Datasets:

mimiklee

/

masterthesis_clean_text_20k

like 0

Introduction Currently, one in eight Americans is over 65 years of age [1]. By 2050, more than 33 million persons in the United States will be aged 65-74 years, while 30 million will be between 75 and 84 years, and 20.5 million will be over the age of 85 years [1]. The quality of older adults' lives is receiving increased attention. A recent report from the National Center for Chronic Disease Prevention and Health Promotion [2] noted that QOL decreased in the US population as a whole from 1993 to 2001. Both the increase in the proportion of the population over 65 years of age and the reported decrease in QOL underscore the need to examine factors associated with QOL in older adults. Quality of life is defined by the World Health Organization [3] as "individuals' perception of their position in life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards, and concerns." The WHO further notes that QOL is a concept with several domains, including physical and mental health, social functioning, and emotional well-being. QOL has been conceptualized in many ways depending on the discipline, paradigm, and time frame of research examining QOL [4]. QOL measurements also vary. Broad measures such as the WHOQOL-BREF instrument include several dimensions of QOL in one scale. Other instruments such as the SF-36 scale are health-related measures. Some are disease specific such as the European Organization for Research and Treatment of Cancer QOL questionnaire. Finally, there are domain-specific measures such as the Rand Social Support Scale [4]. The broader QOL measures have been most widely used. Proponents of these broad measures note that QOL is an overarching concept that reflects the physical, mental, and social aspects of an individual's life [5]. Others, however, point out that broad QOL measures may obscure important nuances [6]. For example, an elderly person with chronic illnesses may have poor health-related QOL but high non-health-related QOL. Therefore, in any study, it is important to decide whether a broad QOL measure or measures of specific dimensions are most appropriate. Similarly, there are discussions as to what constitutes old or elderly [7]. Most developed countries have accepted the chronological age of 65 years as a definition of 'elderly' or older person. It is often associated with the age at which one can begin to receive pension benefits. This common use of a calendar age to mark the threshold of old age does not assume that biological aging starts at 65. There is great variability in biological aging, and it is generally accompanied by decline in one or more of a person's abilities in the physical and cognitive domains [8]. It is well established that decline in one or more of a person's abilities is associated with reductions in QOL [5]. However, the degree to which and the dimensions of QOL affected will vary depending on the extent of decline in abilities and other factors specific to individuals. Understanding the connections between biological aging, individual factors, and the dimensions of QOL is important for planning of interventions and services to enhance QOL in older adults. In the study reported here, Andersen's behavioral model [9] was used to assess factors associated with QOL in older adults. The model depicts how contextual and individual characteristics, which include predisposing factors such as demographics, enabling factors such as social, human, and material resources, and need factors such as number of chronic conditions, activities of daily living , and mental status, influence health behaviors and QOL . Health behaviors include physical activity and health care utilization while QOL includes the following: health-related QOL , social functioning, and emotional well-being. Previous studies have found that both QOL and HQOL are associated with predisposing and enabling factors including age gender, ethnicity, education, income, marital status, and family size [2,[10][11][12][13][14][15]. In addition, HQOL has been shown to be influenced by the number and severity of chronic diseases, though social functioning was less affected by these [16,17]. Moreover, lower ADL function, depression, and memory problems have been linked to poorer QOL [15,[18][19][20]. Finally, health behaviors such as physical activity and health care utilization have been associated with QOL [21][22][23][24][25]. However, the relationship between health care utilization and QOL has varied. Less use of health care services was associated with better QOL in asthma patients and female veterans, but for male veterans and health care services obtained specifically for preventive services, QOL was worse [23][24][25]. In summary, all the variables in the Andersen model have been linked to QOL. However, the relationships between the different dimensions of QOL in older adults and various contextual and individual characteristics and health behaviors are not clear. Therefore, this study examined three dimensions of QOL in community-dwelling older adults. We hypothesized that controlling for predisposing factors, a person's needs , and health behaviors would be linked to HQOL, social functioning, and emotional wellbeing. --- Methods This descriptive, correlational study used secondary data from the National Health and Nutrition Examination Survey from 2005 to 2006. The NHANES from the National Center for Health and Human Services [26] is a national survey designed to assess the prevalence of major diseases and risk factors for diseases, with the aim of assessing nutritional status and its associations with health promotion and disease prevention. Data are collected through both personal interviews conducted and physical examinations carried out at mobile examination centers. The NHANES examines a nationally representative sample of approximately 5,000 individuals each year from the civilian, non-institutionalized US population. The sample is chosen in four stages; first strata are determined based on geographic location, and then, a random sample of 15 US counties is selected within strata . Stratified sampling is used to capture all ages of the US population. To produce reliable statistics, NHANES oversamples persons 60 years and older, African Americans, and Hispanics. For this study, we used data from personal interview modules for individuals 65 years or older. The study included 16 variables. Some were extracted directly from the data set, and others were created as we explain below. --- Measures --- QOL In Andersen's behavioral model, outcomes include perceived health and satisfaction [9]. We expanded these outcomes to the broader concept QOL. Participants' QOL was measured using 3 variables: HQOL, social functioning, and emotional well-being. The HQOL score was a composite of the number of physically and mentally unhealthy days in the past month. In two NHANES questions, participants were asked how many days during the past 30 days their physical health was not good and how many days their mental health was not good. These single items and the composite score have been validated in test-retests and against reported health conditions, physical exams, and broader instruments such as the SF-36 and the WHOQoL-BREF [27,28]. The distribution of HQOL clearly showed two modes with 56.6% of the sample reporting no days of physical or mental unhealth and 12% reporting 30 unhealthy days. In order to improve the data with respect to the shape of the distribution, a revised HQOL variable was created with 6 levels: no unhealthy days = 0, one to five = 1, six to ten = 2, eleven to fifteen = 3, sixteen to twenty-nine = 4, and thirty = 5. For social functioning, we created an index using two questions: number of close friends and how often a person attended church or religious services. A social functioning index based on previous NHANES data using four questions has shown a high level of predictability of the social functioning dimension of QOL [29]. However, two of those questions were not available in the NHANES 2005-2006 data, so a two question index was created. After examining distributions of the number of close friends, participants could feel at ease with, could talk to about private matters, and could call on for help, and we assigned a value of 0 for no close friends, 1 for one close friend, 2 for two close friends, 3 for three close friends, and 4 for four or more close friends. For yearly church or religious services attendance, we assigned a value of 0 to never, 1 to once, 2 to two, and 3 for three or more. Thus, the index score for social functioning ranged from 0 to 7 with higher scores representing better social functioning. The emotional well-being score was also created using two questions. The first question asked whether there was anyone to provide emotional support. A value of 1 was assigned for yes and a value of 0 for no. The second question asked whether more emotional support had been needed in the last year: an answer of no = 1 and a yes = 0. The index score thus had a possible range of 0-2 with higher scores, indicating better emotional well-being. --- Health behaviors Health behaviors are, according to Andersen's behavioral model, defined as actions people do to stay healthy or improve health. For this study, health behaviors included physical activity and health care utilization, both of which have been linked to increased age [9,[21][22][23]. While the NHANES data included other health behaviors, these two behaviors seemed most likely to be affected by increasing age. Physical activity was operationalized as level of physical activity performed on average each day, ranging from 1 = sits during the day and does not walk about very much, 2 = stands or walks about a lot during the day, but does not have to carry or lift things very often, 3 = lifts load or has to climb stairs or hills often, to 4 = does heavy work or carries heavy load. Health care utilization was coded as number of times a respondent had received any type of health care in the last year: no health care utilization = 0, one visit = 1, two to three visits = 2, four to nine visits = 3, ten to twelve visits = 4 and thirteen or more = 5. --- Contextual and individual characteristics In Andersen's behavioral model, contextual and individual characteristics are conceptualized as factors that impede or enhance individual's health behaviors, most notably health care utilization and subsequent other outcomes related to health and satisfaction [9]. The model divides the characteristics into predisposing demographic, enabling socioeconomic, and need for care factors. In our study, predisposing, enabling and need factors were captured by seven variables. Predisposing factors included demographic variables such as age, gender, and race/ethnicity. Age was a continuous variable from age 65 to age 85 or above. Race/ ethnicity was classified as Hispanic, non-Hispanic white, non-Hispanic black, and other. Enabling factors were operationalized as social, human, and material resources, including individuals' immediate social connections such as number of people in the family, marital status, education, and poverty level. Social resources were defined as number of people in the family and marital status. Number of people in the family was the total number of related people in the household including the respondents; scores of 1-6 reflected the actual number and 7 was used for seven or more people. Marital status was coded as married or not married . Human resources included education, coded as a high school degree or more or less than a high school degree. Material resources were based on a poverty income ratio. This ratio, provided in NHANES, was the ratio of the family income to the poverty threshold. A lower number represented a lower income. Needs factors were operationalized as number of chronic conditions, ADL function, and mental status. We created a score for number of chronic conditions using participants' answers to questions about whether they had ever been told by a doctor that they