps obtained with the pedometer, available only for transition and informal unit . We analysed the number of daystotal, working and non-working-registered, as well as the number of steps on each one. The sample presented a median of total steps/day of 5234.8 , with no significant differences between the two units, even though the transition unit presented a higher median. As for the number of days with registry, the informal unit had a lower median of working days and a higher median of non-working days, comparing with the transition unit . By gender, men walked significantly more than women , with a difference in the medians of approximately 2000 steps/day. --- Relationship between physical activity in leisure time and the perception of security and lack of infrastructures Possible relationships between physical activity in leisure time and the need for increased security, and lack of some infrastructures were explored using some binary variables already described in Table 2. Stratifying by urban unit and/or gender, we found some significant associations. Among all participants in the informal unit, physical activity in leisure time presented a weak negative association with perception of need for more security . Overall and among men in the transition unit, there was a positive association between physical activity in leisure time and the need for more sport facilities , respectively). There was a positive association between physical activity in leisure time and the need for more gardens and green spaces for men and overall in transition unit and also for men of the formal unit ). Men who engage in physical activity during leisure time seem to require more infrastructures in their neighbourhood. Exploring this type of association with another binary variable "walking and cycling" for at least 10 minutes continuously to get to and from places?) shows a similar trend. Stratifying by gender, a negative association between "walking and cycling" and the need for more security appears for women . Stratifying by gender and urban unit, the significance of this association was verified for women of the transition and informal units . The relationship between "walking and cycling" and "gardens and green spaces" presented significant positive associations for both men and women of the transition and informal units. Again, participants who walk and bike tend to report a need for more gardens and green spaces. Observation during fieldwork revealed unequal distribution of infrastructures for physical activity within the city. Focus groups allowed us to understand some determinants of physical activity, according to the perceptions of the populations of the three areas under study. Discussions of the focus groups seem to reinforce that insecurity of the city is a potential explanatory factor for not practicing physical activity, particularly in the informal unit, as suggested by some participant opinions: "Also in our neighbourhood, where we are, due to the level of insecurity, people do not feel safe to go out and walk." "I've been the victim of robbery. I was walking." "It is late for me and there is no security on the street . . ." "There's a lot of insecurity in the --- Discussion In addition to the differences in terms of urbanization of the three units, Table 1 shows how the inhabitants differed across urban units for some variables. With the exception of gender distribution, we found statistically significant differences in terms of socioeconomic variables , family-related variables and also in terms of some health-related variables . In the informal unit, inhabitants presented an unfavorable situation. Overall, self-reported hypertension percentage was 15.7%, varying significantly among urban units, with 19.3% in the formal unit, 11.4% in the transition unit and 22.5% in the informal unit. This overall percentage was similar to the percentage of 14.5% found in Cape Verde, in 2007, for adults between 25-64 , but is very dissimilar to the percentage-34.9%found when hypertension was measured using conventional procedures [44,45]. Doulougou et al [46] studied the prevalence of hypertension in formal and informal areas of the city of Ouagadougou in Burkina Faso and found a global prevalence of hypertension of 18.6%. Although without significant difference between urban areas, a higher percentage was found in the formal area compared to the informal area which had a younger age structure. In our study, although formal unit presents an older age structure than the informal unit, even so participants of the informal unit reported percentages of chronic diseases higher than in the formal unit . Doulougou et al [46] found lower values compared to our study, without differences between informal and formal areas . Our study concerns a very small scale, where the three urban units are geographically near each other. Despite this proximity, some related aspects of the nutritional status of participants varied significantly between urban units. Gender disparities in overweight and obesity are also marked in Cape Verde. This fact is described for some developing countries, particularly in the North Africa and Middle East, where the cultural environment favors a larger body size because is linked to fertility, prosperity and healthfulness in women [47]. In fact, in Africa it seems that obesity is more prevalent in middle-aged women in urban areas [48][49][50]. These results are consistent with other studies in populations of African countries [51,52]. Women of the three urban units present significant differences in terms of BMI calculated from self-reported measures, fat mass, waist circumference and waist-to-height ratio . The percentage of overweight and obesity based on self-reported measures differed across urban units, despite no significant differences found in the smaller samples with measures collected by nutritionists. It is possible that the reduction of the sample size from first to second stages, the small sample size of the formal unit, and the influence of the BMI in the participation in the second stage may explain theses findings. In a national study [44,45], based on self-reported weight and height, the percentages of overweight was 36.9% and obesity was 10.5%. Also in Cape Verde, Ng et al [53] reported percentages of overweight around 44.0% [41.3, 47.0] and obesity of 15.4% [13.9,17.1], for women more than 20 years old. For men, these percentages were 31.8% [29.4,34.3] and 7.0% [6.2,7.8], respectively. Our study indicated 42% of overweight and obesity among women and among men 35.5% . Our value is lower than the previous one, particularly for women. However, we found very discrepant values between transition and informal unit . In Ouagadougou , Doulougou et al [46] described lower percentages of overweight and obesity overall, with a higher percentage of overweight and obesity in the formal area than in the informal area . Discrepancies between urban and rural areas and also across urban areas within a country or different countries are pointed out by several authors . Thus, local communities would benefit from local studies that show particularities of a city at a micro-level scale. No significant differences were found among urban units, in terms of cardiometabolic risk of women, according to WHO or WHtR definitions, with 74.6% and 72.5%, respectively. For men, it should be noted that these percentages were lower 22% and 35.9% . In Tunisia, Ati et al [37] reported higher percentages, based on WHtR, for women and men between 35 and 70 years old. Ware et al [38] found lower values for women in a study with participants from rural and urban areas of South Africa in terms of WHtR and waist circumference. Within the multifactorial nature and the complexity of health problems, neighbourhood environment factors such as safety from crime and traffic were associated with overweight among Nigerian adults [54]. Among others, Sallis et al [13] describe the role of built environments in physical activity, and indirectly in obesity and cardiovascular diseases. In some developing countries under nutrition transition, a greater impact on the physical activity of women has been highlighted [47]. Doulougou et al [46] found an association between physical inactivity and hypertension in both formal and informal areas. In our study, for women, overall physical activity was 67.4% , with differences among urban units . For men, it was 85.2% , without significant differences among urban units. The difference between the percentages of women and men who reported physical activity in leisure time was particularly discrepant . Pedometers also indicated men walked significantly more than women , with a difference of approximately 2000 steps/day. In this context, recent studies show the link between pedometers and health and promotion health messages, but there is a lack of this type of data from developing countries, specifically African countries [55]. The few studies are focused on subgroups of population or associated to intervention studies, with small sample sizes. Walking has been suggested as a good from of physical activity in particular in developing countries, because is inexpensive and accessible for a vast population . However, particularly in the cities, there are many aspects to be considered to promote walking, with local urban planning and security being two of them. A borderline association was found between the perception of more security and physical activity in leisure time in the informal unit. Study participants who reported insecurity tend to be inactive in leisure time. Insecurity was also negatively associated with walking and cycling for women in transition and informal units. The qualitative study also reveals this concern, with some women justifying their lack of physical activity by citing insecurity of the environment and absence of adequate infrastructures. Participants who practiced physical activity in leisure time or walking and cycling tended to report the need for more infrastructures in their neighbourhoods. Young men of the transition unit with higher academic qualifications revealed more involvement in physical activity in leisure time. On the other hand, older women of the informal unit who perceived insecurity tended to be inactive. Cape Verde is described by Zoettl [57] as the country of "morabeza" . However, it has been facing problems of youth delinquency and gang-related violence, in the last decade. In 2007, a report conducted by United Nations and Cape Verde Ministry of Justice [58] highlighted that the most of the crimes experienced by the population did not necessarily involve violence. The reporting rates were much lower than in other African countries and among the lowest in the world, with the exception of sexual offences. Nevertheless, the crime level in Praia was higher than in other two African capitals-Gaborone and Maseru . Both citizens and police respondents ranked unemployment and drug consumption as the main causes of crime [58]. In our study, the highest unemployment rate was found in informal unit . The participants reported to like very much living in these neighbourhoods and/or city and they had positive views regarding the evolution of the neighbourhoods in the last 5 years. However, the percentage of the inhabitants expressing this positive evaluation was the lowest in the informal unit. The need for more security in the neighbourhood and/or city appeared as the main concern of the participants of the three urban units, notwithstanding with significant differences among units, with a higher percentage among respondents of the informal unit. Several studies, for example, an American study by McGinn et al [59], showed that both real and perceived crime levels in neighbourhoods discouraged physical activity of their residents. Oyeyemi et al [54] found that in Nigeria, consist with some previous studies, a neighbourhood with crime at night may have negative influence on physical activity. The fear of crime can impact physical activity and decrease confidence and remove the desire to go outdoors. In the urban area of Pelotas, Brasil, Mendes et al [60] reported lower percentage of the physical activity in leisure time and transportation in groups with high levels of insecurity. However, the association between physical activity and perceived insecurity was not significant. As reinforced by Kjellstrom and Mercado [1], violence and crime in poor urban areas have serious implications for trust and well-being. In the State of African Cities Report [61], women of urban poor areas are stated to be most at risk from disease and other social and environment problems. --- Strengths and Limitations In the UPHI-STAT research project we developed a sampling frame based on the geographical coordinates of private households in each area, combining GIS and statistical software. This strategy may be applicable to similar settings, where the identification of geographical coordinates of households is possible. A stratified random sample of size n = 1912 was obtained using proportional allocation. In each household, one adult was selected at random, from the list enumerated of all adults . Each participant was interviewed in a first contact or using two more additional contacts, in case of failure of the first or second contacts. After the UPHI-STAT questionnaire was administered, the selected adult was invited to be observed by a nutritionist in a nearby location, to collect data about dietary consumption, anthropometric measurements and body composition by bioelectric impedance. In the third stage, walking was measured via pedometers. In Cape Verde, to our knowledge, there were no previous studies using pedometers. The use of appropriate equipment is particulary relevant in the African context. Many studies are focused on self-reported and measured BMI and do not explore other important body composition measures. It is well-known that BMI does not differentiate between fat mass and fat-free mass, being a poor predictor of body fat [62]. Among several methods, bioelectrical impedance analysis is previously described [62] as safe, non-invasive, quick, and relatively inexpensive. However, in an African setting the last characteristic is not always valid. All equipment of the UPHI-STAT project was bought in Portugal due to their unavailability in Cape Verde. Calibration was performed in Portugal and some variations may have occurred during air transportation, although all measures were considered to avoid posterior variation in measurements. Testing was performed with all four sets of equipment and differences were not detected. Although this type of equipment is portable, we did not use this function in a door-to-door strategy, due to human resources and logistic reasons, including the availability of electricity in each house , privacy and protection of the information in a suitable space, and also the risk of theft within some areas of the city. Other important reasons, described among others by Dehghan and Merchant [63], are the influence of environment factors in bioelectrical impedance. These authors summarised other factors that can impact in bioelectrical impedance results and they do not recommend their use in epidemiological studies which involve participants from diverse populations. In our case, the measures were taken under controlled conditions and the homogeneity of the population of this small city seems to be ensured. From the statistical point of view, probably a door-to-door strategy would have provided a larger sample size in this second stage, but on the other hand, the accuracy of the measures would be affected. Thus, we opted for the second option, despite knowing that the sample size would be reduced. In fact only 599/1912 went to the measurement points. Keeping in mind future actions to prevent overweight and obesity, an advantage of this option is the identification of characteristics of the inhabitants that should be considered in future interventions designed specifically for these studied urban units or similar ones. In terms of the second stage, the results presented in Table 3 show a higher participation of the inhabitants of informal and transition units compared with formal unit. Women were more likely to visit the nutrition team than men. As the time required to complete this stage was around 30-40 minutes, including going from their house to the nutritional status evaluation site, it is logical that unemployed participants, students and other workers visited the nutrition team more, compared to participants who reported to be working. Another important variable with a particular meaning, was BMI based on self-reported measures which was associated with adherence to the second stage. This association seemed to be stronger than the association between adherence and chronic conditions. This result may suggest a certain awareness of body image issues, probably related to concern about health status. In the first stage , a discrepancy between the percentage of female and male was observed. Perhaps, in the sampling process to locate a man more contacts are needed or men of these urban units do not participate as much as women in health studies. Another study on Santiago Island [64] found even more discrepant values 68.7% vs 31.3%. Conditional on participation in the first stage, in a second stage , the participation was 75.1% for women and 24.9% for men. As the second stage required participant time, and our teams worked mostly during the day, this difference may also be due to a lower employment rate for women. However, the literature indicates that women are more likely to participate in health studies . Nevertheless, there are variations across different contexts, specifically, in Africa, Ati el al [37] highlighted the same trend in a obesity study. The migratory condition of the Cape Verdeans constitutes another potential explanatory factor. Given the bias of the second stage, in which BMI based on self-reported measures influenced participation, it is expected that overall adiposity and abdominal adiposity would be overestimated. On the other hand, BMI calculated from self-reported measures leads to underestimation of overweight and obesity because women observed in our study, despite reporting their weight correctly, tend to overestimate their height. This is a common issue reported in literature, varying across different groups [66,67]. In spite of these findings, there is a scarcity of data on African context. According to Puonane et al [68], African women saw themselves less obese compared to Caucasian women. Others studies about the perception of height, weight, and overweight and obesity in different ethnic groups, collected from the SUNSET Study, showed that participants from African origin presented the most negative perceptions [66,69]. In spite of the aforementioned limitations, through a mix of quantitative and qualitative research methods, this study adds important findings regarding an understudied African country and provides new insights for development of future research. --- Conclusions Our findings indicated that residents of the informal unit and women are the most disadvantaged groups in terms of health outcomes and socioeconomic variables. Focus group discussions and the intensive fieldwork reinforced the higher participation of residents of the informal unit and women in all stages, suggesting the practicability of health promotion campaigns, taking into account the potential of social capital of the informal settlements and the role of the woman in family and society in Cape Verde. The qualitative study also helped us to understand how this particular context can be used to involve women in health promotion in their neighbourhoods, taking into account their decision-making power, their multi-dimensional role in purchasing, processing and preparing food as the pillar of familial food security and also their contributing via non-formal economic activities for their families. On the other hand, the built environment and insecurity of the neighbourhood/urban unit are also modifiable factors in order to promote healthy lifestyles . The role of urban planners and developers, in planning and/or modifying the built environment, to improve populations' health and reduce harmful exposures is still not well established in Africa. The complexity and specificity of the dynamic of each city highlight the importance of local studies to bring informal settlements into health statistics and to promote better health in these poorest communities. The link between health planning, urban planning and security of the city needs to be reinforced to minimize health, social and gender inequalities. --- Due to ethical restrictions imposed by ethical committees, data are available upon request from the UHPI-Stat project. To request the data, readers should contact the principal investigator, Luzia Gonçalves ---

The lack of high-quality data to support evidence-based policies continues to be a concern in African cities, which present marked social, economic and cultural disparities that may differently impact the health of the groups living in different urban contexts. This study explores three urban units-formal, transition and informal-of the capital of Cape Verde, in terms of overweight/obesity, cardiometabolic risk, physical activity and other aspects related to the urban environment.Quantitative and qualitative research methods were used in this intra-urban study. A proportional stratified random sample (n = 1912 adults), based on geographical coordinates of private households, was selected to apply the UPHI-STAT questionnaire. In a second stage (n = 599), local nutritionists collected anthropometric measurements (e.g., height, waist circumference) and body composition by bioelectric impedance (e.g., body weight, body fat, muscle mass). In a third stage, pedometers were used to count study participants' steps on working and non-working days for one week (n = 118). After a preliminary statistical analysis, a qualitative study was developed to complement the quantitative approach. Generalized linear models, among others, were used in the multivariate analysis.

Introduction Adverse childhood experiences , which include forms of abuse, neglect, and household dysfunction , have been shown to influence academic achievement, health behaviours , and various health conditions throughout the life course . Setting the stage for lifelong health and social outcomes, ACEs merit attention both from the moral perspectives of child welfare and health justice , and from perspectives of life course health promotion and societal prosperity . On average, children living in low-income households tend to experience a greater number of ACEs than their higher-income peers . In Scotland, where ACE prevalence is high [i.e. 65% of children experience one or more ACEs by the age of 8 years ], 53% of children in the highest income households are ACE-free by age eight, compared to 8% in the lowest income households . These inequities beg the question of whether the relationship between low-income and ACE incidence can be mitigated. A growing body of theoretical and empirical work suggests that the experience of low-income, and its association with health outcomes, can vary according to the relative generosity of state investment in benefits, social policies, and resources. For example, the generosity of unemployment benefits is known to influence psychological distress and self-reported health among the unemployed. It is possible, therefore, that the association between low-income and child adversity could be mitigated by protective social, economic, or infrastructural resources. The assumed modifiability of poverty experience-in terms of its impact on childhood adversity-underpins many of the UK and Scotland's policy recommendations for investment in early childhood development, as a means of reducing health inequities across the life course. Reports published between 1980 and 2014 all discuss promoting access to adequate and affordable housing, transportation, childcare services, and recreational opportunities to ensure children are assured the ''best start in life'' . These features of local environments have been identified as upstream causes of health inequalities, and interventions on the latter domains are believed to be among the most effective at reducing life course health inequities . The objectives of this study were therefore to assess whether certain community resources are associated with lower ACE incidence in households above and below the poverty line and to assess the extent to which income inequalities in 8-year cumulative ACE incidence could be eliminated if all had access to these community resources. --- Methods This study used data from the Growing Up in Scotland birth cohort, which followed children born in 2004/2005 yearly, for seven sweeps of data collection, until 2013/2014 . With a sampling frame based on Scotland's universal Child Benefit-described in detail previously -GUS documents the health, behavioural, and social characteristics of a nationally-representative sample of Scottish children. A total of 2816 children who had participated in all seven sweeps of data collection and for whom relevant ACE, income, and covariate data were available were included in this study. --- Measures Outcome measure: adverse childhood experience incidence In GUS, data were available to measure seven experiences that are typically counted when assessing the burden of adverse childhood experiences : child physical abuse or emotional neglect ; mother's experience of domestic violence ; parents' reported use of street drugs or elevate use of alcohol [C 14 units per week ]; separation, divorce, or incarceration; and elevated affective symptoms in the past week [i.e. a score of 36 or higher on the SF-12's mental health component , or of 1 standard deviation above the mean on a restricted Depression, Anxiety and Stress Depression scale ]. Each ACE is described in detail in the Supplement's eMethod 1. Since ACE-related items were asked inconsistently through the study cycles, we measured 8-year cumulative ACE incidence by summing the ACEs present for each child throughout the study period, as has been done previously . Two incidence cut-offs were applied. First, a cut-off of zero versus one or more ACEs was used to capture both a relevant, if ambitious, objective for child welfare policy, and because children with one or more ACEs-although at lower risk for future social and health issues than their high-exposure peers -will likely experience a larger share of the population burden of future outcomes simply because they are more numerous . Second, a higher cut-off was used to reflect findings that multiple, concurrent adversities place children at higher risk of negative health and social outcomes throughout the life course . The three ACE cut-offs were used in lieu of a higher cut-off due to the smaller number of ACEs measured in this study , and the limited number of respondents reporting three or more ACEs by the age of 8 years. --- Exposure measure: low income Total household income after tax and other mandatory deductions, in pounds sterling , was measured at baseline and equivalized according to the number of household members based on their age . To allow for the application of a weighting-based mediation analysis approach, a dichotomized measure of income was used. Household income was dichotomized according to the UK's low-income cut-off value in 2004/2005, of £11,000 per year . --- Mediator measures: access to community resources Five community resources were considered as potential mediators and modifiers of the association between low-income and ACE incidence. These included self-reported access to: non-precarious housing, a local park or play park, transportation services, childcare services, and formal in-person breastfeeding counselling. These factors were selected both because they had been acknowledged by governments of Scotland and the UK as relevant areas of intervention for the promotion of child health and well-being , and because the GUS questionnaire included sufficient items to operationalize each measure. Factors were considered separately both because of the potential heterogeneity in protective pathways they represent, and to facilitate interpretation. All were measured at baseline and were operationalized as dichotomous measures to allow the application of weightingbased mediation analysis approach. Households were considered to have access to nonprecarious housing if, when asked ''In the first 3 months [or at the time of the survey], was there anything […] you found particularly difficult?'' respondents did not identify ''Accommodation or housing problems'' as something that the household found particularly difficult and-following the UK's Office for National Statistic definitions of deprived housing -did not report an absence of central heating, nor residence in a shared dwelling , a caravan, mobile home, or houseboat when asked about their current housing situation. Households reporting any of the latter experiences were considered to not have access to nonprecarious housing. The measure of access to a local park was based on responses to the question ''Is there a public park or play park within 10 min walk of here [your residence]?'' Those who reported ''Yes'' were considered to have access to a local park, while those reporting ''No'' were not. Park-related analyses were restricted to those living in urban settings. Households were considered to have access to transportation services if when asked ''In the first 3 months [or in the last 3 months] how much of a problem was lack of suitable transport?'' they answered, ''Not a problem.'' Those who responded ''Bit of a problem'' or ''A big problem'' to the latter question, as well as those who listed ''I would have transport difficulties getting to a provider'' as the main reason they were not accessing childcare for the child were considered to not have access to suitable transportation services. For the measure of access to in-person breastfeeding counselling, mothers were considered to have access to the latter resource if they responded ''Yes'' to the question ''Did you receive any help or advice about breastfeeding at the time of the child's birth?'' and indicated that the source of the advice was a midwife, health visitor, a professional from the National Childbirth Trust or another voluntary group or organization, or another health professional. Those who reported ''No'' to the question and those who indicated that the source of the advice was from books, magazines, leaflets, friends, or family members were categorized as not having access to formal in-person breastfeeding counselling. Lastly, for the measure of access to childcare services, respondents who answered ''Yes'' to the question ''Do you currently get help with childcare for child on a regular basis?'' and answered ''No'' when asked ''If it was available and you could afford it, would you use a different kind of childcare provider as your main childcare provider for child [e.g. childminder, a Local Authority playgroup, preschool, creche or nursery, a community playgroup or preschool, a workplace creche or nursery, a Family Centre, or a child-carer]?'' were considered to have access to childcare services. Those who responded that they did not get help with childcare on a regular basis, or were receiving help but answered that if alternative affordable childcare services were available, that they would use them, were considered to not have access to childcare services. --- Covariates Analyses were conditioned on factors that were assumed to be confounders of the associations between low income, the five community resources, and cumulative ACE incidence . Covariates included mother's or stepmother's ethnicity , age at birth of first child , employment during pregnancy , area of residence , level of education , and the child's sex . Descriptive analyses were performed on the sample's characteristics . We used income-stratified identity-link Poisson regression models to assess the associations between resources and ACE incidence across income groups. These models were weighted using the product of individuals' inverse probability weight for low-income exposure and GUS longitudinal weight-which accounted for selection at baseline and non-response until the final survey sweep . These IPW ensured that income groups were balanced in terms of the measured covariates. Details on the construction of IPW are described in the Supplement's eMethods 2. Then, to assess the extent to which income inequalities in ACE incidence could be eliminated if all had access to any of the five resources, we estimated a metric known as the ''Proportion Eliminated'' in the mediation literature . The PE is estimated by first taking the difference between the total association between low income and ACE incidence , and the association between low income and ACE incidence that would remain if all had the mediating resource ; this difference is then divided by the TE [i.e. PE = /TE] . For TE and CDE estimation, we used an IPW-weighted identity-link Poisson regression-based approach-the theory behind which has been described previously . For the TE estimate, we regressed ACE incidence on income-weighting the model using the product of individuals' IPW for low-income exposure and GUS longitudinal weight. For the CDE estimate, we regressed ACE incidence on income, resource absence, and the product between the latter . The latter model was weighted using the three-way product of individuals' IPWs for low-income exposure and for resource absence, and GUS longitudinal weight. The coefficient associated with low-income and resource presence captures the income inequality in ACE incidence that remains if all had access to the resource . All confidence intervals for the TE, CDE, and PE were estimated using the bootstrap method . Additionally, three sensitivity analyses were conducted. First, to assess whether the associations between low-income and the incidence of each ACE were relatively homogeneous, we specified IPW-weighted models for the incidence of each ACE, separately. Second, we applied VanderWeele's method to assess how large the associations would have to be between an unmeasured factor and both the resource and ACE incidence for the true CDE to be null, despite non-null estimates . Lastly, since the use of IPW relies on the assumption of practical positivity [i.e. that propensity scores used to estimate IPW are neither 0 nor 1 ], we performed descriptive analyses of all estimated propensity scores . All analyses were completed using R, version 3.4.1 . --- Results --- Baseline characteristics Overall, a greater proportion of mothers with the following characteristics were part of low-income households: those who were not white; had not completed Highers ; had not been working at any point during their pregnancy; had their first child before the age of 20 years; lived in urban areas; had not been offered breastfeeding counselling; and had difficulties accessing transportation or childcare services . --- Cumulative ACE incidence By 8 years, approximately 57% of children in the sample had experienced at least one ACE, and 9% had experienced three or more ACEs . The proportion of children who had experienced no ACEs was 12% among those living in households below the poverty line, and 47% in households above the poverty line . Significant associations between low income and each of the adverse childhood experiences were observed . Overall, cumulative ACE incidence was lower among those who reported having access to housing, transportation, and breastfeeding counselling . --- Protective role of community resources above and below the poverty line Above the poverty line, access to housing, transportation, and breastfeeding education was associated with lower ACE incidence . The latter resources were associated with a 10-, 14- and 9-percentage point lower incidence of 1 or --- Proportion of income-based ACE inequality eliminated TE, CDE, and PE estimates for each resource are summarized in Table 3 . With the higher ACE cut-off , only transportation was associated with a statistically significant PE value. If all families had access to transportation, 21% of the income inequality in higher ACE incidence could be eliminated . With the lower ACE cut-off , all PE confidence bounds crossed the null value, except those for housing . The PE pertaining to housing was negative , indicating that if all had access to housing, the income-based inequality in cumulative incidence of 1 or more ACEs may in fact increase. This finding is in line with findings from the above-stratified analyses: housing is associated with a 10-percentage-point lower incidence of 1 or more ACEs among wealthier households and a null association among low-income households . --- Sensitivity analyses Analyses suggest that the associations between an unmeasured factor and both transportation access and ACE incidence , and between an unmeasured factor and both housing and ACE incidence , would -11.0 -13.9 -8.0 -3.9 3.9 -2.6 4.5 0.4 RD cumulative incidence risk difference, expressed as percentage-point difference; CI confidence interval a Cumulative incidence risk differences are weighted for mother or stepmother's ethnicity, age at birth of the child, employment status at during pregnancy, area of residence, level of educational attainment, and child's sex b Data are restricted to those living in urban settings have to be of RR = 3.0 to explain away the observed CDE estimates . RR = 3.0 is larger than the observed association between income and incidence of either 1 or 3 or more ACEs . Lastly, analyses of propensity scores indicate strong covariate balance after weighting, and of practical positivity for both exposure to low income and exposure to the resources measured . --- Discussion In a population-based sample of Scottish children, we found that access to housing, transportation, and breastfeeding education were associated with lower ACE incidence among households above the poverty line. Below the poverty line, only transportation access was associated with lower ACE incidence. We estimated that if access to transportation was held fixed across the entire population, approximately 21% of the income-based inequality in cumulative incidence of 3 or more ACEs could be eliminated. Our finding of the protective role of transportation is in line with those of previous studies. Transportation resources can mitigate the association between distance to health or social services and service utilization and enable families in accessing employment, food, and leisure facilities . That access to housing and breastfeeding counselling was also associated with lower ACE incidence-at least among higher-income households-is also consistent with observations in the extant literature. Housing is known to influence general family well-being by shaping residents' mental health, sense of self-worth, and offering a stable environment from which to pursue training, employment, and parenting responsibilities . Breastfeeding resources may represent a proxy for the density of other protective health resources, such as access to nurses. Perinatal nurse visits are known to be beneficial for mothers' health and well-being . Several theories may explain the observed association between transportation and lower ACE incidence. Transportation