had a particular condition. Thirteen conditions, including pulmonary, cardiovascular, cancer, stroke, kidney, diabetes, liver, and arthritis illnesses, were scored for a score range of 0-13. ADL function was the sum of 16 items on the Activities of Daily Living scale, which measures constructs associated with locomotion and transfers, household productivity, social integration, and manipulation of surroundings [30]. Items scored from 1 to 4 where 1 = no difficulty, 2 = some difficulty, 3 = much difficulty, and 4 = unable to do for a total score range from 16 to 64. Higher scores indicated lower ADL function. Mental status was operationalized as memory problems and depression. One question asked whether a person had memory problems or not. Depression was the total score on the depression module in the Patient Health Questionnaire, a self-administered version of the PRIME-MD diagnostic instrument for common mental disorders. The module contains 9 DSM-IV items asking participants whether they have been bothered by feeling down, depressed, or hopeless in the last 2 weeks. Item scores range from 0 = not at all to 3 = nearly every day for a total scale score of 0-27 [31]. Higher scores indicate more depression. --- Analyses Means and percentages were used to summarize the characteristics of study subjects. Linear regression was used to examine associations between the independent variables described earlier and the three QOL dimensions. For each of the three QOL measures, we first examined bivariate relationships by linking each of the needs and health behavior variables to the specific QOL measure. We next constructed multivariate models by linking all the needs and health behavior variables together to the specific QOL measure. Both models controlled for age, gender, race, family size, marital status, education, and poverty index. All descriptive and regression analyses took into account the complex NHANES survey design, including stratification, clustering, and unequal weighting. Hence, analyses were performed in SAS survey procedures using the whole NHANES data set with a subpopulation statement for people 65 years and older. --- Results The 2005 NHANES sample contained 911 adults 65 years or older with no missing values in the 16 variables described earlier . Our initial sample was 1,189, but 278 had missing data on one or more of the 16 variables we extracted or created. Missing data were more common in older people who were less likely to be married and had worse ADL function, more memory problems, lower physical activity levels, and worse QOL in all three dimensions. The majority of the final sample of 911 were female , and their average age was 73.6 years. Most identified themselves as white , with 8.2% black and 5.1% Hispanics. Their family size was an average of 1.9 people, and 57.8% were married. The majority had a high school degree or more education and were above the poverty threshold . On average, these older adults reported 2.5 chronic conditions and no difficulty to some difficulty with ADL function . Only 14% reported memory problems and few had been depressed in the last two weeks . The average score on physical activity indicated that most of the sample stood or walked a lot during the day, but did not have to carry or lift things very often. On average, participants received health care two to nine times a year. The HQOL score was 1.22, indicating an average of more than five but less than ten unhealthy days in the last 30 days. The average score on social functioning was high at 5.16, indicating that most had at least two close friends and had attended church or other religious services more than once in the last year. The average score on emotional well-being was 1.74, indicating that most people had someone to provide emotional support and received all the emotional support they needed in the past year. When examined individually, all the needs and health behavior variables were significantly associated with at least one QOL dimension, controlling for age, gender, race, family size, marital status, education, and poverty index . Specifically, lower ADL function, memory problems, and more depression were associated with poorer QOL in all three dimensions. A higher number of chronic conditions were associated with both poorer HQOL and social functioning, while greater health care utilization and lower physical activity were associated with poorer HQOL. In the full models, when all needs and health behavior variables were included, poorer HQOL was associated with having a high school degree or higher, lower ADL function, and more depression . When full models for social functioning and emotional wellbeing were analyzed, social functioning was higher in people who were older, female, black, married, with a high school education or higher, and with better ADL function, no memory problems, and greater health care utilization. Emotional well-being was lower in people who were Hispanic and black, with lower ADL function, more depression, and less physical activity. --- Discussion This nationally representative sample of community-dwelling people 65 years and older reported high QOL in all three dimensions. In bivariate analyses, the need variables, ADL function, memory problems, and depression were associated with all three QOL dimensions, while the number of chronic conditions was associated with HQOL and social functioning. However, in our full models, only ADL function was significantly related to all three dimensions of QOL. In previous studies, ADL or physical function was associated with social functioning in older women. Our results that memory problems were associated with social functioning, and depression with HQOL and emotional well-being are somewhat different from previous studies where memory problems defined both as working memory and executive function and depression were associated with only HQOL [18][19][20]. Our finding that number of chronic conditions was not associated with any QOL dimension is also different from previous studies that found number of chronic diseases associated with HQOL in older adults [13,18,32]. The differences in results from bivariate and full model analyses suggest that the needs variables may affect each other, a relationship that needs further examination. Future studies are needed to explore how different needs affect each other and different dimensions of QOL. Even so, our findings underscore the importance of diagnosis and treatment of lower ADL function, memory problems, and depression to improve the dimensions of older adults' QOL. The two health behaviors examined were associated with one QOL dimension, but neither was associated with HQOL. Less physical activity was associated with better emotional well-being. Thus, these older adults experienced more emotional well-being when they did not have to do higher levels of physical activity such as heavy lifting or strenuous activity on a regular basis. The second health care behavior examined, greater health care utilization, was linked to lower social functioning. This association may have been a function of limited time for social activities when more time is spent on contacts with heath care providers. Another plausible explanation is that higher health care utilization may have been a proxy for more extensive health problems, which themselves could limit social functioning. More in-depth examination of physical activity using other measures as well as health care utilization and QOL is warranted. There were several limitations to our study. First, the findings were based only on subjects who had complete data. The initial sample was 1,189, but 278 had missing data, When we compared those with missing data and those included in the study, we did find differences. Since people with missing data scored worse on QOL, using only the sample with no missing data provided an estimate of better QOL than it should be for the whole representative sample of community-dwelling older adults. In addition, results presented from regression analyses assumed that data were missing at random; however, we cannot test for that, so our results might have been different if this assumption was not met. In our analysis to examine missing data patterns, we found that people with missing data were older, less likely to be married, had worse ADL function, more memory problems, lower physical activity, and lower QOL. Second, all our QOL measures were generated from population-based measures. Thus, the study was limited to testing associations between variables reflecting computed and transformed QOL measures. While the two item HQOL has been validated against longer HQOL measures [27, 28], our transformations have not been validated. Also, the two-item index we computed for social functioning did not capture the multifunctionality of social functioning. It would have been preferable for the NHANES data to include all four domains in the Berkman-Syme Social Network Index, including church or other religious group membership and membership in other community organizations [33]. Third, in the NHANES survey, memory problems were measured by self-report, rather than an objective measure. Thus, it is possible that older adults underreported or were unaware of their memory problems. Fourth, because the NHANES database limits the response choices for the categorization of age that includes a category of over the age of 85, it is possible that the oldest-old are not adequately represented in our findings. Finally, we used cross sectional data so we are reporting associations, not causal links. Longitudinal analyses are needed to examine temporal relationships of these variables with QOL. Nevertheless, the study found differences in the contextual and individual characteristics and health behaviors associated with the dimensions of QOL in these community-dwelling adults 65 years and older. The finding that different dimensions of QOL were associated with different factors confirms the emerging consensus that QOL is a multidimensional construct and needs to be measured as such [4]. The findings are also useful for designing interventions to improve QOL in the growing population of older adults. Since depression, memory problems, and ADL function were all associated with some dimension of QOL, future interventions to improve QOL in older adults should include interventions such as screening and treatment for these problems. Conceptual model. Adapted from Andersen's behavioral model [9]. Factors associated with quality of life in older adults in the United States --- Table 3 Multivariate results for health-related quality of life, emotional well-being, and social functioning in US adults 65 years and older

Purpose-According to the World Health Organization, quality of life (QOL) includes physical and mental health, emotional well-being, and social functioning. Using an adaptation of Andersen's behavioral model, we examined the associations between the three dimensions of QOL and needs and health behaviors in a nationally representative sample of adults 65 years and older.Examination Survey (NHANES) was used. NHANES over-samples persons 60 years and older, African Americans, and Hispanics. Frequencies and distribution patterns were assessed, followed by bivariate and multiple regression analyses. Results-These older adults reported high levels of QOL. However, associations between needs and health behaviors and QOL varied across dimensions. Activities of daily living (ADL) were associated with all three dimensions. Depression was associated with two dimensions and memory problems with one dimension. Physical activity was linked to social functioning, and health care utilization was linked to emotional well-being. Conclusions-The differences in associations with different dimensions of QOL confirm that this is a multidimensional concept. Since depression, memory problems, and ADL function were all associated with some dimension of QOL, future interventions to improve QOL in older adults should include screening and treatment for these problems.