may help parents' to gain a sense of control over their lives-offering opportunities for decisional latitude regarding work and daily activities . A second theory is that inadequate transportation is itself a stressor for low-income families. Walking as a compulsory mode of transport in disadvantaged areas is often accompanied by physical fatigue and psychosocial stress , which may lead to greater household dysfunction. Third, transportation resources allow families to access a wealth of protective resources beyond their local area . Lastly, it may be that transportation resources How community resources mitigate the association between household poverty and the incidence… enable parents-specifically mothers-to remove themselves and their children from adverse situations, thus protecting their children from ACEs . These theorized mechanisms merit attention in future work. In contrast, there are several potential explanations why statistically significant associations between parks, childcare, and ACE incidence were not observed. First, it may be that park proximity does not guarantee park use or quality . Parks in low-income areas in the UK tend to be of lesser quality than those in higher-income areas . Similarly, childcare may not be protective if its quality is low [a well-documented issue for poor Scottish families ] or if the parents cannot afford a critical ''dose'' of childcare [i.e. several half-days weekly from ages 2-4 years ]. Second, it may be that these two factors play a more important protective role later in the child's life than at the time at which they were measured . Future studies on these more nuanced elements are recommended. This study requires replication in other settings before strong recommendations can be made. If other studies confirm protective effects of similar service features, specifically for ACE prevention, intervening on community-level resource availability could represent a valuable interim measure to mitigate the association between childhood poverty on ACEs. While obviously second best to the complete elimination of poverty, local measures to promote resources for families may hold promise in settings lacking the political will for formal wealth redistribution. These study's findings are bound by certain limitations. First, baseline marital status was not included as a covariate due to its collinearity with the dichotomized exposure measure of household income. Approximately, 80% of single mothers were below the poverty line. This exclusion and other unmeasured factors may have contributed to residual confounding in the study. Factors that were not included in the study, primarily due to limitations in data availability, were paternal characteristics, parents' own histories of adverse childhood experiences or behaviours . Though sensitivity analyses suggest that the present study's estimates are likely robust to unmeasured confounding, we recommend that, if available, future studies adopt a life course exposure perspective and consider these additional sources of bias as well as potential time-varying confounding. A second limitation pertains to the ACEs measured. Though we attempted to maximize the comparability of the ACEs measured in this study with those of previous studies , missing in this study are ACEs of material neglect, sexual abuse, and emotional abuse , as well as ACEs that might occur after age 8. These exclusions may lead to potential underestimation of childhood ACE incidence in the Scottish population. ACE incidence estimates may also be affected by higher observed attrition rates in the GUS cohort among younger mothers and households living in deprived areas . In this study, we sought to minimize this potential source of bias using the GUS longitudinal survey weights . Furthermore, though available ACE measures were compiled here to form one single outcome measure of cumulative ACE incidence-in order to both accommodate data availability and account for the observation of the detrimental impact of concurrent adversities on health and social outcomes -this outcome operationalization can also lead to the interpretation of the moral or sociopolitical equivalency of each experience. The potential stigmatizing effect of this grouping of experiences is a limitation of both this study and its predecessors and may be a relevant topic of future enquiry. A third limitation is the potential measurement bias associated with self-reported data. Previous studies have noted differential self-reporting of neighbourhood characteristics across socioeconomic groups, wherein populations living in more deprived contexts under-report deprivation experiences, in part out of a need for self-preservation . It is possible, therefore, that income inequalities in resource access may have been underestimated in this study. To address these concerns, future studies may benefit from using multi-item indices to measure resource access . Lastly, the IPW-weighted models used rely on several operational assumptions. Where possible, we attempted to assess the potential sensitivity of findings to violations of these assumptions. Importantly, our findings appear robust to potential unmeasured confounding, and violations of practical positivity. In conclusion, this is, to our knowledge, the first study to explore how community resources may mitigate the association between household poverty and cumulative ACE incidence. Of the resources assessed, only transportation was associated with lower ACE incidence in households above and below the poverty line. We estimated that a substantial portion of the income-based inequality in the 8-year cumulative incidence of three or more ACEs could be eliminated if all had access to adequate transportation. Though our findings require replication, they offer an initial body of evidence that can inform interventions to prevent both ACE incidence and income disparities in ACE incidence. ---

To assess what proportion of the association between household low income and incidence of adverse childhood experiences (ACE) would be eliminated if all households had access to housing, transportation and childcare services, breastfeeding counselling, and parks. Methods Using Growing Up in Scotland birth cohort data (N = 2816), an inverse probability-weighted regression-based mediation technique was applied to assess associations between low-income status (\ £11,000 in 2004/5), resource access, and cumulative 8-year ACE incidence (C 1, C 3 ACEs). Resource access was measured based on households' selfreported difficulties (yes/no) in accessing housing, transportation, childcare, and breastfeeding counselling, and park proximity (within 10 min from the residence).The protective effects of resources were heterogeneous. Only access to transportation was associated with lower ACE incidence in both low-and higher-income households. If all had access to transportation, 21% (95% CI 3%, 41%) of the income-based inequality in incidence of 3 or more ACEs could be eliminated. Conclusions While second best to the elimination of child poverty, measures to improve families' access to community resources such as transportation may mitigate the effects of poverty on ACE incidence.

INTRODUCTION In developed countries strong inverse associations between socio-economic indicators and obesity in adulthood have been demonstrated. Cross-sectional evidence continuously shows higher obesity prevalence in disadvantaged adults compared with those who are more affluent [1][2][3][4] although these associations appear to be strongest in women. Longitudinal studies also show that weight gain over time and risk of obesity in later life is greatest in those of lowest socio-economic position. [4][5][6] There is also recent evidence to suggest that the disparity in obesity between those of the highest and those of the lowest socioeconomic position may continue to widen in the future. 7 Reducing socio-economic inequalities in health, and risk factors for disease, is a major public health concern; however, in terms of obesity there is a distinct lack of evidence of the types of intervention that are effective in reducing such inequalities. [8][9][10] This article aims to systematically review the best available evidence of the effectiveness of interventions in reducing socio-economic inequalities in obesity among adults. --- How interventions can impact on inequalities in obesity Interventions can be characterized by their level of action and their approach to tackle inequalities. Whitehead 11 describes four levels of interventions to tackle inequalities: strengthening individuals , strengthening communities , improving living and work environments and promoting healthy macro-policy . According to Graham and Kelly, 13 these interventions are underpinned by one of the three different approaches to health inequality: disadvantage , gap , or gradient ) Interventions are thus either targeted or universal . The aim of this review was to systematically examine the effectiveness of interventions operating via different approaches in reducing socio-economic inequalities in obesity among adults. A companion paper examines interventions for reducing socioeconomic inequalities in obesity among children . --- MATERIALS AND METHODS The review was carried out following established criteria for the good conduct and reporting of systematic reviews. 14,15 The full review protocol is published elsewhere 16 and is registered with the PROSPERO International Prospective Register of Systematic Reviews . The full review is available to view at http://www.phr. nihr.ac.uk/funded_projects/obesity.asp. 17 --- Data sources The following electronic databases were searched from the start date up to the 11 October 2012: MEDLINE, EMBASE, CINAHL, PsycINFO, Social Science Citation Index, ASSIA, IBSS, Sociological Abstracts and the NHS Economic Evaluation Database . We did not exclude papers on the basis of language, country or publication date. The electronic database searches were supplemented with website and grey literature searches. --- Types of intervention Our review examined interventions at the individual, community and societal level which might reduce inequalities in obesity among adults . We defined individual level interventions as those that included individualized/one-toone health promotion, education, advice, counselling or subsidy and were conducted in a health care or research setting, or in participant's homes. Community level interventions were defined as group-based health promotion, education, advice, counselling or subsidy only interventions, or interventions conducted in a community setting . Societal level studies were split into two sub-groups: Societal-environment level interventions, which were defined as those that included a change in environment or access to environment; and Societal-policy level interventions as macro-level policies such as taxation, advertising restriction or subsidies. Interventions were also classified in terms of whether they took a gradient approach and included participants of all socio-economic status or a targeted approach, that is, aimed at low-SES participants only . Measures and proxy measures of SES were income, education, occupation or area level disadvantage. Interventions that involved drugs or surgery, and laboratorybased studies, were excluded from the review. Our review considered strategies, which might reduce existing inequalities in the prevalence of obesity , as well as those interventions, which might prevent the development of inequalities in obesity . --- Types of studies Our review included randomized and non-randomized controlled trials that included either a non-treatment control group or standard treatment group, and prospective and retrospective cohort studies, with or without control/standard treatment groups, and prospective repeat cross-sectional studies with or without control/standard treatment groups . Only studies with duration of at least 12 weeks were included. For the purpose of this article only the best evidence available for each intervention level is reported; however, this included randomized and non-randomized controlled trials as well as uncontrolled prospective cohort studies . --- Types of outcome measure Studies were included if they reported a primary outcome that is a proxy for body fatness and if they examined differential effects with regard to socio-economic status or were targeted specifically at disadvantaged groups or were conducted in deprived areas. --- Data extraction and quality appraisal The initial screening of titles and abstracts was conducted by one reviewer with a random 10% of the sample checked by a second reviewer. Data extraction was conducted by one reviewer using established data extraction forms and independently checked by a second reviewer. The methodological quality of the included studies was appraised independently by two reviewers using the Cochrane Public Health Review Group recommended Effective Public Health Practice Project Quality Assessment Tool for Quantitative Studies. 18 Any discrepancies were resolved through discussion between the authors and, if consensus was not reached, with the project lead. --- Analysis and synthesis Our full review 17 used very broad study inclusion criteria and conducted a very wide search to capture the entire evidence based on the effects of interventions to reduce inequalities in obesity among adults. We examined the full papers of all studies which fit our population, intervention, design and outcome inclusion criteria, even if there is no mention of socio-economic inequalities in the abstract. By adopting this strategy we were less likely to exclude studies which undertook subgroup analyses by socio-economic status but did not publish the findings in the abstract. This resulted in a very large evidence base that was much larger than anticipated. To make sense of it for policy and practice, this article focuses only on a narrative synthesis of the 'best available' international evidence for each intervention type. Best available evidence was defined in terms of both study design and study quality by each intervention type. We considered experimental study designs as providing better evidence than observational study designs , and 'strong' quality studies as providing the best evidence, followed by 'moderate' and then 'weak' quality studies. Only those studies that provided the highest quality evidence for each intervention type are synthesized in this paper. --- RESULTS Our database searches indentified 70 730 records. After title and abstract screening 3142 papers were retreived. Supplementary searching revealed four additional studies that met the inclusion criteria. After full paper screening, the 'best available' evidence for each intervention level was obtained from 20 studies . For the individual and community level interventions, the 'best available' evidence is provided by strong quality, experimental studies . For the societal-environmental level interventions a moderate quality experimental study was the strongest identified. For the societal-macro-policy interventions only weak quality observational studies were located and these therefore provide the 'best available' international evidence. The descriptions and findings of the 'best available' evidence studies are summarized in Tables 123. Overall, this shows that the various interventions either reduced inequalities in obesity or had no effect, with no studies reporting a negative impact . Summaries of each study and their overall impact on SES inequalities in obesity are reported in the tables. Individual Four of the five 'best evidence' individual level studies were from the USA, with the remaining study from the UK. All of the studies had all women or majority women participant groups. Four of the studies examined tailored weight loss programmes delivered via primary care for low-income groups . One was a randomized controlled trial 19 of 106 lowincome African-American women that found that after a 6 month intervention, the intervention group demonstrated a significant weight loss compared with the control group . A small randomized controlled pilot study of overweight post-partum women living in areas of moderate to high deprivation in the UK 20 found that after a 12-week intervention, body weight loss was significantly greater in the intervention compared with the comparison group , with significant improvements in BMI and percentage body fat too. There were no significant differences in waist circumference or physical activity. A cluster randomized controlled pilot study 21 investigated the effects of a 6 month tailored weight loss programme for 51 low-income adults considered at high risk of diabetes in four areas of the USA. Post intervention, 25% of the intervention group achieved a clinically significant weight loss compared with only 11% of the control group. However, physical activity and nutritional changes were similar among control and intervention groups. Intention-to-treat analysis from another randomized controlled trial 22 investigating the effects of a 6 month tailored weight loss intervention showed that after 9 months , weight loss among 86 low-income African-American women was significantly higher in the intervention group than the control although this was not sustained at 12 or 18 months . One randomized controlled trial 23,24 investigated the effects of a 3 year weight gain prevention educational intervention among low-and high-income women in the USA . After 1 year the intervention improved weight among high-income women only, but after 3 years, there was no significant intervention effect on weight for either high or low-income groups. --- Community The vast majority of the 'best evidence' community level studies were from the USA and all except three studies had only women participants or a female majority, some had very small sample sizes as they were pilots and many had active controls . A randomized controlled trial conducted 25 compared the effects on weight loss of a professionally delivered behavioural therapy active control with a 4 month behavioural therapy and social support intervention among 136 low-income men and women in the USA. The intervention showed short-term weight loss but this was not sustained at 7 and 10 month follow-up . A small randomized controlled pilot study 26 investigated the effects of a 16 week culturallyadapted weekly behavioural weight loss group programme with motivational interviewing vs a behavioural weight loss group programme plus health education in 33 lower income African-American women . There were significant positive effects post-intervention on weight loss, calorie intake, percentage calories from fat and fruit and vegetable servings per day in both groups. Together these findings suggest that such interventions have short-term but not long term positive effects on weight loss. The three studies that examined group lifestyle counselling interventions compared with active controls were conducted among low-SES populations . One found a short-term effect on weight loss among low-income Latino women while the other two found no intervention effect. A randomized controlled trial 27 evaluated a 1 year communitybased, culturally tailored group lifestyle counselling intervention among 288 low-income Latinos in the USA. The intervention group lost significantly more weight postintervention compared with the control group participants with a significant decrease in BMI . A randomized controlled trial 28 in a lowincome urban area of Australia examined the intervention effects on 149 participants of a one-off group lifestyle counselling session on the reduction of cardiovascular risk factors among those at high risk . The control group received a pamphlet about reducing cardiovascular risk factors. After 6 months post-intervention, there were no significant weight changes either between or within groups. A randomized controlled trial 29 investigated a 12 month extended care group counselling intervention compared with an education-only control condition following a weight loss intervention for 224 obese women living in low-SES areas of the USA. The interventions had no overall effect on weight regain compared with the control group. The three studies that evaluated community-based groupbased health education interventions also followed targeted approaches. A small randomized controlled pilot study 30,31 compared a 20 week community centre delivered, culturallyadapted group nutritional education intervention against individual-counselling on weight loss among 19 low-income Mexican-American women in the USA. There were no significant differences in weight loss in the intervention group compared with the control after post-intervention (2.9 kg vs 1.3 kg, Prevention or treatment intervention . c Disadvantage/gradient approach to inequality . d Po 0.05.This is the relative mean differences between intervention and control at the longest follow-up. e +, positive intervention effect so it reduces obesity-related outcomes in low-SES groups or reduces the SES gradient in obesity-related outcomes; 0, no intervention effect or no effect on SES gradient in obesity-related outcomes. Reducing inequalities in adult obesity FC Hillier-Brown et al P = 0.47). A randomized controlled trial 32 investigated the effects of a 3-month peer-educator led group-based dietary skills development intervention for 239 obese low-income African-American women in the USA. The control group received a selfhelp workbook and a half-day workshop. There were no significant differences between intervention and control groups for changes in BMI or weight post-intervention. A randomized controlled trial 33 examined the effects of a 25-day group-based health education intervention among 400 women from a low-income area of Turkey. The intervention group reported significant changes in lifestyle behaviours leading to reductions in BMI and an increase in those with an ideal BMI after 6 months . The final community-based group health education and counselling intervention study was a small randomized controlled trial that took a universal approach. 34 It compared the effects by social class status of weekly community diet clubs in the UK that used two different weight reducing diets: low carbohydrate compared with low fat/high fibre. Moderate weight losses were observed among 119 participants in both diet groups after 3 months but there was no significant difference at 1 year follow-up. Both diets worked equally well among the higher SES participants . Among the lower SES participants the low carbohydrate diet was particularly effective at 3 months but this effect was lost at 1 year. Two community level studies examined workplace delivered group interventions with divergent results. One study of blue collar male workers in the USA found positive long term effects of an extensive and complex health promotion, counselling and physical activity intervention on weight loss and weight maintenance. A study of blue collar women employees found no effect on BMI of a lay health advisor programme over 5 years. A cluster randomized trial 35 investigated the effects of four 3 year wellbeing interventions among 690 overweight , blue collar workers across four manufacturing worksites in the USA. After the 3 years, there was a significant intervention effect as the control group gained weight, intervention groups A and B experienced no weight change and intervention group C lost weight . Another cluster randomized controlled trial conducted by Campbell et al. 36 examined a 5-year workplace health promotion programme among 859 low skill women workers in the USA. The intervention consisted of two strategies: individualized computer-tailored health messages; and a lay health advisors programme. Data were collected at 6 and 18 months during the programme and there were no significant changes in BMI in either of the study groups at either time points. The final two community level studies were conducted in the USA and examined family based group education interventions that were delivered in school/classroom setting . One was a randomized controlled trial 37 that investigated the effects of a 12-week intensive classroom-based physical activity intervention compared with a less intensive activity control group among 92 daughter and mother dyads in a low-income Latino community in the USA. There were no significant differences in maternal BMI at the end of the 12-week intervention. The other was a randomized controlled study 38 to test the effectiveness of a professionally delivered, tailored group education intervention to prevent heart disease in 242 adults in the USA with low-income and low-literacy levels ). It comprised a 6-week, tailored classroom-based intervention followed by a 12-week maintenance intervention. It was compared with nontailored general nutritional group interventions. There were no significant changes in BMI in SNAP compared with the control after either of the intervention periods. Societal The 'best available' evidence for the environmental interventions comes from one moderate quality experimental study that took a universal approach and examined a multi-faceted workplace weight prevention intervention. The 'best available' evidence for the macro-level interventions comes from two weak quality observational studies that took a targeted approach and examined effects of the USA Food Stamp Programme, a welfare programme for people with a low-income or with no-income. All three studies were from the USA and two included only women participants. The environmental intervention study was a cluster randomized controlled trial of 648 participants 39 in the USA that investigated the effects of a 2-year multi-faceted workplace weight prevention intervention on hospital employees. The intervention included a social marketing campaign, interpersonal support groups and environmental strategies to promote physical activity and healthy eating. This included stairway signs, cafeteria signs, farmer's markets, walking groups, challenges, workshops, educational displays, newsletters, project website, project information centre and print materials. The study found that, postintervention participation reduced BMI. However, higher educated participants more likely to prevent weight gain than lower educated participants and so the intervention did not reduce inequalities. One of the macro-policy intervention studies was a retrospective cohort study 40 of 5503 USA women that found that a $2000 increase in food stamps had no significant effect on weight change unless women were persistently food insecure whereby a $2000 increase was associated with a significant increase in weight . A retrospective repeat cross-sectional study 41 of low-income men and women participants however found no significant effects of the USA Food Stamp Programme on obesity. --- DISCUSSION What works in reducing inequalities in obesity? for whom? and where? Evidence of the effectiveness of individual level interventions to reduce inequalities in obesity among adults was only found for tailored weight loss programmes targeted at low-income groups particularly those delivered in primary care settings. These appear to have positive effects on weight outcomes in the short term . Similarly, we only found evidence of effectiveness for community-based behavioural weight loss interventions and community diet clubs . However, again these positive effects on obesity-related outcomes were only short term in nature with no longer term positive effects evident. This is in keeping with previous research into the general effectiveness of obesity-reduction interventions, which has also found short-term benefits of interventions with weight regain in the longer term. 42,43 The community level workplace studies suggested that longer term positive effects on obesityrelated outcomes require more complex, multi-faceted interventions. This is in keeping with the Foresight obesity review which highlighted the complex multi-factorial nature of inequalities in obesity and therefore the need for sophisticated and longer term interventions to reduce them. 44 However, in our review, only limited and weak quality evidence was found of more upstream interventions and the results were inconclusive over the longer term. It is important to also reflect on 'for whom' and 'where' the interventions were-or more usually were not -effective. The 'best available' international evidence was typically of interventions in the USA perhaps reflecting their more sophisticated approach to experimental design. The 'best available' evidence was also populated by studies with exclusively or majority female participant groups. This is in keeping with the results of general obesity studies where only 10-30% of participants in weight loss programmes are men. 45 Often the interventions were targeted at Afro-American or Latin American women. The findings of effectiveness are therefore very much limited to effectiveness among low-income women in the USA. In terms of 'where' interventions were effective then the 'best available' evidence suggested that primary care delivered interventions and those based in community settings were the more effective. --- Implications for Research The nature of the evidence base has a number of implications for public health researchers. Most notably, while we found a very large international evidence base, the quality of the evidence found was largely observational and of moderate to weak quality. These studies did not generally contribute to the 'best available evidence' . There were few studies of societal-level interventions which might be expected to have more of an impact on the gradient in obesity. 16 This was particularly the case in terms of the macro-policy level and the few studies that were found did not necessarily have obesity as their main outcome or indeed motivation behind the intervention . Similarly, the majority of interventions that were evaluated took a targeted approach to reducing SES inequalities in obesity, with only a minority of studies examining the effects of interventions across the SES gradient. The targeted approach has limitations as even when interventions are effective among lowincome groups they are only able to reduce the health inequalities gap, they have little effect on the wider social gradient. Studies were almost exclusively based on women. The findings of effectiveness are therefore limited to women given that weight loss is embedded in socio-cultural contexts, including those relating to gender. 46 The focus on women, however, may be explained by the stronger associations between SES and obesity observed in woman. 1,4 We also found no studies that assessed the cost-effectiveness of interventions and meta-analysis could not be conducted given the heterogeneity of the studies. Our results show that there is a clear need for more experimental studies of the effectiveness and cost-effectiveness of interventions to reduce inequalities in obesity among both men and women, and especially in terms of macro-level interventions that potentially address the entire gradient. Implications for Public Health Our review has found a large international evidence base but only limited effectiveness of interventions with the potential to reduce SES inequalities in obesity among adults. Most notably, primary care delivered tailored weight loss programmes targeted at individuals from low-income groups and community-based behavioural weight loss interventions and community diet clubs appeared to have some effectiveness-at least in the short term-among low-income women. These interventions may therefore be worth commissioning by those who wish to target services at low-income women or at women in deprived areas. However, to be effective in the longer term, such interventions will need to be of a longer duration and supplemented with subsequent weight maintenance interventions. They may also need to be adapted to be effective among men. --- Strengths and Limitations This review included an extensive and thorough search of the literature, along with a broad inclusion and exclusion criteria, in order to find the best available evidence at each level of intervention. To ensure reliability, quality check procedures were carried out that included double screening and checking by independent researchers at the title, abstract and full paper screening, data extraction and quality appraisal stages. However, there are also some limitations of this review. We found that the quality appraisal tool, although designed for appraising public health interventions, tended to produce stronger quality scores for those studies following a more clinical model and that some questions were unhelpful. The definitions used for each level of intervention resulted in the majority of interventions being categorized as community level. We also found that some complex interventions containing different elements were difficult to categorize, although we tried to be consistent in our assignment. For this review we did not use ethnicity alone as a proxy measure for SES; therefore, we may have excluded some important studies that did so , particularly those that explored differential effects by ethnic group. --- CONCLUSION Our review found a large international evidence base in which the 20 better quality studies suggested that individual, community and societal-level interventions which aim to prevent, reduce or manage obesity at least do not increase inequalities; some universal interventions reduced the gradient in obesity, and many targeted interventions were effective in decreasing obesity among lower socio-economic groups. Among adults, primary care delivered tailored weight loss programmes targeted at individuals from low-income groups and community-based behavioural weight loss interventions and community diet clubs appeared to have evidence of effectiveness-at least in the short term-among low-income women. More research is needed to determine long term effectiveness, as well as more studies among male populations and of macro-level interventions. contributed to analysis, synthesis and interpretation. J-MC-N contributed to data collection and synthesis. AK contributed to analysis. HJM designed and conducted the searches and contributed to data collection. CDS provided methodological, analysis and synthesis input. All authors contributed to revised successive drafts of this article, and approved the final version and submitted for publication. --- CONFLICT OF INTEREST The authors declare no conflict of interest. ---