Introduction Suicide remains an important public health issue in China, although the suicide rate has decreased substantially over the past decades [1]. In 2017, the estimated mean suicide rate was 7.2 per 100,000 persons in China, which has declined by 65% compared to the rate of 20.9 per 100,000 persons in 1990 [2]. However, there was still an upward trend in 15-34 rural males in suicide rates from 2005 to 2019 [3]. Studies conducted in China identified some unique risk factors for suicide, such as being female [4], rural residence [5], single status [5], religious belief [6], mental disorders [6], high chronic stress [7], low social support [8], high impulsivity [9], severe psychological strain [10], conflicts with family members [5], and previous suicidal behavior [11]. A growing body of studies used the strain theory of suicide to explain suicide in China and showed evidence that psychological strains were significantly associated with suicide [2,12,13]. For suicide risk factors at the individual-familial level, most studies explored the impact of socioeconomic status on suicide, and few studies analyzed the impact of position in family on suicide. For example, Kim et al. studied the relation between socioeconomic inequalities and suicidal ideation, parasuicide, and completed suicide in South Korea [14]. They found that both absolute and relative inequalities in socioeconomic position were highly correlated with suicidal ideation and mortality. Moreover, Iemmi et al. conducted a systematic review to understand the association between suicidal ideations, behaviors, and economic poverty in low-income and middle-income countries [15]. They found that poor economic status, diminished wealth, and unemployment were associated with suicidal ideations and behaviors. There are few studies on the relationship between socioeconomic status and suicide risk among the Chinese population. For example, low educational level, unemployment, and low income were suggested to be risk factors behind suicide [12,16]. Zhang and Tao investigated the relationship between family socioeconomic status and the psychopathology of Chinese college students and found that perceived family socioeconomic status was positively correlated with suicide ideation among college students [17]. In this study, we aimed to provide new evidence of the impact of personal position in family on suicide risk among Chinese rural youth. There is politics in the family, which is called family politics. Politics is not only about who is more powerful in the family but also about whether family members can be harmonious. In the power structure of this family, some members are more powerful in the family, while some are in a disadvantageous position. Mental disorders and suicide are often the result of games of power in family politics [18]. In the power structure of this family, family members have different contributions to the family, and thus their positions within the family might be different. There are many factors that could affect the family position which might be related to perceived honor, finances, gender, family roles, and so on. Different individuals and families may value family position from different aspects, and the data in this study reflects the subjective general estimation from the respondents on family position. Recent psychological and sociological researches have shown that subjective experience has direct impacts on personal behaviors, such as personal health and health behavior [19]. This study focused on the general estimation of family position from family members, and the factors affecting the estimation of family position warrant future researches. We also explored potential mediating mechanisms for the relationship between position in family and suicide risk from the perspectives of mental health, social support, and coping strains, which were neglected in the existing literature. We used the psychological autopsy method to compare the position in family between young adults who committed suicide and living controls from the same village. The PA method, known as "a procedure for the reconstruction of suicidal death through interviews with survivors" [20,21], has been widely used in Western countries for suicide studies. This method is shown to be an equally applicable method for the study of completed suicides in China [16] and has been suggested to be the optimal research method given the environment of Chinese rural suicides [22]. --- Methods --- Study Population and Design. The data for this study was derived from a case-control psychological autopsy study. The survey was designed to explore potential risk factors for suicide among Chinese rural young adults, such as mental disorders, impulsivity, and strain. The target population of the study was rural young adults aged 15-34 years old who died by suicide in comparison with community-living controls from the same village. In this study, we focused on the association between family position and suicide risk. --- Sampling. The survey was carried out in three provinces in China, including Liaoning, Hunan, and Shandong. A total of 16 counties were randomly selected in the 3 provinces, with 6 in Liaoning, 5 in Hunan, and 5 in Shandong. Within each county, all village doctors were trained on study procedures and were required to report suicide deaths to the local Centers for Disease Control and Prevention by telephone or fax within 24 hours after the suicide occurred. If suicidal deaths were not recognized by any health agency, village treasurers , who collected fees for each burial or cremation and were aware of all deaths in the village, were required to notify the county CDC. To ensure that no cases of suicide were missed, we conducted investigations with the village board and villagers whenever necessary. Subsequently, all the suicide information gathered at the county CDCs was transferred monthly to the provincial CDC. Finally, we collected 393 suicide cases among 15-34-yearolds from October 2005 to June 2008, with 178 female and 214 male subjects. We randomly selected living comparison individuals within the same age range and the same county, based on the 2005 census database of the 16 countries. Finally, 416 comparison subjects were collected, with 214 females and 202 males. --- 2. 3. Information Sources. For each suicide case and each control, we carefully selected two informants to interview who were 18 years or older and most familiar with the subject's life and circumstances. The first informant was a parent, a spouse, or an important family member. The second informant was a friend, coworker, or neighbor. The comparison individuals were also interviewed. --- Interviewing Procedures. Informants were first approached by the local health agency or the village administration and were notified of the upcoming interview. Upon their agreement through written informed consent, the interview was arranged 2-6 months after the suicide. Informants had the opportunity to decline participation, and none of them declined participation. The control and informants were interviewed face to face by the trained interviewer in a private place in a hospital/clinic or their home. The average duration time for each interview was 2.5 hours. --- Measures. The main outcome variable in this study was the dummy variable of suicide risk. The key variable of interest was the personal position in family, which was assessed by the question "How do you evaluate his/her position in the family?" Respondents were asked to tick one out of the five proposed answers: "highest/higher/general/lower/lowest." We combined the first two categories into "high," and the last two into "low." Personal economic status in a family is measured by personal annual income, whether an individual has income or not, and whether individual income is higher than the family average level or not. We estimate the role of personal economic status in the family on suicide risk to identify whether the impact of personal position in family on suicide stems from personal economic status in family. Control variables in the analysis included gender, age, educational level, marital status, work status, and family economic status. Age was set as a dummy variable: "<25" and "≥25." Educational level was categorized into three groups: "elementary school and below," "primary school," and "high school and above." Marital status was categorized into two groups: "currently not married" and "currently married." Work status was dichotomized as "employed" and "not employed." Family economic status was measured by family annual per capita income with the Chinese renminbi , which was divided into three levels: "<10,000 yuan," "10,000-19,999 yuan," and "≥20,000 yuan" . Furthermore, this study tried to explore potential mediating mechanisms for the relationship between position in family and suicide risk from the perspectives of mental health, social support, and coping strains. First, mental disorders were defined as 1 if the case/control had at least one category of mental diseases diagnosed by psychiatrists and 0 if otherwise. We used the Chinese version of the Structured Clinical Interview for the DSM-III-R to measure axis I diagnoses for cases/controls [23,24]. Diagnoses were made by the psychiatrists on each interview team in consensus meetings at which all responses from each informant were presented by the interviewers. Five categories of mental disorders were diagnosed, including mood disorders, schizophrenia and other psychotic disorders, anxiety disorders, substance use disorders, and other axis I disorders . Second, social support was assessed with the perceived social support subscale in the Duke Social Support Index [25], which reflects the participants' perception of his or her social support from family members and friends . The level of social support was divided into three groups based on the 33.3rd and 66.6th percentiles among the whole sample. The Chinese version of the DSSI has been validated in earlier studies [26,27]. Third, the coping strain was measured by the Coping Responses Inventory , which was developed by Moos [28]. When facing a life crisis, people would experience coping strain if they were not able to cope with it [10]. Subjects were assumed to suffer coping strain if they had low coping skills. The CRI asked respondents to evaluate the target's frequency of engaging in 48 skill activities. Out of the 48 items, 24 measured positive skill activities which were protective factors of suicide and 24 measured negative ones reflecting risk factors . The responses to questions on negative skill activities were assigned to an integer value of 0 to 3, and the responses to the questions on positive skill activities were recoded as 3 to 0. Finally, we obtained total scores of coping strain by adding up the overall score of the 48 items, with higher scores indicating higher coping strain. The approach coping subscale has been tested in the rural Chinese population with good validity and reliability [10,29]. PA was a systemic method to study risk factors preceding suicide and was commonly used in Western countries [30,31]. Information on demographics, family and society circumstances, and personal psychological status was collected from proxy respondents which made the information relatively objective [30,31]. In PA studies, including the current one, many information was collected via the informants of participants. Therefore, the validity of using proxy data and the reliability and validity of the methodology and instruments were tested before the study [30][31][32]. The results indicated that PA was a valid method to study suicide in China [30][31][32], and the information provided by the informants might not be perfect but was reliable for PA study in China [30][31][32]. 2.6. Statistical Analyses. STATA was used for data analysis. T-test and chi-square tests were used to compare differences in continuous and categorical variables between cases and controls, respectively. For study outcome part 1, we employed multiple logistic regression models to estimate the coefficients of position in family in relation to suicide, adjusting for individual and family characteristic factors. For study outcome part 2, we tried to distinguish the role of personal economic status and noneconomic status in the family on suicide risk. For the first step, we added separate variables of personal annual income and personal relative income in the family instead of personal status in the family to estimate the effect of personal economic status in the family on suicide risk. Since personal status includes the individual's economic and noneconomic status, the coefficient of personal status in the family represents the role of personal noneconomic status in the family if personal economic status in the family is controlled. For the second step, we added both measures of personal status in the family and the personal economic status in the family within the model to estimate the coefficient of personal status in the family after adjusting for personal economic status in the family. If the coefficient of personal status in the family remains significant, it provides evidence that personal noneconomic status in the family plays a significant role in suicide risk. Hence, in the second step, we added variables of individual income, whether one has individual income or not, and whether individual income is higher than the family average level or not into the model gradually from the first step to estimate the role of individual economic position on suicide risk. This result would help to 3 Depression and Anxiety interpret whether the effect of position in family on suicide risk is related to personal economic status. For study outcome part 3, we performed sensitivity analyses by examining heterogeneity among subgroups by marital status and educational level. For study outcome part 4, we employed the strategy of identifying mediating mechanisms using the stepwise regression method. We added covariates gradually into the model from the first step and then examined the association among those covariates and suicide risk, as well as how the coefficients of position in family would change. The final model included all indicators simultaneously. The added covariates included mental health, social support, and coping strain. For the strategy of identifying a mediating mechanism, if the covariate is added into the model and the coefficients of position in family become smaller or less significant, then the covariate is proved to be an important mediator in the pathways of position in family affecting suicide risk. This strategy of identifying mediating mechanisms has been widely used in the literature [33][34][35]. In addition, we employ the Karlson-Holm-Breen method which was developed by Karlson et al. [36] to further explore the contributions of mediators to the association between low position in family and suicide risk. We also disentangle the contributions of mediators to identify which of the mediators contribute most to the confounding. --- Results --- General Information. The data collection yielded 392 suicide cases and 416 community living controls in rural China. All the subjects were aged between 15 and 34 years at the time of death or interview. As shown in Table 1, suicide cases were more likely to have a low position in family. For example, 11.24% and 16.82% of female and male cases had lower positions in family, while the corresponding proportions for living females and males were only 1.87% and 1.49%, respectively. Suicides were more likely to occur among the unmarried. There was no significant difference in working status between suicide cases and controls. Both suicide females and males had lower educational levels and were more likely to be from families with lower annual income than controls. There was no significant difference in personal annual income between female cases and female controls, while male cases had significantly lower annual income than their controls. Compared with living females, more female cases had a lower annual income than the family average annual income , while there was no significant difference between male cases and their controls. Compared with living controls, more cases had mental disorders, lower social support, and lower coping skills. --- Multivariate Analysis 3.2.1. Impact of Position in Family on Suicide Risk. We performed logistic regressions to estimate the association between position in family and suicide for females and males separately and reported ORs in Table 2. The table showed suicide risk estimates for demographic factors, family income, and position in family, with the first two columns for females and the last two columns for males. As Table 2 illustrates, males aged 25 years and above had a significantly higher risk of suicide than males younger than 25 years old , while it was not significant among females . Lower educational levels and unmarried status were strong predictors of suicide, and the effects were stronger among males than among females. Family economic status was highly associated with suicide. Females and males from low-income families were more likely to commit suicide than individuals from high-income families . Column 2 and column 4 in Table 2 showed that the ORs of low position in family