BACKGROUND: Socioeconomic inequalities in obesity are well established in high-income countries. There is a lack of evidence of the types of intervention that are effective in reducing these inequalities among adults. OBJECTIVES: To systematically review studies of the effectiveness of individual, community and societal interventions in reducing socio-economic inequalities in obesity among adults. METHODS: Nine electronic databases were searched from start date to October 2012 along with website and grey literature searches. The review examined the best available international evidence (both experimental and observational) of interventions at an individual, community and societal level that might reduce inequalities in obesity among adults (aged 18 years or over) in any setting and country. Studies were included if they reported a body fatness-related outcome and if they included a measure of socioeconomic status. Data extraction and quality appraisal were conducted using established mechanisms and narrative synthesis was conducted.The 'best available' international evidence was provided by 20 studies. At the individual level, there was evidence of the effectiveness of primary care delivered tailored weight loss programmes among deprived groups. Community based behavioural weight loss interventions and community diet clubs (including workplace ones) also had some evidence of effectiveness-at least in the short term. Societal level evaluations were few, low quality and inconclusive. Further, there was little evidence of long term effectiveness, and few studies of men or outside the USA. However, there was no evidence to suggest that interventions increase inequalities. CONCLUSIONS: The best available international evidence suggests that some individual and community-based interventions may be effective in reducing socio-economic inequalities in obesity among adults in the short term. Further research is required particularly of more complex, multi-faceted and societal-level interventions.

in the individual life course. Whereas the major theoretical frameworks that have been used to study adjustment to retirement assume that retirement processes are related to experiences in the past, empirical insights regarding earlier life experiences and retirement adjustment are limited. The central question of this study is: To what extent and how can variation in retirement adjustment be explained by earlier life experiences? In the literature on retirement adjustment , several qualitative studies have pointed at the importance of life histories for understanding perceptions of retirement and adjustment . Earlier life experiences in the work, family, leisure, and health spheres seem to be associated with the ease of adjusting to retirement. However, insights regarding earlier life experiences and retirement adjustment based on quantitative studies are scarce. Only few studies explicitly pay attention to more distal life experiences. These studies examine the impact of either work histories or family histories on retirement quality and retirement adjustment problems . This study aims to contribute to the literature on retirement adjustment in three ways. First, compared with earlier studies on life histories and retirement adjustment, we will build to a greater extent on the life course proposition of "multispheral development" . Consistent with this proposition, we will not solely focus on earlier life experiences in one life sphere but simultaneously test the impact of earlier life experiences in the work, health, and family spheres on adjustment difficulties. Given that predictors of retirement adjustment might differ between men and women , we will pay attention to gender as a potential moderator of life history effects on retirement adjustment. Second, the retirement transition involves two developmental challenges: adjustment to the loss of the work role and the development of a satisfactory postretirement lifestyle . Most retirement adjustment studies are based on general measures of psychological comfort such as happiness , morale , life satisfaction , or retirement satisfaction , which do not distinguish between these developmental processes. This study aims to improve our understanding of the first developmental challenge-adjustment to the loss of the work role-by directly asking retirees about the extent to which they miss aspects of work since they retired. Missing work has been studied as a predictor of postretirement morale , satisfaction with retirement , and intentions to unretire , but relatively little is known about the factors that predict this developmental facet of the postretirement process itself. Third, instead of using a general measure of missing work after retirement , we will pay attention to the multidimensional nature of the adjustment process. The loss of the work role might imply multiple changes, such as the loss of income, social contacts, status, daily structure, and purposeful activity. Not only might the ease of adjustment differ across these dimensions but also predictors might differ. As Taylor, Shultz, Spiegel, Morrison, and Greene argue, "a composite criterion that simply combines different dimensions may mask more complex relationships between the predictors of adjustment and particular facets of adjustment" . In this study, we focus on three work-related aspects that retirees might miss after retirement-money/ income, social contacts via work, and status-which resemble the dimensions that Van Solinge andHenkens distinguish in their measures of preretirement anxiety regarding the loss of the work role. Especially when examining the role of earlier life experiences in the adjustment process, it is important to study these dimensions separately, given that the direction of some relationships can be hypothesized to differ between dimensions. This article is based on panel data collected in 2001, 2006-2007, and 2011 among 1,004 Dutch older persons, who were all employed at the first wave of data collection and fully retired within the observation period. Retrospective information on earlier life experiences provides the possibility to study the relationships between life history experiences and adjustment. Resources and retirement transition characteristics-which are established correlates of retirement adjustment -will also be taken into account in the analyses. In the Netherlands, all individuals are covered by a flat-rate basic public pension scheme, and about 91% of employees are covered by earnings-related occupational pension plans in which participation is mandatory. Income replacement rates are relatively high . In recent decades, there has been a strong "early exit culture" in the Netherlands --- Theoretical Background The main theoretical perspectives that have been used to study variation in retirement adjustment are role theory, continuity theory, and the life course perspective . Role theory assumes that the transition into retirement might be especially difficult for individuals who are highly invested in their work role and for whom the work role is central to their self-identity . Continuity theory generally suggests that most adults will be able to achieve positive results adapting to life transitions because during their earlier life they have developed relationships, activities, frameworks of ideas, and adaptive skills that create continuity in their lives when making these transitions . The life course propositions of lifelong and multispheral development imply that specific life periods cannot be understood thoroughly without information on preceding experiences in different life spheres . To integrate these theoretical frameworks, Wang and colleagues propose a resource-based dynamic perspective for studying adjustment to retirement. In this perspective, adjustment is conceptualized as a process, which is dependent on individual resources and changes in resources. The extent to which retirees miss money/income, social contacts, and status can also be expected to be dependent upon the amount of financial and social resources offered by work, changes in these resources due to retirement, and the availability of alternative resources. Moreover, the importance individuals attach to specific work-related resources may play a role. --- Work History In the literature, two main arguments can be found that link work histories to retirement adjustment. Based on a financial argument, it can be expected that employment histories characterized by continuity and upward mobility are positively related to retirement adjustment . Given that pension benefits are dependent upon income and years of service, retirees with these work histories are likely to have an advantageous postretirement financial situation, which might facilitate adjustment to retirement . We therefore hypothesize that retirees who have worked continuously, full-time, or followed an upward career path are less likely to miss the money/income provided by work than those who had a more discontinuous career . Via a nonfinancial argument, adjustment to retirement can be expected to be relatively difficult for retirees who followed a continuous or upward work trajectory. These retirees may be highly attached to their jobs and might have had fewer opportunities to invest in alternative roles over the course of their working life . In that respect, they might perceive the social changes associated with retirement as troublesome. We hypothesize that retirees who have worked continuously, full-time, or followed an upward career path are more likely to miss work-related social contacts and status than those who had a more discontinuous career. --- Health History The health situation of retirees is often found to be an important resource that enables retirement adjustment . Insights regarding the effects of health problems earlier in life are limited though. The experience of severe health problems earlier in life can be expected to increase expenditures and suppress earnings , which might affect retirees' financial situation and adjustment negatively. We hypothesize that retirees who experienced severe health problems in midlife are more likely to miss the money/income provided by work than those who did not experience these health problems . Workers who have had health problems earlier in life might experience more difficulties adjusting to the social dimensions of the retirement transition as well. During midlife, they might have had fewer capacities to develop alternative roles, activities, and relationships next to work compared with those who did not experience health problems. Furthermore, persons in poor health might be less capable of replacing lost relationships and sources of status by new ones, which might make the retirementrelated loss of these social resources relatively difficult. It can be expected that retirees who experienced severe health problems in midlife are more likely to miss work-related social contacts and status than those who did not experience these problems. --- Family History In studies on retirement adjustment, it is generally hypothesized that married retirees experience less adjustment problems than unmarried retirees . The broad categories of whether retirees are married capture, however, a lot of diversity in terms of marital histories, which might be associated with retirement experiences . Individuals who have ever been divorced have been found to have significantly lower wealth in preretirement years than the continuously married group, although remarriage partly offsets the negative divorce effects . A divorce earlier in life might also result in a relatively large drop in terms of income after retirement for the partner that earned the most during the marriage because of pension sharing. We hypothesize that retirees who have ever been divorced-both those who repartnered and those who remained single-are more likely to miss the money/ income provided by work than the continuously married group . Divorces are often accompanied with changes in social networks . Although divorced persons are more involved with friends than persons in their first marriage, divorces negatively affect neighborhood contacts, participation in clubs , and outdoor recreation . Repartnering, however, seems to reverse negative effects of divorce on social integration. Moreover, having a partner can be expected to offer access to relation-specific resources and to offer a stable role or identity . Especially for divorced persons who remained single, therefore, the social contacts and status provided by the work role can be expected to be highly relevant. We hypothesize that divorced retirees without a partner are more likely to miss work-related social contacts and status than those who have continuously been married or repartnered after divorce. --- The Role of Gender In the literature on retirement adjustment, two main arguments can be found on the role of gender . On the one hand, women might experience less difficulties adjusting to the loss of the social dimensions of work than men, given that they have more experience in terms of role transitions and career interruptions, and might be more inclined to perceive the family role as their primary role. On the other hand, it can be expected that women experience more financial adjustment difficulties when leaving the work role compared with men, given that they might be more financially vulnerable due to their more interrupted work careers, employment in secondary labor market positions, and lower likelihood of being married. However, given that the previously discussed life history factors will capture many of these differences between men and women, gender differences in terms of adjustment to the loss of the work role-net of the life history effects-are expected to be limited. It might be the case, however, that the impact of certain earlier life experiences on adjustment differs between men and women. Previous research has shown that the financial status of women in later life is more strongly affected by prior marital dissolution than the financial status of men and persists until remarriage . Therefore, it can be hypothesized that the effect of being divorced and single on missing the money/income provided by work is stronger among women than among men . With respect to the social contacts dimension of missing work, the impact of being single and divorced can, however, be expected to be stronger among men than among women . The experience of a divorce has been found to have a significant positive effect on support from colleagues and acquaintances among men but not among women . This suggests that colleagues are particularly an important source of social support for divorced men, which might make the social changes due to retirement challenging. --- Design and Methods --- Sample The NIDI Work and Retirement Panel data are three-wave panel data collected by the Netherlands Interdisciplinary Demographic Institute. In 2001 , data were collected among a random sample of civil servants aged 50-64 years working for the Dutch central government and all workers aged 50-64 years of three large Dutch multinational private sector organizations . A mail questionnaire was sent to 3,899 older workers; in total, 2,403 questionnaires were completed . In 2006-2007, a follow-up study was carried out among surviving and traceable participants of the first wave. A total of 2,239 questionnaires were mailed out, of which 1,678 were returned . The third round of data collection took place in 2011 among all 1,638 surviving and traceable respondents of the second wave. The Wave 3 questionnaire was returned by 1,276 respondents . The base sample for the analyses consists of 1,080 respondents who shifted from being in paid work at Wave 1 to being fully retired at either Wave 2 or Wave 3. Given that this study focuses on adjustment to retirement, those who did not make use of an retirement arrangement but stopped working because of unemployment or disability were excluded from this base sample. Respondents for whom information on the dependent variables is missing or who did not answer the central questions regarding mid-career experiences were eliminated from the sample. This results in an analytic sample of 1,004 retirees. On average, respondents were retired for 2.5 years when they answered the adjustment questions. --- Measures --- Dependent Variables To measure adjustment to the loss of the work role across dimensions, fully retired respondents were asked during Waves 2 and 3 to report to what extent they miss various aspects of work since they stopped working. We used the responses provided at the study wave immediately following the respondent's full retirement. Missing money/income and missing social contacts via work were both measured by one-item indicators. Missing status was measured by a two-item scale , which was constructed by taking the mean score of items that ask about the extent to which respondents miss self-esteem and prestige/status since they stopped working . Response categories ranged from 1 to 5 but were reversely coded in the analyses. High scale scores indicate that respondents miss the specific work aspect very much. Social contacts are the work-related aspect that respondents are most likely to miss , followed by financial resources and status . In the multivariate analyses, we standardized the dependent variables to obtain effect sizes for the dummy variables. --- Independent Variables To measure continuity of the work career, respondents were asked to indicate the age at which they started working and for how many years in total they have been out of the labor market after that . This information was used to calculate the number of years spent in the labor market at retirement. Specific work and health experiences in midlife were measured by two questions that asked for several life experiences-such as employer change, part-time work, and severe health problems-whether respondents had these experiences before age 40 and between ages 40 and 50. We constructed a dummy variable per life experience, indicating whether the respondent has had the particular experience before age 50 . Information about upward mobility was acquired via the question "how would you characterize the course of your career between ages 40 and 50" . To measure marital histories, information about the marital and partner status is combined with retrospective information about whether respondents have ever been divorced. The following categories were distinguished: married/cohabiting, never divorced; married/cohabiting, ever divorced; no partner, never married; no partner, ever divorced; no partner, widowed. In the analyses, we control for the respondent's gender, the study wave at which the dependent variables were measured , and the time elapsed since the respondent made use of an retirement arrangement. In addition, given that resources and retirement transition characteristics are established correlates of retirement adjustment, we take preretirement financial resources, preretirement perceived satisfaction with life, subjective health, voluntariness of the retirement transition, and age at retirement into account. Table 1 presents the means, standard deviations, coding, and wording of survey questions for all variables. In general, item nonresponse was low and was dealt with by using multiple imputation procedures . The variables with missing cases were imputed 25 times using information from the dependent, independent, and control variables. Thereafter, the regression models are estimated for all these 25 data sets and the results are combined . --- Analyses The relationships between earlier life experiences and the different dimensions of missing work after retirement were analyzed by estimating linear regression models and combining the estimation results by seemingly unrelated estimation . SUE is an appropriate technique when estimating different equations based on the same data. It combines the parameter estimates and variance matrices of the separate regression models , thereby allowing to test cross-equation differences between coefficients . To deal with the structure of the data , we control for organization in the analyses and used standard errors that allow for intradepartmental correlation . --- Results Table 2 presents the SUE results for the different adjustment dimensions. The models are estimated in two steps. In the first step, the relationships between earlier life experiences and missing money/income , social contacts via work , and status are examined. In the second step , resources and retirement transition characteristics are added to the equations. --- Life History Experiences In Model 1a, the extent to which retirees miss money/ income is regressed on information about earlier life experiences and control variables. The results show that experiences in both work and family spheres are associated with missing financial resources after retirement. As predicted in Hypothesis 1a, those retirees who experienced an upward career path-either a steep or a more gradual upward trajectory-are less inclined to miss money/income after retirement than those who did not experience upward mobility. Examination of interaction effects with gender suggests, however, that the effect of gradual upward mobility is significantly stronger for men than for women = .35, t = 2.39, p = .017). The steep upward mobility effect does not differ significantly by gender. As shown in Model 1a, the coefficients of years in the labor market, midlife employer change, part-time work, and severe health problems are not statistically significant. Regarding marital histories, the findings indicate that retirees who repartnered after divorce are more likely to miss financial resources than those who have continuously been married, as expected in Hypothesis 3a. Divorced retirees who remained single do not differ significantly from the continuously married group in terms of missing financial resources. The results regarding missing work-related social contacts after retirement are presented in Model 1b. None of the effects of the studied work and health history experiences is statistically significant. Marital histories, however, are found to be associated with missing work-related social contacts. As expected in Hypothesis 3b, divorced retirees without a partner are more likely to miss work-related social contacts than their continuously married and repartnered counterparts. Moreover, they are more likely to miss work-related social contacts than single never married retirees . Whether we used the scores provided at Wave 1, 2, or 3, is dependent upon the type of variable and the moment at which the respondent shifted into full-time retirement. For those who were already fully retired at Wave 2, we used the Wave 2 measures of partner status, health, and retirement transition characteristics. For those who transitioned into full retirement between Waves 2 and 3, we used the Wave 3 measures of these variables. The earlier life experiences, preretirement financial resources, and preretirement perceived satisfaction with life were measured at either Wave 1 or 2. --- SE Coef. --- SE Coef. --- SE Coef. --- SE Coef. --- SE Coef. statistically significant. Regarding marital histories, the results show that single divorced retirees are more likely to miss work-related status than continuously married, single never married , and single widowed retirees. The difference with the repartnered group is in the expected direction but not statistically significant at the 5% level. --- SE With respect to gender, the results show that men and women do not differ significantly in their likelihood of missing money/income, social contacts, and status after retirement. Inspection of interaction effects suggests that the impact of marital experiences differs between men and women on some dimensions. As expected in Hypothesis 4b, the effect of being divorced and single on missing social contacts is smaller for women than for men = -.45, t = -1.92, p = .055) but is strictly not significant. For the other adjustment dimensions, the impact of being single and divorced does not differ between men and women. The effect of being repartnered after divorce on missing status is significantly smaller for women than for men = -.39, t = -2.30, p = .022). F tests of cross-equation differences between coefficients show that the effect of a steep upward career path differs significantly across all three studied dimensions [money/ income vs social contacts , money/ income vs status , and social contacts vs status ]: Those retirees that experienced a steep upward career path are less likely to miss money/income after retirement but more likely to miss status, whereas no association is found with missing social contacts. The coefficient of being single and divorced is significantly larger for missing status than for missing money/income . Another noteworthy result is the role of the control variable time elapsed since retirement, which differs significantly between the money/ income and social contact dimensions . The findings show that the more years have elapsed since retirees made use of an retirement arrangement, the less likely they are to miss work-related social contacts. For the financial dimension, this effect is not statistically significant. --- The Role of Resources and Transition Characteristics In Models 2a-2c, resources and retirement transition characteristics are added to the equations. Preretirement financial resources are negatively associated with missing money/ income after retirement. For preretirement satisfaction with life, a negative effect on missing financial resources is observed as well. Those who had more financial resources and were more satisfied with life in preretirement years are less likely to miss financial resources after retirement. For the social contacts and status dimensions, these effects are not statistically significant. The perceived health situation of the retiree is relevant for all studied adjustment dimensions. Retirees in good health are less likely to miss work-related money/income, social contacts, and status compared with those in poor health. Also, a voluntary retirement transition is related to fewer adjustment difficulties on all studied dimensions. Those who retired at a relatively older age are less likely to miss money/income than those who retired earlier. For the social adjustment dimensions, the effect of age at retirement is not statistically significant. Comparing the effects of earlier life experiences between the life history models and the expanded models provides insights into the extent to which the effects of earlier life experiences are mediated by the established correlates of adjustment. The results show that the effects of career path and marital history generally remain statistically significant when taking resources and retirement transition characteristics into account, suggesting that these life history effects cannot be fully explained by the established correlates of adjustment. --- Discussion The difficulties retirees experience when adjusting to the loss of the work role are often assumed to be dependent upon experiences earlier in life. In line with the life course notion of multispheral development , this study shows that earlier life experiences in both work and family spheres are associated with missing work after retirement. Regarding work histories, the findings show that retirees who had a steep upward career path in midlife are less likely to miss money/income, equally likely to miss social contacts, and more likely to miss status compared with those that did not experience upward mobility. These findings clearly point out that retirees can miss work for different reasons, depending on their career path in midlife. Probably, mid-career pathways "set the stage" for experiences during late careers and one's postretirement situation-either by limiting or promoting resources and opportunities-and consequently shape retirement experiences. Marital histories were found to play a role as well. This study shows that divorced retirees without a partner are most likely to experience difficulties adjusting to the social changes accompanied with the loss of the work role. They were not only more likely to miss social contacts and status compared with continuously married retirees but also compared with single never married retirees, suggesting that among those living without a partner it is important to take diversity in terms of marital histories into account. Also, the long-term negative financial consequences of divorce experiences are reflected in the data. Those retirees who repartnered after divorce were found to be more likely to miss financial resources after retirement compared with continuously married retirees. Generally, men and women did not differ in terms of their difficulties adjusting to the loss of the work role, although the implications of marital experiences were found to differ slightly by gender. As expected, being divorced and single has a slightly stronger impact on missing work-related social contacts for men than for women. Paying attention to the multidimensional nature of adjustment appears to improve our insights into the postretirement process. The results show differences across dimensions in terms of the incidence of adjustment difficulties and processes over time. Moreover, predictors were found to differ across adjustment dimensions, suggesting that some effects might have been overlooked when using a combined measure of missing work after retirement. Generally, social contacts were found to be the work-related aspect that retirees are most likely to miss. The longer individuals are retired, the less likely they are to miss work-related social contacts, which might either suggest that retirees compensate work-related contacts by other contacts or perceive work-related contacts as less important over time. For financial resources and status, the incidence of adjustment difficulties is lower and no time effect is observed. This may indicate that for many persons-at least in our Dutch sample-retirement is not necessarily associated with an important loss of financial resources or status. The lack of a time effect could suggest, however, that for those retirees who do miss financial resources or status, the likelihood of experiencing difficulties does not decline the longer they are retired. Probably, income and status remain important needs among retirees , which may be relatively difficult to compensate for after retirement. No support was found for the hypotheses regarding various work and health history factors . These findings may suggest that not the amount of time employees have spent in the work role but rather the investments they made within the work role shape their retirement experiences. For the interpretation of the research findings, it is important, however, to take the Dutch country context into account. With regard to retirement savings and income, much more is organized at a collective level in the Netherlands than, for example, in the United States where individual workers mainly carry the risks and responsibilities . Replacement rates are relatively high and the income poverty rates among the elderly are low . In this respect, especially adjustment to the loss of the money/income provided by the work role might be relatively easy in the Netherlands, and its relationships with earlier life experiences might be relatively weak. Whether the incidence and predictors of the different dimensions of missing work after retirement are similar in other countries is an important question for future research. Studying other routes of exiting the labor market, such as disability or unemployment, might also be a relevant venue for future research. Leaving work due to disability or unemployment is likely to result in much less favorable outcomes than the retirement experiences examined in this study, given its relatively poor prospects and inherently involuntary character. Three limitations of this study should be kept in mind when interpreting the findings. First, we used rather broad retrospective questions to measure work and health histories, which might not have captured the meaning of the work role in sufficient detail. Moreover, several life history measures focus on the period before age 50 and therefore did not capture the years between age 50 and retirement. It might be the case that the impact of midlife experiences cumulates during one's late career and that these late-career experiences are more influential for shaping postretirement experiences. In future studies, it would be interesting to examine the role of both mid-and late-career experiences to disentangle their relative importance for explaining postretirement adjustment. Second, even though the retirees in the study sample form a highly diverse group in terms of earlier life experiences, resources, and retirement transition characteristics, they were all employed at four organizations. Therefore, the sample is not representative for Dutch older workers. Third, even though availability of information about the extent to which retirees miss work-related aspects is an important strength of the data, it should be noted that missing financial resources and social contacts were both assessed by single-item measures. For future research, it is advisable to develop multi-item scales to measure the three adjustment dimensions examined in this study, as well as other adjustment dimensions . Despite these limitations, this study shows that adjustment to the loss of the work role is a multidimensional process embedded in the individual life course. The findings of this study raise important issues for policy and practice. For policymakers, the findings point out that changing life course experiences might have important implications for retirement quality of future cohorts. Whereas the lives of Dutch men and women born between 1931 and 1940 generally reflected the standard life course, life courses destandardized among cohorts born after 1950. Variation in behavior increased, for example, divorces became more common . These developments might have important implications for the retirement experiences of future cohorts, given that divorced retirees were found to be most likely to experience difficulties adjusting to the loss of the work role. For retirement counseling, the results highlight the importance of not solely focusing on the current situation of older individuals but to view retirement as an integral part of the individual life course.

Although the process of adjustment to retirement is often assumed to be related to experiences earlier in life, quantitative empirical insights regarding these relationships are limited. This study aims to improve our understanding of adjustment to the loss of the work role, by conceptualizing retirement as a multidimensional process embedded in the individual life course. Design and Methods: Analyses are based on panel data collected in 2001, 2006-2007, and 2011 among Dutch retirees (N = 1,004). The extent to which retirees miss aspects of the work role (money/income, social contacts, status) is regressed on information about earlier life experiences, resources, and retirement transition characteristics. Results: The incidence of adjustment difficulties varies across dimensions. Predictors differ as well. A steep upward career path is associated with fewer financial adjustment difficulties but with more difficulties adjusting to the loss of status. Compared with continuously married retirees, divorced retirees without a partner are more likely to miss the social dimensions of work and those who repartnered are more likely to miss financial resources. The longer individuals are retired, the less likely they are to miss work-related social contacts. Implications: Changing life course experiences might have important consequences for retirement processes of future retirees.

Background In the past decade, public health strategies in Sub-Sahara Africa have focused on increasing young people's access to HIV testing and treatment services because of the high HIV incidence and low uptake of testing in this sub-group [1,2]. Adolescence is widely recognised as a period of physical and mental development, a time of testing boundaries, increasing independence and risktaking, which elevates their vulnerability to HIV infection [3][4][5]. Malawi has a young and growing population. In 2018, a fifth of the population was aged 15-24 years and nearly 32% of new HIV infections occurred in this group and a majority among female adolescents [6][7][8]. Young people in Southern Malawi have an increased vulnerability to HIV as they are less likely to use condoms and more likely engage in sexual activity at a younger age [8]. Additionally, pre-marital sex is strongly disapproved of which creates a "culture of silence" whereby the negative consequences of sex are emphasised so young people fear discussing sexual health issues with their parents or peers [9]. Myths and misconceptions about HIV transmission and treatment spread among adolescents and further elevate HIV risk [10]. There are multiple reasons why young women and girls in Malawi experience this high HIV burden including being more likely to experience sexual violence, intergenerational relationships and experiencing a younger age of sexual debut and marriage than their male counterparts [9]. Furthermore, adolescents in Malawi experience lower HIV testing coverage then their adult counterparts, exemplified in a 2015-16 survey where 79% of adults had ever received HIV testing and results in comparison to only 40.4% of adolescents [11]. Reaching young people with HIV testing and prevention and care services is a priority of the Malawi HIV and AIDS Strategy [12]. HIV self-testing at home has played a key role in increasing testing uptake in this group [1,13]. Choko et al. found that adolescents in Malawi, had the highest uptake of HIVST over a 2-year period, due to increased confidentiality, convenience and ease of use and 56% of users, of all ages, linked to treatment [13]. As a result of this data, HIVST for young people became a key focus in Malawi for the HIV Self-Testing Africa Initiative with implementation led by Population Services International a locally based international NGO working on the ground through community-based distribution agents on short term contracts, who distribute the HIVST kits to people of all ages in the community. Whilst distributing these tests the CBDA's advise and give information to all users on how to link to formal health services for confirmation testing, however they are not trained in professional post-test counselling. Included in the HIVST kit is a referral card which all users present at the formal health services. Despite the high acceptance of HIVST, young people in Malawi are less likely to link to formal health services post-HIVST [14,15]. This creates a treatment gap between the adult and younger population evidenced in 2016 where 41% of HIV positive adolescent and young Malawian people aged 15-24 were aware of their status and receiving Antiretroviral Therapy in comparison to of 69% adults [11]. Reported barriers for young people to accessing general sexual and reproductive health services in Malawi include; fear of HIV-related stigma and discrimination, the lack of confidential and "youth-friendly' SRH services and the associated cost of transport [16][17][18][19][20][21]. Reportedly, receiving STI treatment and therefore keeping their STI status private is a motivation for young Malawian people to access sexual health services as this preserves their social status in the community [17], the same hasn't yet been revealed, however, for young people's reasoning in linking to HIV services. The U = U campaign is a powerful message for people living with HIV to encourage linkage to post-testing services and adherence to treatment [22]. Yet, young people in Malawi may not have knowledge of this campaign as in 2016, just 41% of women and 44% men had a "comprehensive knowledge of HIV"which did not include knowledge of the U = U campaign [8]. We investigated self-tested young people's perceptions on the barriers and facilitators to linking to services after HIVST. --- Conclusions: Young people are happy to self-test for HIV but faced barriers to link to services following a self-test. Potential interventions for improving linkage suggested by this analysis include the establishment of youth-friendly linkage services, enhanced lines of communication between young people and healthcare providers and prioritising linkage for future interventions when targeting young people following HIVST. Keywords: Adolescents, Young people, HIV self-testing, Malawi, Linkage, Community-based health --- Methods A cross-sectional, inductive, qualitative approach was utilised to examine young people's perceptions about the reasons for linking, or not linking, to confirmatory testing and treatment after receiving a positive self-screen result. --- Study setting From the most recent national demographic health survey in 2015-16, the southern region of Malawi has the nationally highest HIV prevalence [8]. Two villages within the Southern Machinga district were chosen for sampling locations as unpublished process data from STAR demonstrated that young people who selftested from the Southern Machinga district had comparatively low rates of linkage to services. Village 1 was purposively chosen to sample participants because it had the lowest rates of linkage in the district and, therefore, likely held the richest data. Village 2 in the same district was chosen for comparison as it was the closest village with higher rates of linkage to Village 1. --- Sampling strategy To approximate information saturation in the short time frame provided, rapid appraisal methods were used with swift participation sampling rather than other naturalistic methods [23,24]. Young people were sampled purposively, where predicted information-rich participants were recruited by local CBDAs, who would verbally describe the study so the young people could provide informed consent. Young people aged 16-24 were sampled, with a deliberate majority of females and efforts to include younger adolescents . To avoid stigma associated with study-participation, participants were not required to reveal their HIV status or whether they linked to services post-HIVST. Healthcare workers and CBDAs were sampled from the village's health centres, balanced in age and gender. Included healthcare providers were of the same cadre to lessen professional hierarchy bias. --- Data collection Rapid appraisal methods; such as in-depth interviews with purposive opportunistically sampled participants and focus group discussions of key informants, aim to assess and address a community's health needs within a limited time frame to give practical recommendations and was therefore chosen to approximate information saturation in the short time frame provided [24]. In-depth interviews were adopted with the 16-24-year-old study participants as these allowed for an in-depth exploration of their individual perspectives in relation to their underpinning characteristics and beliefs [23]. Furthermore, as this involved discussion of sensitive topics like HIV testing, a focus group discussion would likely inhibit responses about personal behaviour due to the lack of confidentiality. FGDs was chosen to explore the collective perspectives of the HIVST suppliers; healthcare workers and CBDAs , to additionally observe any conflicts or agreements between the villages and professions [25]. IDIs and FGDs were held in the participants' preferred languages of Chichewa or Chiyao by a research assistant who had previously lived and worked in the area to further build trust with the participant but also impart contextual knowledge to the main researcher , who took observational notes and made further iterative enquiries to further explore previously unexpected topic areas. A semi-structured interview topic guide was used to guide discussions, with open questions such as "What are your opinions of local health facilities and how they treat young people?" and further solution-orientated questions, for instance "if you could organise how young people link to services after HIVST, what would you do?". The topic guide also included methods of free-listing, ranking and scenarios as described by Palmer [24]. Free listing involved the participants listing any reasons young people would link to services , these were then ranked by perceived importance to give means of comparison. These reasons and rankings were referenced for further exploration within the interviews. All participants were presented with hypothetical scenarios and asked to describe what would happen in their community, for example, "What would a young person do if they went to a health facility to get treatment and they saw someone else their age?"; "Would it be different if they knew them?". This methodological triangulation helped build rapport, facilitated further exploration and safeguarded the data's trustworthiness [26,27]. --- Data analysis A thematic framework approach was used to analyse the qualitative data and observe differences between participants' perspectives. After transcription and translation of data the lead researcher first familiarised and immersed herself in the data. From this immersion, a coding framework was developed using a priori/deductive codes and each IDI and FGD was double-coded after the transcripts had been imported in NVivo 11 . Categories of coded data were grouped to reveal overarching themes. Participant's responses were compared between villages, young people, healthcare workers and CBDAs. Data triangulation was also stratified by the socio-demographic characteristics of; age, sex, village and HIV status where the latter had been disclosed voluntarily. Ethical approval for this project was given by the Liverpool School of Tropical Medicine on the 29th March 2018 and in Malawi by the College of Medicine Research and Ethics Committee on the 13th April 2018 under research with the STAR affiliated with Malawi-Liverpool Wellcome Trust . --- Results We included 41 participants . Thirteen IDI's were conducted with the young participants and an additional 28 individuals participated in four FGD's with two groups of healthcare workers and two groups of CBDAs from each village. Although the young study participants were not required to reveal their HIV status, 12 out of 13 participants voluntarily disclosed this during the IDIs. By chance, most young people sampled from Village 1 had a HIV + status whereas as the majority from Village 2 had a HIVstatus. Confidentiality and not wanting to disclose HIV status, social support, communication, and attitudes/perceptions toward the health facilities emerged as key themes on factors influencing linkage to healthcare following a positive HIVST result at home for young people. --- The contextual impact on the assurance of confidentiality Most participants agreed that young people would not link to formal health services following a positive HIVST result due to the anticipated negative social consequences of inadvertent disclosure of their HIV status. Interestingly and in contrast to the majority, three young people from Village 1 described how early linkage to confirmatory testing and subsequent treatment preserved their serostatus privacy; --- "You try to hide it from people because when you wait for you to be sick people get to realize what is really wrong with you" A healthcare provider from Village 1 described that the facility held youth-specific days for HIV services supported young people's access to post-HIVST services due to the social support from their peers and youthfriendly health providers. Four of the seven young people and CBDAs from Village 1, however, narrated that this approach as an obstacle as it is not effective in safeguarding their privacy in their serostatus or that they were accessing post-HIVST services, and preferred using healthcare centres further away but with less visible routes; "When people in the village see [young people] going on that day, they definitely judge they are going for treatment and mock them" In conflict with this position, nearly all young people ranked the long distance to healthcare centres and associated transport costs as a highly important barrier to linkage. One CBDAHIVST provider from Village 1 described young people's frustration on the lack of a convenient and private approach for linking to services for those self-testing HIV positive; "[Young people] tell us that 'you have a self-testing method that preserves privacy so how come you cannot give us treatment that does too?'" --- Availability of social support and trusting relationships The presence or absence of social support from a trusted individual was described as an important factor for youth to access confirmatory testing and treatment following HIVST. There were conflicting views on parental support; three young females expressed concerns that their parents may compromise their unwillingness to disclose their serostatus publicly. Eleven young people, both CBDA and the village 2 healthcare FGD participants, observed that CBDAs were well placed to foster trusting relationships with young people and could encourage accessing formal health services following HIVST because they lived and worked in the community and could be accessed in the village for advice without young people experiencing fear of inadvertent HIV status disclosure; --- "Even when [my friend] has some health problems, he consults the CBDA, I even saw him yesterday going to the CBDAs house" . Despite their placement in the community, CBDA's were perceived as confidential figures as the young people had observed them talking to their peers and their peer's serostatus was not revealed within the community. Four young people described peer support as a facilitator to linkage following HIVST but these were often informal and with no mention of sexual partners. For example, one young self-tester from Village 1 narrated that some young PLHIV created a secret support group to encourage each other's treatment adherence. However, many young people and participants in the four FGDs discussed how the general perception that HIV treatment is not a cure leads to a fatalistic attitude even among peers; --- "[Young people] don't go because of peer influence, when they disclose to their friends, [they] influence