were 7.1 for females and 9.1 for males. In other words, compared with females having high positions in family, females with low positions in family were 7.1 times more likely to commit suicide. --- Role of Personal Economic Status on Suicide Risk. Table 3 showed that the coefficients of personal economic status variables were not significant at the 5% level both for females and males , and coefficients of position in family remained the same as those in Table 2 . It indicated that personal income was not directly associated with suicide risk after controlling for family income per capita income and other control variables, indicating that the impact of personal position in family on suicide did not stem from personal economic status. --- Subgroup Analysis. Probing for heterogeneity , we found that position in family had a more significant impact on suicide risk among the married population than among the unmarried population . For education heterogeneity, we found that the impact of low position in family on suicide risk was strongest among people with a primary school level of below , followed by people with a middle high school level , and people with a senior high school level and above . --- Mediating Factors. Furthermore, we explored the association between mental health, social support, coping strain, and suicide risk and examined how the coefficients of position in family would change when these variables were added to the model. In Table 5, we added the above variables gradually to identify the potential mediators for females in columns 1-4 and males in columns 5-8. were included in the model. In columns 2-4, new covariate variables were added into the model gradually, including mental disorder, social support, and coping strain. In column 4, all the covariates were included in the model.) --- Depression and Anxiety We added the variable of mental disorders in column 2 for females and column 6 for males and found that females and males with mental disorders were 13.36 times and 32.51 times more likely to commit suicide in comparison to living subjects, respectively . The OR magnitude and significance of low position in family declined after the variable of mental disorders was added in the female model , while the corresponding coefficient was barely affected among males . It indicated that a low position in family might influence suicide risk by affecting the mental health of females. We further added social support variables in column 3 for females and column 7 for males in Table 5. Social support was significantly associated with suicide risk for both females and males. For females, the coefficient of low position in family declined substantially after controlling for social support variables and was no longer significant . For males, there was also a sharp decline in the coefficient of low position in family . The above results indicated that social support might play an important role in mediating the effect of a low position in family on suicide for both females and males. Finally, we added the coping strain variable in column 4 for females and column 8 for males. Females and males having high coping strain were 29 times and 18 times more likely to commit suicide than living controls, respectively. The coefficient of low position in family for males declined to a certain extent after controlling for coping strain . It indicated that coping strain could be a partial mediator in the association between low position in family and the risk of suicide for males. Based on the stepwise regression method to study the mechanism effect, we further employ the KHB method [37] to explore the contributions of mediators to the association between low position in family and suicide risk. Tables 6 and7 reported the average partial effects of a low position in family on suicide before and after controlling for potential mediating factors. On average, the probability of suicide was 13.7 times higher for females with a low position in family compared to those with a high position in family. After controlling for mental health, social support, and coping strain, the difference in suicide probability was no longer significant among individuals with low and high positions in family. Similarly, after controlling for the above three potential mechanisms, the probability of suicide reduced from 34 times to 4 times for males with low positions in family, and the corresponding significance was reduced to a 5% significance level. The above results indicated that mental health, social support, and coping strain are the main mediators underlying the association between suicide risk and low position in family for both females and males. Further, we explore which mediators contribute most to the confounding in Table 7. Column 1 for females and column 2 for males showed the contribution of each mediator to the indirect effect . The results indicated that low coping strain played the most important role underlying the association between low position in family and suicide risk for both females and males, and mental disorders problem contributed to the second, and the third one is low social support. --- Discussion This study provided new evidence that position in family was significantly associated with elevated suicide risk. Females with low positions in family were 7.1 times more likely to commit suicide compared to females with high positions in family. It indicates that 7 Depression and Anxiety position in family is significant in predicting suicide among Chinese young adults. Our results were consistent with the findings of Fei [18], which indicated that suicide was often a result of games of power in family politics. On the one hand, when the power balance is broken in the family and family harmony is unsustainable, suicide might happen. On the other hand, if one family member had an obviously disadvantageous position with less moral capital and economic capital, in other words, he/she owned low power in the family, thus he/she would be confronted with a higher risk of extreme behaviors including suicide. Our study provided evidence that subjective experience of family position has direct impacts on suicide behaviors, which is consistent with Simandan's [19] findings. Our results also showed gender differences, with males more likely to be affected by family position which could be due to the fact that males value "face" more than females do in China [38]. We have to mention that low position in family and suicide may affect each other. Personal position in the family of people who died of suicide might be rated low due to stigmatization, a phenomenon that was common among patients with mental disorders [39,40] and people who died of suicide [41,42]. For example, patients with mental disorders might perceive stigma from other family members [40], and family members might also feel ashamed of suicide behaviors in China [43]. The stigma problem is therefore a big challenge for psychological autopsy studies on suicide. We also provide new evidence that personal economic status in a family is not significantly related to suicide risk after controlling for family income, personal position in the family, and other factors. One possible reason might be the impact of household division of labor. The Chinese culture has a gender division of household labor in the family, especially in Chinese rural families. Normatively, the males head the family, and they are supposed to take charge of external matters and be responsible for the orderly management of the family, while the females have to shoulder more domestic work, such as cooking, laundry, and taking care of family members [44][45][46]. Family members who fail in their "expected tasks" , hence, would encounter a lower position in family regardless of their income. Our findings indicate that in the Chinese traditional context, personal noneconomic status in the family might be more important in predicting suicide risk than economic status. However, we could not rule out the possibility that the general average household/personal income in the locations where the study was carried out has very little variability itself. Hence, our results on the effect of personal economic status in family on suicide risk should be interpreted with caution, and more robust evidence is needed in future research. Subgroup analysis revealed that low position in family had a more significant impact among the married population. A possible explanation might be that the transition from unmarried to married brought more complicated family relationships, and people with low positions in family might undergo negative effects from the other members of the family. If appropriate coping skills and social support are absent, these young married people might become vulnerable and be at high risk of suicide. Results of subgroup analysis also indicated that the impact of low position on suicide risk was stronger among people with low education levels. This might be explained by their low income level, tough life, lack of social support, and limited coping skills. The government and social organizations should pay more attention to this disadvantaged population and provide them with psychological counselling and support. Furthermore, we identified the potential mechanism underlying the association between position in family and suicide risk, which was neglected in the existing literature. Our results indicate that mental disorders, low social support, and low coping skills might be the potential mediating factors that link low position in family to suicide. In the power structure of this family, when family conflicts happen, both female and male young adults with low family positions easily plunge into complex and confusing personal dilemmas. This could cause strong mental stress and even mental disorders. If they have a low level of social support and inadequate coping skills, they might conduct extreme behaviors. For suicide intervention, our findings suggest that family education on how to maintain family harmony, to deal with family conflicts, and to improve coping skills might be useful. For families with severe family conflicts, local community officers should pay more attention and provide help in mediating and counselling. The findings in this study have significant implications for future suicide prevention/intervention and future 9 Depression and Anxiety research. First, for families with severe conflicts among family members, the community neighborhood committee should pay more attention and provide prompt help and psychological counselling. This could help dilute family conflicts and avoid extreme risky behaviors. Second, interventions for family members with low family positions should be considered. For example, we find that young married people with low family positions are at higher risk of suicide behavior, which indicates that education programs and counselling on marriage and family are quite important. The government and social organizations could provide training classes for young married couples on how to get along with their spouse's family members, how to adapt to the new family, how to deal with family conflicts, etc. In addition, families with lower economic status tend to have more life difficulties which might lead to family tensions, and people with lower income levels usually have limited education attainment and coping skills to deal with family conflicts. Therefore, more welfare, social security, and social support should be provided to these disadvantaged people. Third, for precise and tailored suicide prevention, identifying factors associated with low family position is important, as this might give further clues to the mediators. There are many factors that might be related to one's family position. For example, some family members might be the main source of income for the family and thus earn a high position in family [47]; some family members might have high moral capital and reputation in the family which could help to maintain and improve family harmony and thus might have a high position in family [48]; in some families with son preference, women might have a low family position if they gave birth to daughters and had no son [49,50]. The reasons for the low family position might be complex which warrants further investigations. There were several limitations in this study. First, the nature of the case-control design could result in recall bias due to the retrospective assessments by the participants. Besides, the participants might be emotional and hard to be interviewed due to bereavement shortly after the suicide death. Therefore, we conducted the interview 2-6 months after the suicide death, when the impacts of bereavement and recall bias were relatively small. The problem of recall bias is also a major challenge for other psychological autopsy studies of suicide [51]. Second, we explored three potential mediators underlying the association between position in family and suicide risk. Despite the evidence, the coefficients should not necessarily be interpreted as causal effects because mediating factors and an individual's position in family are correlated. Thus, a causal inference is necessary for further study, and more work needs to be done with respect to the gender difference in mediating mechanisms. Third, the data used in this study was collected nearly ten years ago. China was one of the few countries where suicide rates were higher among females than among males, and suicide rates have declined with an increased male-tofemale ratio in the past decades [3]. However, the gender ratio of 1.56 is still smaller than that in Western countries where the ratio is nearly 3 to 4, which means female suicide rates are yet relatively high in China [2]. Therefore, the findings in this study are still meaningful for suicide prevention among Chinese females. Future studies using more recent data to further examine the association between family position and suicide are needed. To the best of our knowledge, this is the first study exploring the association between personal position in family and suicide risk among the Chinese population. The results demonstrate a significant impact of personal position in family on suicide risk among Chinese rural young adults. People with lower positions in family suffer a higher suicide risk. We also provide evidence that the possible mechanisms underlying the effect of personal position in family on suicide risk include mental status, social support, and coping strain. The findings have important implications for suicide prevention in China and other developing countries; tailored intervention within the family should be considered, and more attention and help for family members with low family positions from the community and society are warranted. --- Data Availability Data sharing is not applicable to this article as no new data were created or analyzed in this study. --- Ethical Approval This study was approved by the medical ethics committees of Buffalo State, State University of New York, and Shandong University. --- Disclosure None of the funding agencies had any role in study design, in the collection, analysis and interpretation of data, in the writing of the paper or in the decision to submit the paper for publication. --- Conflicts of Interest The authors have no conflicts of interest to report. --- Authors' Contributions

Introduction. Suicide remains an important public health issue in China. Existing literature on the relationship between individual-familial variables and suicide risk mainly focused on family socioeconomic status, and few studies analyzed the effect of position in family on suicide. In this study, we aimed to explore the association between position in family and suicide among Chinese rural youths. Methods. We conducted a case-control psychological autopsy study. The data collection yielded 392 suicide cases aged 15-34 years and 416 community living controls within the same age range. Personal position in family was assessed by the question "How do you evaluate his/her position in the family?" and categorized as high, general, and low to reflect the relative position in family. Logistic regression models were conducted to explore the association between position in family and suicide risk. Results. The results showed that compared with females (males) having a high position in family, females (males) with a low position in family were 7.1 (9.1) times more likely to commit suicide (p < 0 01). Mental disorders, social support, and coping strain were potentially important mediating factors linking position in family to suicide, with certain heterogeneity among males and females. Low coping strain played the most important role in underlying the association between a low position in family and suicide for both females and males, accounting for 55% (28%) of the contribution to the total effect for females (males). Subgroup analysis revealed that low position in family had more significant impacts on suicide risk among married youths and those with low education levels. Conclusion. The effect of position in family on suicide should receive greater consideration when predicting suicide in rural China. Possible mechanisms underlying the effect of position in family on suicide include mental status, social support, and coping strain.