them to not to go to the hospital… Because they say the virus has no cure " Both village healthcare centres were said to have "expert client" volunteers in the facilities who were presented to patients as "treatment role models". The healthcare workers and some CBDAs were convinced that expert clients had an important role to play in facilitating health-service linkage and treatment adherence among HIV self-tested youth; --- "When youths see these HIV positive people living happy and healthy, they get encouraged and see no reason of isolating themselves from treatment. " In contrast, "expert clients" were ranked of low importance by the young people, and were even described as "ineffective" by a healthcare worker in Village 2, who narrated that no expert client is under 25-years-old and their 'counselling' is only confined to the health facility. --- Communication between CBDAs, healthcare workers and NGO staff The most notable difference between the villages was the relationship between healthcare workers, CBDAs and NGO staff. In Village 2, this relationship was a facilitator, but a barrier in Village 1. A healthcare worker from the Village 1 health centre lamented that the CBDAs didn't counsel young people when they received a positive result, even though CBDAs are trained in HIVST distribution, not professional post-test counselling. This suggests that the CBDAs' role was not fully explained to the healthcare workers in Village 1. A young woman from Village 1 described how the fractured relationship between CBDAs and healthcare workers hindered her access to post-HIVST services when she presented at the facility; --- "[The healthcare workers] didn't believe that I have done self-testing, [my CBDA] gave me a referral card, I said I don't see any reason to not believe me. " The healthcare workers in Village 2 described how they experienced a similar scenario and the NGO organised a meeting to explain the roles of the professionals to each other and confirmed referral methods. In this meeting, it was decided that mobile phones would be used to support referrals of young HIV self-testers between CBDAs and healthcare workers based at a facility. However, a meeting of this nature had not taken place in Village 1. Participants in all the four FGDs desired more interaction and communication between CBDAs and health workers to ensure seamless referrals and to prevent future tension, as articulated; "We were not involved in any of the meetings which the project implementers conducted with the CBDAs, so we had no clue on what role we were playing and what role were the CBDAs playing. " Furthermore, this lack of communication between the healthcare providers, CBDAs and self-tested young people contributes to a failure in recognising young people's barriers to post-HIVST services linkage. The young people and CBDAs ranked the following four factors of high priority for linkage whereas the healthcare workers ranked these factors as low priority. --- Role of health centres in promoting linkage Seven young people and all of the FGD participants ranked healthcare services as the most important facilitator to ensure linkage to confirmatory testing and HIV treatment and care. The need for a confirmatory test was said to encourage young people to link in order to dispel anxieties on whether young self-testers performed the self-test accurately. Furthermore, eight young people stated they, and other peers, linked to health centres after observing HIV symptoms and to "stay healthy". No young person mentioned the U = U campaign as an incentive for linkage, however, this was also not probed for. The FGD participants described the healthcare workers providing confidential counselling and treatment as another facilitator. Most young people described healthcare workers as respectful or "welcomed them properly". Noticeably, young people and healthcare workers both described this respect as reciprocal; --- "It depends on the attitude of the [young] person …. We don't shout at them but on the issue of respect it depends on their attitude According to our job and hospital rules we still respect [by not shouting or getting angry with] them whether they respect us or not" --- Influence of demographic characteristics on linkage When the perspectives were stratified by gender, it was observed that young males were more concerned with the impact on their societal status in the community or loss of sexual partners from a potential inadvertent HIV status disclosure during linkage. In contrast, young females were afraid of the social consequences in terms of losing potential future marriage proposals and financial security. Additionally, young women described how they were less able to afford any transport costs as they held less lucrative jobs living with their parents; "A boy would emigrate and work at tobacco farms to buy a bike while a girl wouldn't" More females stated having "strength of mind" as an important facilitator. This may relate to a CBDA's observation in Village 1 that females are more likely to have a close relationship with their parents at home, disclose their status and receive encouragement to have the 'strength of mind' to link to formal health services. For the young men, the story was different as illustrated in a quote below; "It is really difficult for young people, mostly boys, they take time to disclose their results to their parents, they are hard to convince them to accept it, it is like you're working with a lion" In terms of age, study participants falling within the 16-19 and 20-24-year-old age groups held similar perspectives on linking to post-HIVST services. However, this is likely due to the lack of comparative data as only three participants were aged 16-19, and they seemed to find it difficult to fully articulate their reasoning for linking to services or not. One young person from Village 2 described how adolescents find the associated travel costs a larger obstacle than their older counterparts as they are more likely to be financially dependent on their parents. Additionally, a healthcare worker noted how the cultural perceptions of adolescents contribute towards undermining agency to make independent decisions; "According to our Malawian culture, a person who is sixteen is still a little child…, he needs a guardian to guide him" The village of residence had a bearing on the ability to link to services following HIVST. A major difference between the villages was that young people from Village 2 described living a "sexually-risky" lifestyle as a driver to link to post-test confirmatory testing and treatment services as illustrated in this quote; "Most male young people who go to the lake earn huge amounts of money which influence them to indulge in sexual relationships with all kinds of people and when they test themselves Negative that's when they come for verification" . This "sexually-risky" lifestyle is likely related to the available income from fishing activities in Village 2, and not in Village 1, so young men from this village were more likely to engage in transactional sex. Furthermore, more young people in Village 1 discussed issues relating to travel "visibility" as there is a single route to the nearest health-centre which the local community lives alongside. Meanwhile in Village 2, there are multiple, less visible routes to the health facility. --- Discussion This study is one of few to describe young people's perceptions of barriers and facilitators to linking to services following HIVST. Similarly to Malawi adults and echoing previous findings, confidentiality and the self-preservation of their serostatus pervaded young people's reasoning for linking to post-HIVST services [18,28]. This was gendered and affected by young people's age, reiterating Hatchet et al. [29], as adolescents and young males were less likely to disclose their status to others and so experience less support in linking to post-HIVST services. Our finding that young people emphasized having "strength of mind" as a facilitator highlights how young people feel expected to shoulder the burden of linking to formal healthcare services by overcoming a multitude of barriers. An unexpected barrier was that of poor communication/relationship between CBDAs working in the community distributing HIVST kits and the healthcare workers based at a clinic who provide follow-up services. This finding shows a "fragmented" approach to implementation, with a focus on increasing HIV testing uptake but little emphasis on subsequent linkage to post-HIVST services like confirmatory testing or treatment. This lack of joined up thinking can negatively impact sexual health outcomes [30]. Community health programmes that effectively work with close-to-community providers have been shown to successfully reach pregnant adolescents or young people who were lost to follow-up from HIV care, by building on the important interface role that close to community providers have between the health system and the communities they serve [31,32]. Deliberately aligning tasks to national community-health programmes is likely to ensure improved community support to posttest linkage in adolescents during a period when international funding for bespoke support is declining [31][32][33]. Our findings on the importance of the CBDA cadres give further evidence how it is vital to understand the local context to address issues of professional distrust or unfamiliarity which affect upon service delivery performance and impact [31,32]. Youth participation should be a programme priority to increase young people's access and use of SRH services [34][35][36]. However, this study showed a lack of communication between the young people and the healthcare workers. In Kenya, a study reported that having stakeholders and a young person's "advisory group" involved in the planning of treatment distribution, improved linkage by 41% [37]. Therefore, this should be a priority in future HIVST implementation to increase young people's linkage to post-HIVST health services. As with previous findings, [18,38], this study also highlighted how CBDAs were motivators and preferred by young people because they were trusted and lived in the villages with the young people, whereas this trust in healthcare professionals was only experienced by those who linked to services. In this study, 'expert clients' roles were described as confined within the health facilities with a limited role in the community. This is a lost opportunity, as multiple studies [21,29,39] have reported how integrating peer supporters, especially PLHIV, into the healthcare system during follow-up visits can be especially effective in increasing young people's linkage. Hence, they could effectively promote linkage for the young people by working in conjunction with the CBDAs in the community. Expert clients and/or CBDAs could also provide a means of transport which has also been found to increase linkage for young people [40,41] and help to improve communication between those in the community and healthcare providers based at the referral facility. --- Methodological limitations Using CBDA recommendations for purposive sampling may have created a sample bias towards the young people's positive CBDA descriptions as the CBDAs likely chose participants with whom they had a positive relationship. Furthermore, there was an unintended location bias whereby most young people sampled from Village 1 disclosed a HIV + status and in Village 2 a HIVstatus. All the young PLHIV had linked to post-HIVST services and as such their perspectives and experiences may have been different from those of the individuals who failed to link. Most of the HIV negative young people's perspectives were based on other's experiences or hypothetical thinking. This lack of comparison with experiential data was mitigated through the triangulation of data from different sources and the range of participants provided a holistic view of linkage reasoning from the HIVST supply and demand perspectives for young people. Despite the sensitive nature of the discussions with the young people, the trust gained with the researchers in the limited time frame is evidenced through the majority of participants self-disclosure of their serostatus. This may be due to the researchers being balanced in gender and of a relatively younger age, 25 and 26. --- Conclusions In conclusion, HIVST provides confidential and convenient testing which increases uptake among young people. However, linking to confirmatory testing and treatment following a self-test at home remains a daunting challenge as the onus of responsibility to link rests on the young people and their ability to navigate multiple and complex barriers. This study shows the importance of communication between target beneficiaries and the different health providers involved within a health intervention delivered at community level. It also highlights young people's reasoning for linking to services post-HIVST, including the differing contextual and gendered perspectives, which future Malawian policy makers and implementers can use to implement effective interventions with a targeted response to encouraging different groups of young people's linkage post-testing. Implementing these recommendations in the national scale-up of HIVST in Malawi would maximise the benefits of young people's high uptake of HIVST; so more young people receive treatment to achieve viral suppression and, ultimately, reduce the national incidence and prevalence of HIV in Malawi. --- --- Abbreviations ART : Antiretroviral therapy; CBDAs: Community-based distributing agents; FGD: Focus group discussions; HIVST: HIV self-testing; IDI: In-depth interviews; MLW: Malawi-Liverpool Wellcome trust; PLHIV: People living with HIV; SSA: Sub-Saharan Africa; STAR : HIV self-testing Africa Initiative; WHO: World Health Organisation. --- --- Funding The project was funded by MLW and self-funded by the main researcher, Lisa Harrison. --- --- --- Competing interests The authors declare that they have no competing interests. ---

Background: Young people, aged 16-24, in southern Malawi have high uptake of HIV self-testing (HIVST) but low rates of linking to services following HIVST, especially in comparison, to older generations. The study aim is to explore the barriers and facilitators to linkage for HIV prevention and care following uptake of HIV self-testing among young Malawians.We used qualitative methods. Young people aged 16-24 who had received HIVST; community-based distribution agents (CBDAs) and health care workers from the linked facilities were purposively sampled from two villages in rural southern Malawi.We conducted in-depth interviews with thirteen young people (9 female) and held four focus groups with 28 healthcare workers and CBDAs. Young people strongly felt the social consequences associated with inadvertent disclosure of HIV sero-status were a significant deterrent to linkage at their stage in life. They also felt communication on testing benefits and the referral process after testing was poor. In contrast, they valued encouragement from those they trusted, other's positive treatment experiences and having a "strength of mind". CBDAs were important facilitators for young people as they are able to foster a trusting relationship and had more understanding of the factors which prevented young people from linking following HIVST than the healthcare workers. Young people noted contextual barriers to linkage, for example, being seen on the road to the healthcare centre, but also societal gendered barriers. For example, young females and younger adolescents were less likely to have the financial independence to link to services whilst young males (aged 19-24) had the finances but lacked a supportive network to encourage linkage following testing. Overall, it was felt that the primary "responsibility" for linking to formal healthcare following self-testing is shouldered by the young person and not the healthcare system.