Introduction Oral health has been considered a multifaceted concept that is continuously influenced by the values and attitudes of people and communities, reflecting their physical, social, and psychological attributes that are essential to quality of life [1]. Thus, quality of life has been considered an important attribute of patient's evaluation of their oral health and reorientation of health services [1]. Oral health-related quality of life is a multidimensional construct that reflects the extent to which oral disorders affect individuals' daily functions and well-being [2]. Therefore, oral health emerges as a positive concept highlighting the personal and social resources available to individuals, such as social capital. Social capital can be defined as the resources embedded in social networks, which can be accessed and mobilized through the existing social ties within or among one's networks [3][4][5][6]. The previous literature suggests that social capital may be distinguished into structural and cognitive dimensions, which correspond to the quantitative and qualitative aspects of social capital, respectively [3]. The former dimension refers to the extent and intensity of one's participation in associations or other forms of social activity. The cognitive dimension is less tangible and relates to people's perception of interpersonal trust, solidarity, and reciprocity. Thus, social capital involves the quantity and quality of accessible resources and benefits among individuals or groups according to their social networks [3][4][5][6]. Social capital and its proxy indicators have been associated with different clinical and subjective oral health measures. The previous literature has shown that individuals with lower social capital were more likely to have dental caries, gingivitis, and worse selfrated oral health and OHRQoL [7][8][9][10]. Previous studies have also verified the relationship of different dimensions of social capital with general health, such as frailty [11], mental health [12], and self-perceived health [13]. Other research has also highlighted the impact of different dimensions of social capital on oral health outcomes in adults and elderly people, including edentulism [14], dental pain [15], and self-perception of oral health [16]. Although previous studies have established the relationship between social capital and oral health outcomes, the possible impact of social capital on OHRQoL during adolescence has been little explored. Adolescence is characterized by biopsychosocial changes, when individuals tend to initiate and establish their own social relationships according to their involvement and engagement in different social groups [17,18]. Further investigations aiming to enhance the understanding of which aspects of social capital are more relevant to oral health in this age group can be useful to develop public health actions to promote oral health and quality of life. Furthermore, elucidating the impact of different aspects of social capital on health has been encouraged by the previous literature [19]. Thus, this study aimed to evaluate the relationship of structural and cognitive social capital with OHRQoL among adolescents. It was hypothesized that adolescents with lower structural and lower cognitive social capital are more likely to report a greater impact on OHRQoL. --- Methods --- Study Design and Sample This was a cross-sectional study nested in a 10-year cohort study with preschool children from the city of Santa Maria, Brazil. Baseline data collection was conducted in 2010 and follow-ups were in 2012, 2017, and 2020. Further methodological details of this cohort study are described elsewhere [10,20]. This study used data from the data collection carried out in 2020. The baseline sample was obtained during National Children's Vaccination Day in health centres that were responsible for the vaccination coverage of 90% of children in the city. All healthcare centres with dental offices were initially selected as sampling units. Subsequently, every fifth child in the vaccination line was systematically selected and invited to participate. The selected health centres were located in different regions of the city involving populations with different socioeconomic backgrounds. A total of 639 children aged from 1 to 5 years were examined in 2010. Of them, 429 adolescents were assessed in 2020. All individuals who participated in the baseline were invited to participate in 10-year follow-up, with ages ranging from 11 to 15 years old during this period. Data collection of this wave was performed between November 2019 and January 2021, though it was interrupted between March and October 2020 due to the COVID-19 pandemic. Adolescents were evaluated in their schools or households, according to the information available in the database. Participants and their families were contacted via social media if they were not found using the initial strategy. The power of the study was calculated in OpenEpi [21] through a post hoc power calculation and considered an alpha error probability of 0.05 and a 95% confidence interval. Overall, adolescents' CPQ11-14 scores were compared between the non-exposed group and the exposed group according to all social capital variables used, resulting in a study power that ranged from 80 to 100%. --- Data Collection and Variables All data collection procedures were based on the international standardized criteria for oral health surveys [22][23][24][25]. Data were collected through interviews using structured questionnaires and through oral clinical examinations performed by a trained research team. OHRQoL was assessed using the Brazilian short-version of Child Perceptions Questionnaire 11-14 [22]. This questionnaire is composed of 16 items, grouped into 4 different domains: oral symptoms, functional limitation, emotional well-being, and social well-being. All questions are responded to considering a Likert scale from 0 to 4 points. The overall CPQ11-14 score is made by the sum of the scores of the 16 items. The overall CPQ11-14 scores were used in the data analysis and can range from 0 to 64 points, with higher scores indicating a lower level of quality of life. Individual social capital was assessed considering the structural and cognitive dimensions [3]. The structural social capital reflects the extent and intensity of participation in associations or other forms of social activity, whereas the cognitive side reflects people's perception of interpersonal trust, solidarity, and reciprocity [3,6]. The structural dimension was evaluated according to the social networks with the following questions: "How often do you attend group religious activities?", and "In the last 12 months, how often have you visited or received visits from friends and neighbours?", with the possible response options including the following: at least once a month; and less than once a month or never. The cognitive dimension was assessed by the following questions: "Do you think your friends and neighbours can be trusted?", yes or no; "Do people in your neighbourhood have good relations with each other?", yes or no; and "If something bad happens to you, would someone help you in this situation?", yes or no. These items have been commonly used as proxy measures of structural and cognitive social capital according to the previous literature [7,9,10,23,24]. Demographic, socioeconomic, and clinical data were collected as possible confounding factors on the relationship between social capital and OHRQoL [2,10]. Demographic characteristics included sex and age . Socioeconomic status was assessed according to monthly household income based on the sum of all income sources of the family in the previous month in Brazilian Reais . For data analysis, household income was categorized into <1 Brazilian minimum wages and ≥1 BMW. One BMW was the equivalent to USD 220 in 2020. Dental caries were assessed with calibrated examiners using the International Caries Detection and Assessment System criteria [25], in which all dental surfaces were evaluated. Dental caries with cavitated lesions were grouped as present or absent . A total of seven examiners conducted clinical examinations. All clinical examiners were dentists who were previously trained and calibrated for ICDAS evaluation, totaling a process of 36 h, including theoretical training, photographic image evaluation, exercise with exfoliated teeth, and clinical evaluations. The inter-and intra-examiner Kappa coefficients for ICDAS ranged from 0.70 to 0.96. --- Ethical Aspects This project was approved by the Research Ethics Committee of the Federal University of Santa Maria . All guardians signed a written informed consent agreeing to have their son take part in the study. The adolescents also signed a consent form to participate in the study. --- Data Analysis Data were analysed using the STATA 14 . A descriptive analysis was carried out for the main sample characteristics. OHRQoL were also described according to sociodemographic, social capital, and clinical variables. The characteristics of adolescents between those who completed the follow-up and dropouts from the original baseline sample were compared using the Chi-square test and the t-test for categorical and continuous variables, respectively. Similar tests were used to compare data of adolescents before and during the COVID-19 pandemic. The study outcome was the overall CPQ11-14 scores. All descriptive analysis was performed considering the sample weight by 'svy' command on STATA for complex data samples. The simple comparison between overall CPQ11-14 scores according to sample characteristics was performed using the Mann-Whitney test. Multilevel Poisson regression analysis was performed to estimate the relationship of structural and cognitive social capital on the OHRQoL of adolescents. The construction of multilevel analysis considered adolescents nested into 15 neighbourhoods . The analysis structure considered the fixed effect with the random intercept. Demographics , socioeconomic , and the clinical variable were considered as possible confounders, as suggested by the previous literature on the predictors of OHRQoL [2,10]. Variables that presented p ≤ 0.20 in the unadjusted analysis were included in the adjusted model. The results are presented as Rate Ratio and 95% confidence intervals . --- Results The sample comprised 429 adolescents, representing 67.1% of those evaluated at the cohort's baseline. There were no statistical differences between the analysed sample and dropouts individuals for the evaluated characteristics . Data obtained before and during the COVID-19 pandemic did also not differ . Table 1 presents the descriptive characteristics of the adolescents. The sample was balanced between boys and girls, and the mean age was 12.6 years. Most individuals were from families with monthly household income > 1 BMW . According to structural social capital variables, 45.2% and 95.1% of the adolescents attend religious meetings and visit friends and neighbours at least once a month, respectively. Considering cognitive social capital, 48.7% reported trusting in friends and neighbours, 57.8% believed that neighbours had good relationships, and 90.1% reported support during hard times. OHRQoL scores were reported according to sociodemographic, social capital, and clinical characteristics . The mean of the overall CPQ11-14 scores was 11.2 . Higher OHRQoL scores were found in girls, individuals from families with low household income, and those who presented with dental caries. OHRQoL scores were also higher in individuals with lower structural and cognitive social capital considering all indicator variables. The highest OHRQoL scores were among participants reporting no support during hard times . Table 3 presents the unadjusted and adjusted analysis between the social capital variables and OHRQoL scores. All independent variables were associated with OHRQoL in the unadjusted analysis. In the adjusted analysis, adolescents who attended religious meetings less than once a month or never presented with OHRQoL scores 12% higher than their equivalents . Considering the cognitive social capital variables, adolescents who did not trust their friends and neighbours and those who believed that their neighbours did not have good relationships were also more likely to present with worse OHRQoL. Adolescents reporting no one to support them during hard times presented with OHRQoL scores 24% higher than those who had this support . In addition, female adolescents and those from low-income families and with dental caries were also more likely to report poor OHRQoL. --- Discussion This study aimed to evaluate the different roles of structural and cognitive dimensions of social capital on OHRQoL among adolescents. Our findings confirmed the hypothesis that both dimensions of social capital were associated with OHRQoL. Despite that the previous literature has evidenced the association between these social capital dimensions and oral health [11][12][13][14][15][16], the role of these factors in OHRQoL during adolescence has not been explored yet. Overall, our findings showed that OHRQoL was poorer in individuals who presented with lower levels of structural and cognitive social capital. In this study, social capital variables were measured through indicators of social networks, social support, and trust, which have been considered proxy measures to assess social capital at the individual level [23,24]. In general, individuals reporting some source of social capital tend to have better general and oral health outcomes, as reported in previous studies [9,10,[26][27][28]. This may be explained due to the fact that individuals with high social capital tend to experience positive 'peer pressure' to adopt healthy habits, increase the use of health services, and have less stress through the protective effects of social support [29], which may impact their oral health and quality of life. Considering the structural dimension of social capital, our results showed that adolescents who attended religious meetings less than once a month or never presented with poorer OHRQoL than those who attended religious meeting more regularly. The structural social capital reflects the extent and intensity of participation in associations or other forms of social activity [3]. On the other hand, the frequency of visit to friends or neighbours were not related to OHRQoL in the adjusted analysis. This unexpected finding can be explained due to the fact that interpersonal relationships such as visiting friends and neighbours were very common among the participants. According to our results, almost the entire sample reported visits to friends and neighbours at least once a month, which may have influenced the results despite being considered an important proxy measure of social capital [23,24]. In contrast, attendance of religious meetings impacted adolescent's OHRQoL. A previous study suggested that access to dental care and the adoption of healthy behaviour related to oral health are influenced by religiosity, which positively impacted individuals' oral health [30]. It has also been shown that religiosity and the presence of churches in the neighbourhood are related to better oral health outcomes in children and adolescents [9,31,32]. Thus, this type of social tie may generate social resources that benefit the individual by improving social support and coping [29], which directly impacts subjective health outcomes, such as OHRQoL. According to our findings, all cognitive social capital indicators were related to OHRQoL scores. Adolescents who did not trust their friends and neighbours and who believed that their neighbours did not have good relationships with each other presented with worse OHRQoL. The cognitive social capital is considered the qualitative component of social capital and refers to people's perception of interpersonal trust, solidarity, and reciprocity [3]. Previous studies have shown that high cognitive social capital was associated with better subjective health outcomes, such as mental health and self-perceived oral health [12,16], which is in accordance with our findings. These findings may be explained by the fact that high levels of interpersonal and neighbourhood trust may act as a protective factor that attenuates the negative effects of stress through feelings of security and belonging, [29,33] and, consequently, mitigates the impact of oral problems on an individuals' quality of life. Our findings also demonstrated that adolescents who reported a lack of support during hard times presented with poorer OHRQoL. This cognitive indicator of social capital exerted the greatest impact on OHRQoL. Social support exerts a fundamental role during adolescence since it is related to other psychosocial factors, such as sense of coherence, self-management, and personal strength to deal with life problems [34]. These factors have a significant impact on oral health outcomes at this stage of life [35][36][37], where individuals experience biopsychosocial changes and emotional adaptations according to social structures [18,19,36]. Therefore, cognitive social capital seems to represent a relevant aspect of OHRQoL among adolescents. Although this study focused on the different dimensions of social capital, it is important to emphasize the fact that there is a reciprocal relationship between the structural and cognitive dimensions of social capital. In this sense, studies have shown that social networks and social participation may lead to increased levels of social trust [6,38]. Likewise, it is reasonable to assume that individuals with high levels of social trust are more likely to socialize with others across different types of social networks [38]. Notwithstanding, despite the reciprocity among the social capital dimensions, our findings cautiously assume that, in adolescence, trust and social support operates as a central component in the composition of social capital. This study has some limitations that need to be addressed. First, our study considered only some proxy variables of structural and cognitive dimensions of social capital. However, measuring all sources of social capital was out of the scope of this study. The social capital measures examined in this study have been used in previous research [7,9,10,23,24]. Furthermore, the losses to follow-up of adolescents between baseline and the studied sample may have impacted the external validity of our results. However, attrition analyses showed that the analysed participants were similar and representative to the baseline sample when the main characteristics were compared. Another limitation is due to some missing data in our database among followed individuals. However, this concern occurred in few cases and Bootstrap sensitivity analyses confirmed that missing data did not affect the validity of our results. Despite the above-mentioned limitations, this study also has strengths that should be acknowledged, including the selection of a representative sample of the population of the city and the use of robust methods of data collection and analysis. Investigations of social capital and its different dimensions during adolescence is very relevant since the individuals build relationships and networks of trust during this period of life, which can impact their health throughout their life course. Moreover, this study also included patientreported outcomes, which have been increasingly incorporated into epidemiologic studies in oral health. Public health policies to promote social capital in schools or communities are encouraged, since stimulating interpersonal social networks and social trust can directly benefit the general and oral health of adolescents and their living environment. Future longitudinal and interventional studies considering these factors are also encouraged to enhance the understanding of these interrelationships in more depth. --- Conclusions Our findings suggest that OHRQoL was poorer in adolescents who presented with lower structural and cognitive social capital, with the greater impact on OHRQoL related to the cognitive dimension. These conclusions are important for public health policies to promote social capital, aiming to improve oral health and quality of life in this population. --- Data Availability Statement: Data available on request from the authors. --- Acknowledgments: The authors thank all the scholars, their parents, and schools that took part in this study, as well as the Health and Education Authorities from Santa Maria, Rio Grande do Sul, for all information and authorization. --- Author Contributions: J.K.K. designed the study, designed the data collection instruments, collected data, carried out the analyses, drafted, and revised the manuscript. B.B. designed the study, collected the data, and revised the manuscript. M.V.V. and B.E. conceptualized and designed the study and critically reviewed the manuscript. T.M.A. conceptualized and designed the study, coordinated and supervised data collection, and critically reviewed the manuscript. All authors have read and agreed to the published version of the manuscript. Funding: This study was funded by the Conselho Nacional de Desenvolvimento Científico e Tecnológico , by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior-Brasil -Finance Code 001 and Fundação de Amparo à Pesquisa do Estado do Rio Grande do Sul . Institutional Review Board Statement: All procedures performed in studies involving human participants were in accordance with the ethical standards of the Human Research Ethics Committee of the Federal University of Santa Maria , Brazil. Informed Consent Statement: Informed consent was obtained from all individual participants included in the study. ---

This study evaluated the relationship of structural and cognitive dimensions of social capital with oral health-related quality of life (OHRQoL) among adolescents. This was a cross-sectional study nested in a cohort of adolescents from southern Brazil. OHRQoL was evaluated using the short version of the Child Perceptions Questionnaire 11-14 (CPQ11-14). Structural social capital was measured by attendance of religious meetings and social networks from friends and neighbours. Cognitive social capital was evaluated through trust in friends and neighbours, perception of relationships in the neighbourhood, and social support during hard times. Multilevel Poisson regression analysis was performed to estimate the association between social capital dimensions and overall CPQ11-14 scores; higher scores corresponded to worse OHRQoL. The sample comprised 429 adolescents with a mean age of 12 years. Adolescents who attended religious meetings less than once a month or never presented higher overall CPQ11-14 scores. Adolescents who did not trust their friends and neighbours, those who believe that their neighbours did not have good relationships, and those reporting no support during hard times also presented higher overall CPQ11-14 scores. OHRQoL was poorer in individuals who presented lower structural and cognitive social capital, with the greatest impact related to the cognitive dimension.

INTRODUCTION Q UANTITATIVE analysis of epidemic processes such as in- fectious diseases, malware codes, and rumors spreading over physical and online social networks has stimulated intense research activities [1], [2]. Owing to the pervasive use of social media and the abundance of data extracted from several such networks, which for long were merely unavailable, the theoretical perception of epidemic dynamics driven by nodal interactions has refined substantially in recent years [3], [4]. While the vast majority of research has scrutinized only positive social relationships, user pairs may also signify enmity or distrust as perceived in reality. Subsequently, a user may decisively decline to interact with a hostile contact and avoid involvement in further spread of the viral process [5]. Accounting for heterogeneous social interactions is thus, crucial in characterizing social link valence evolution under the influence of individual user's attitudes towards viral spread. Unlike conventional networks, signed SNs evolve based on the structural balance theory, pioneered by Heider [6], where the relationship between any two users in a triad can be impacted by the third user [7]. That is to say, the theory posits that if "the friend of my friend is my friend" and "the friend of my enemy is my enemy", then the resulting triad will be balanced and will constitute an odd number of friendly links. Evidently, SSNs converge to structurally balanced states with minimum social tension by flipping the link polarity to maximize the number of balanced triads [8]. In the jargon of networked epidemics, a handful of works focus on edge sign reconfiguration under the effect of evolving user states. The conditions to attain opinion convergence in generic SNs are obtained in [9] using monotone dynamical systems. Further extended in [10], Shi et al. analyze the asymptotic user state evolution affected by deterministic weights on pairwise interactions by formulating a relativestate-flipping model for consensus dynamics in random SSNs and prove the conditions leading to almost sure convergence and divergence. Their analysis assumes that the initial network structure is always balanced which in truth, may not always be the case. Saeedian et al. [11] study the non-trivial coupled dynamics over a complete signed graph using an energy function. The authors adopt the susceptible-infected epidemic model to study the local and global energy minima of the system irrespective of the possibility of recovery to susceptibility or epidemic alertness. Lee et al. [12] then introduce an adaptive susceptible-infected-susceptible model to reinforce transitivity by rewiring the links between susceptible and infected nodes rather than their signs. Though insightful, the emergent behaviors of the parallel processes in [12] are limited to the population level and do not address the microscopic dynamics inherent in user interactions. Zhang et al. [13] present an approximation algorithm for the minimum partial positive influence seeding problem in viral marketing. Moreover, Li et al. [14] propose a non-stochastic computational model for maximizing polarity-related linear influence diffusion in SNs. However, neither [13] nor [14] subsume Heider's theory in the network structural evolution. Thus far, there exists no work that investigates the intriguing co-evolution of generic SN structures and epidemic dynamics of a reversible process in conjunction with user alertness. Absent in classical epidemic models, awareness towards viral processes is an intrinsic human response that plays the role of a natural immunization strategy. Such change in hu- man behavior can be induced by learning about the contagion spread from others without having to encounter it firsthand. This hence, results in a coupled situation where an infectious person and information about its presence spread simultaneously when humans react to the presence of the infection. In fact, raising awareness is a widely-practiced control strategy in dynamical systems as it alters the progression of the viral spread [15], [16]. Besides the analytical merits, such a refined projection model may serve beneficial to network administrators, social influencers, and decision makers in devising optimal resource allocation policies. In view of this research gap, the main contributions of this work are as follows: • Inspired by Heider's balance theory, a novel energybased framework is proposed to jointly minimize the number of unbalanced triads that contribute to the social tension while mitigating the viral spread in SNs. To capture users' response to such processes, we formulate the susceptible-alert-infected-susceptible epidemic model [15] as a continuous-time Markov chain and derive the stationary probabilities to investigate the virtues of promoting awareness on the network structural evolution. • By incorporating a tuning parameter, we then analyze cases for which the initial fraction of positive links and the initially infected users induce natural immunization by segregating the alerted and infected users into two clusters interconnected via unfriendly links. • Our model is evaluated on three real SSN datasets by employing a time-efficient Monte Carlo simulation method under different parametric settings. --- THE COUPLED NETWORK MODEL In this section, we first formulate the stochastic SAIS model re-purposed for general spreading processes in our analysis. The proposed energy model is then detailed subsequently. --- Stochastic Epidemic Model Description We consider an undirected SSN, represented by the graph G t = , with a set V = {1, 2, . . . , n} of n users that form friendly , hostile , or no social links. The link polarity of user pair ∈ E t at any given time t is denoted by A i,j ∈ {-1, 0, 1} [11]. For epidemic spreading over G t , the SAIS model in Fig. 1 is used, where each user is in the susceptible , alert , or infected state at time t. User i is said to be susceptible if he/she is completely unaware of the spreading process. Since these processes do not propagate over negative links in SNs [17], a susceptible user gets infected with rate β ∈ R + times the number of its infected friendly contacts [15]. A user aware of the process however, is less likely to get infected, with a lower infection rate 0 ≤ β a < β, as compared to a susceptible user. Unlike the irreversible SI model in [11], a susceptible user becomes aware of the process with rate κ ∈ R + times the number of direct infected friends and all infected users may eventually recover back to susceptibility with rate δ ∈ R + . For all i ∈ V, the network state can thus, be expressed formally as the CTMC {X i ; t ≥ 0}, where: X i =    1; if user i is susceptible at time t, 0; if user i is alert at time t, -1; if user i is infected at time t. Using , we now can define the probability of user i being in one of the three epidemic states as S i = Pr[X i = 1], A i = Pr[X i = 0], and I i = Pr[X i = -1] such that for 1 ≤ i ≤ n, S i + A i + I i = 1 always holds. --- Pairwise Spreading Energy Function We now delineate the sign evolution of user interactions in the context of energy. Given the three epidemic states and binary link signs , there exist 12 distinct user pair configurations as shown in Fig. 2. We characterize the viral potency by mapping each user pair configuration , where i, j ∈ V, to the energy landscape as follows: E p i,j      A i,j Xi-Xj 2 4 ; if |X i +X j | mod 2 = 0, A i,j 1-Xi -Xj 2 ; otherwise. ( Based on the functional value of , the configurations depicted in Fig. 2 can be classified as follows: • Balanced edges: As long as configurations S -I and A-I do not flip their edge signs while evolving, the users i and j are in a balanced social relationship and do not engage in the propagation process [11]. Therefore, they exhibit a pairwise energy of E p i,j = -1. • Unbalanced edges: Cases in which a susceptible or alert user is in a friendly relationship with an infected user are socially unstable and are bound to change with time. In our model, S + I and A + I serve as feasible links for epidemic spread and thus, the users are in an unbalanced state with pairwise energy of E p i,j = 1. • Neutral edges: Irrespective of the edge sign, configurations S ± S, I ± I, A ± A, and S ± A do not contribute to the spreading process, and thus, exhibit zero pairwise energy, i.e., E p i,j = 0. Accordingly, the total pairwise spreading energy of network G t , denoted by E p , can be computed as: E p = 1 n 2 i,j i =j E p i,j . --- Triad Structural Energy Function Along with the users' epidemic states, edge sign evolution is also driven by Heider's structural balance criterion. A triad of users in G t , denoted by , is said to be balanced if the product of the constituent triads are balanced. Conforming to Heider's balance theory, the structural status of any triad can be mapped to the energy landscape as below [6]: A i,j •A j,k •A k,i( E i,j,k -A i,j • A j,k • A k,i . With in place, the energy contribution of any balanced triad in G t is E i,j,k = -1 = 1). Fig. 3 showcases all the balanced triads for the SAIS and the SIS models. To converge towards lower energy states , users in unbalanced triads tend to flip their link signs which in turn, affects the configuration of other triads that share common edges with them. Consequently, the total normalized energy of G t , denoted by E △ , is as follows, where -1 ≤ E △ ≤ 1: E △ = 1 n 3 i,j,k i =j =k E i,j,k , Users in triads decide on whether or not to alter their relationships only if the total triad energy of the resulting network is further reduced. Apparently, SNs that manifest triad energy values closer to -1 tend to be socially more stable and thus, pragmatically justified. --- Weighted Network Energy Function We now define the total energy of network G t , given by E, as the weighted sum of the overall pairwise and triad energy functions derived in and , respectively: E = α • E △ + • E p , where α is the tuning parameter used to adjust the energy trade-off between the epidemic spread and the structural balance in the network. Hence, if G t is fully balanced , then E = -1 = 1) , which is more likely to be achieved in smaller graphs. Note that for some fixed α value, attaining the global energy minimum state in which G t is fully balanced depends on the initial fractions of infected users and friendly links [11]. --- STEADY-STATE PROBABILITY DISTRIBUTION Assuming that only one event is triggered in each time step, ∆t ≪ 1, i.e., either the epidemic state of exactly one user pair changes or the edge sign is flipped, the rules defining the user pair transitions are shown in Fig. 4. The balanced edges S-I and A-I transition to S-S and A-S, respectively, with probability δ •∆t or change to S+I and A+I, respectively, with probability 1-δ • ∆t. For unbalanced edges, S+I changes to S-I with probability 1-δ • ∆t - • ∆t • or switches to states A + I, I + I, or S + S with probabilities κ • ∆t, β • ∆t, and δ • ∆t • ∆t), 1 1 1 1 1-2•δ•∆t δ•∆t 1-2•δ•∆t δ•∆t 1-κ•∆t κ•∆t 1-δ•∆t δ•∆t 1-δ•∆t-•∆t κ•∆t β •∆t δ•∆t•∆t) 1-δ•∆t δ•∆t 1 1-δ•∆t-βa•∆t βa •∆t δ•∆t Fig. 4. Transition probabilities for temporal evolution of states in Fig. 2. respectively. Also, A+I switches to A-I, I+I, or A+I with probabilities 1-δ •∆t-β a •∆t•, β a •∆t, and δ •∆t, respectively. Among the neutral edges, S±S, A±A, and S-A flip their edge signs with probability 1 in each time step, S+A changes to either S-A or A+A with probabilities In general, given the pair , the tri-variate CTMC of the form {Z i,j ; t ≥ 0}, where Z i,j = X i , X j , A i,j , defines these conditional transition probabilities as follows: 1 -κ • ∆t • and κ • ∆t • , respectively. Finally, I ± I changes to I ∓ I with probabilities 1 -2δ • ∆t • and 1 -2δ • ∆t, P c,c ′ Pr Z i,j = c ′ Z i,j = c , where c = , c ′ = , and x, y, z, x ′ , y ′ , z ′ ∈ {-1, 0, 1}. Based on , the stead-state probability distribution is derived in Theorem 1. Theorem 1. Let π x,y,z x, y, z ∈ {-1, 0, 1} be the stationary probabilities for the CTMC Z i,j defined above, then we have the following in steady-state: The fraction of susceptible users is y,z π 1,y,z . The fraction of infected users is y,z π -1,y,z . The fraction of alerted users is y,z π 0,y,z . The fraction of friendly links is x,y π x,y,1 . Proof. We use to obtain the elements of the infinitesimal generator matrix Q = [q c,c ′ ] of order 27 as follows: q c,c ′ =    lim ∆t→0 P c,c ′ -1 ∆t ; if c ′ = c, lim ∆t→0 P c,c ′ ; if c ′ = c. From , we now can obtain the stationary probabilities by solving Π • Q = 0 and Π • 1 = 1, where Π = π x,y,z x, y, z ∈ {-1, 0, 1} . Denoted by s ∞ = n i=1 S i /|V|, the fraction of susceptible users is computed as: s ∞ = y,z π 1,y,z . Similarly, the steady-state probabilities for ρ ∞ , a ∞ , and r ∞ can be obtained straightforwardly. --- Algorithm 1 Time-efficient Network Energy Calculation Input: G0 = , α, ρ0, r0, β, βa, κ, and δ. --- Output: Emin. Initialization: ∀, Ai,j ∈ E0 = -1, and e = 1. 1: Calculate E using . 2: e ← E. 3: for t ← ∆t to T • ∆t do 4: Randomly select an edge ∈ Et-∆t. --- 5: Change the state of edge according to Fig. 4. --- 6: Update E using , , and e ← E with probability 0.5. --- 9: else if E < e then 10: e ← E 11: return Emin ← e --- MONTE CARLO METHOD Starting from an initial network state at t = 0, where a fraction of users are randomly infected, we select an edge at random in each evolution step of the simulation and change its state as in Fig. 4. Doing so affects the energy states of all triads that share edge in the long-term which successively, alters the total network energy state. Convergence towards the new network structure transpires as long as the new energy state decreases in each time step [11], i.e., this process continues until the global minimum = -1) or a local > -1) minimum energy state is reached. Computing E using in each time step takes O n 2 + n 3 time. To expedite the computation, we propose Algorithm 1 that evaluates the energy difference of the selected edge between consecutive time steps, t ′ and t ′′ , in O time as: ∆E = α • ∆E △ + • ∆E p , where ∆E △ is the difference in the triad energy, i.e., ∆E △ = 1 n 3 i ′ ,j ′ ,k ′ E i ′ ,j ′ ,k ′-E i ′ ,j ′ ,k ′ = A i,j -A i,j n 3 k ′ =i,j A i,k ′ • A j,k ′. If the state transition does not flip the edge sign, then E △ remains unaltered. Otherwise, flipping the edge sign implies that A i,j = -A i,j , which further reduces to: -2Ai,j k ′ =i,j A i,k ′•A j,k ′; if A i,j = -A i,j , 0; if A i,j = A i,j . Similarly, ∆E p can be computed as follows: ∆E p = 1 n 2 E p i,j -E p i,j . (12 ) --- SIMULATION RESULTS AND DISCUSSIONS We evaluate the proposed energy framework with respect to r 0 , ρ 0 , and α. For Case Study I, we generate a complete network of |V| = 180 users to entail the maximum number of triads in our analysis. We then apply the bootstrap technique to extract i.i.d. samples of smaller connected components of size |V| from the Slashdot081106 [18] and the Bitcoin-OTC [18] 100 runs on a PC with 3.2 GHz Intel Core i9-9900 CPU and 16 GB RAM 1 . --- Case Study I: Results For the synthetic network, Fig. 5 plots the trajectories for a viral outbreak under varying values with r 0 = 0.25, p 0 = 0.15, α = 0.5, β a = 0.3β, and δ = 9 [15], [16]. As shown in Fig. 5a, unlike the SIS baseline , the susceptible fraction under the SAIS model drops to zero with rate proportional to κ. As a result, the users are either influenced by the viral process or alerted thus making them less likely to fall prey in the long run since β a ≪ β. The impact of κ on the alerted and infected user densities are, respectively, shown in Fig. 5b and Fig. 5c. For higher κ values, a larger susceptible fraction is made aware of the spread which, in turn, diminishes the size of the infected population. For instance, in contrast to the baseline, Fig. 5c vividly shows that the infected cluster size decreases by nearly 14% when κ = 2. In spite of setting β = 0, note that there exists a non-zero infected population in the network. This clearly implies that for smaller values of r 0 , the proposed model partitions the network into two distinct clusters: one comprising of alerted users and the other containing infected users. While the users within each cluster maintain a friendly relationship with each other, they are hostile towards users in the other cluster. The virality of the process however, dies out gradually with rise in r 0 and the two clusters eventually merge into a single cluster of alert users. The impact of κ on the number of friendly links is shown in Fig. 5d. Driven by the transitions given in Fig. 4, such behavior is not far from expectation as users are inclined to detach from friends influenced by the spreading phenomena and instead, befriend those who are informed or share common interests to attain social stability. values in steady-state. In Fig. 6a, we observe that the system gravitates towards the jammed states < -1), where mitigating the epidemic is favored over attaining structural balance, for lower values. As α and r 0 goes beyond 0.5 however, it is evident in Fig. 6b that the network tends towards the global minimum energy state to become structurally robust at the expense of further epidemic spread. Interestingly, due to the reversible nature of the SAIS model, the fully balanced complete network progresses to be infectionfree at α = r 0 = 1 as all users eventually become aware of the spread. Hence, the network tends to exploit the negative links to naturally immunize the susceptible users by separating them from the cluster of infected users. Therefore, for any given setting, an optimal pair exists for which the network would contain minimum number of infected users in steady-state and yet, not necessarily be socially balanced. The significance of ρ 0 in the coupled evolution is shown in Fig. 7. For 0 ≤ ρ 0 , r 0 ≤ 0.5, Fig. 7a showcases the impact of negative links in controlling the contagious spread. Further increase in ρ 0 however, yields a fixed fraction of infected users as most of the triads have evolved into a balanced state. Full recovery is attained when ρ 0 ≤ 0.5 and r 0 > 0.5 due to the small infection prevalence and the low infection rate relative to κ and δ. Also, note that for low ρ 0 , the stationary infection density drops to zero, irrespective of r 0 . The process diffuses at its maximum when ρ 0 and r 0 are both high. Fig. 7b shows the trivial effect of ρ 0 on the net network energy for high r 0 , where the triad structural energy is dominant. Around r 0 = 0.5, the network struggles to become balanced as flipping the edge sign conceivably creates more unbalanced triads as compared to other values on the r 0 spectrum. Fig. 8 shows E p and E △ against r 0 for different values in steady-state. As seen in Fig. 8a, the minimum achievable pairwise and triad energy values increase with κ for r 0 < 0.6, whereas the ρ ∞ inversely drops. But for larger r 0 , the impact of β and κ on the energy and ρ ∞ is minimal. That is to say, E p in Fig. 8a increases to zero as the number of infected users becoming aware rapidly grows with increase in friendly links. Contrarily, E △ in Fig. 8b falls to the global energy minimum because the triads gradually evolve to become socially stable as most users have already formed friendly links. --- Case Study II: Results We now assess our model using the real sparsely-connected SL , BC , and the dense CS SN datasets. The total number of triads existing in SL, BC, and CS are 4268, 1588, and 74140, respectively 2 . Table 1 compares the impact of network scalability on the model performance. Fig. 9 depicts the steady-state results of the datasets with respect to generic β and κ values. Consistent with Fig. 6a and Fig. 7a, we see in Fig. 9athat ρ ∞ reduces with rise in awareness for ρ 0 = 0.15, while it remains high and almost the same for ρ 0 = 0.75. As a resultant, the fraction of alerted users increases with κ as shown in Fig. 9b. It is noteworthy that, irrespective of ρ 0 , the gap in a ∞ between CS and the other two datasets gradually decreases. Despite the fewer nodes in CS, the plots reveal the profound impact of κ on the nodal states of densely connected CS that contains more number of triads. Summing up the infected and alerted fractions for each pair also justifies the natural immunization induced by the formation of two-cluster networks. Changes in β and κ however, do not seem to influence the fraction of friendly links in steadystate. Though the epidemic state of nodes are decisive in link sign evolution, we see that for all scenarios the ultimate number of friendly ties is almost the same. Upon reaching steadystate, the fraction of balanced triads, denoted by |△ B |, is close to 100% in Fig. 9d, which indicates a balanced SN structure in accordance with Heider's balance theory. Furthermore, the results empirically show that although the structural energy 2. In general, the sharp upper bound on the number of triads in a graph with n nodes and m edges is n 6 of triads approach a global minimum, E does not reach the global energy minimum state. This is because the pairwise energy tends towards a local energy minimum state when α = 0.5. Also, since κ affects the pairwise spreading energy, E is slightly lower for higher κ rates as plotted in Fig. 9e. Comparing the trends in both rows of Fig. 9, the results for r ∞ and |△ B | are almost alike. With increase in κ, E of CS is higher by roughly 28% for ρ 0 = 0.15 because of the larger population of alerted users being intrinsically immunized as compared to the setting with ρ 0 = 0.75, where the impact of increasing κ is relatively inconsequential. --- CONCLUSION In this paper, coupled dynamics of the SAIS epidemic model and the structural evolution of SSNs was studied. Inspired by Heider's balance theory, a network energy framework was formulated to capture the viral spreading via pairwise user interactions in conjunction with social stability in triad configurations. The superiority of incorporating user awareness in the classical SIS model was fully validated by the Monte Carlo simulation results. Moreover, it was shown that a complete SSN splits into two clusters of alerted and infected users upon reaching a local energy minimum. The alerted cluster density was also found to grow with increase in the initial number of friendly links and a fully balanced SSN becomes infection-free only when the triad energy is considered and all initial user links are friendly. One interesting future work is to leverage network centrality measures other than degree distribution in probing the trade-off between opposing epidemics and social stability in directed and composite SSNs. Better parameter estimations based on network spectral analysis is also of value in adopting effective control strategies.

Prediction and control of spreading processes in social networks (SNs) are closely tied to the underlying connectivity patterns. Contrary to most existing efforts that exclusively focus on positive social user interactions, the impact of contagion processes on the temporal evolution of signed SNs (SSNs) with distinctive friendly (positive) and hostile (negative) relationships yet, remains largely unexplored. In this paper, we study the interplay between social link polarity and propagation of viral phenomena coupled with user alertness. In particular, we propose a novel energy model built on Heider's balance theory that relates the stochastic susceptible-alertinfected-susceptible epidemic dynamical model with the structural balance of SSNs to substantiate the trade-off between social tension and epidemic spread. Moreover, the role of hostile social links in the formation of disjoint friendly clusters of alerted and infected users is analyzed. Using three real-world SSN datasets, we further present a time-efficient algorithm to expedite the energy computation in our Monte-Carlo simulation method and show compelling insights on the effectiveness and rationality of user awareness and initial network settings in reaching structurally balanced local and global network energy states.

PSM Propensity score matching SMI Safe motherhood initiative SSA Sub-Saharan Africa Currently, the global estimates of maternal mortality , although indicative of improvements, remain unacceptably high 1,2 . In 1987, the Safe Motherhood Initiative was launched as an initiative to enhance the quality of maternal health. The focus of the initiative was channeled to people living in low-and-middle-income countries . The SMI envisages that in making an impact to minimize MM, all women must have access to essential health services including family planning 3 . However, many women of reproductive age are not utilizing any contraceptive method which is contributing to the high burden of MM. It has been established that a considerable number of MM would be avoided if the contraceptive prevalence rate increased, and the unmet need for FP decreased 4 . Contraceptive use reduces the risk of unplanned pregnancy 5,6 , and provides substantial social and economic benefits including improved educational and employment opportunities 7 . Additionally, contraceptive use reduces unwanted fertility which is a major public health problem in developing countries 4 . Globally, 172 million women are currently not using any method of contraception even though they desire to avoid pregnancy 8 . In 2013, Darroch and colleagues found that the unmet need for modern contraceptives in sub-Saharan Africa was 60% of the 89 million population 9 . The non-use of contraceptives varies across the continent. However, unmet needs, health concerns, infrequent sex, opposition from others, lack of knowledge, and other less reported factors are the key barriers hindering contraceptive use 10,11 . In low and middle-income countries, it is estimated that a quarter of women of reproductive age wish to avoid pregnancy but are not utilizing effective contraceptive methods 7 . Among women aged 15-49 years in 47 developing countries, this was observed to be higher at an average of 40.9% who needed contraception but are not using any technique 12 . All sexually active women should be informed about their risk of becoming pregnant as well as the many techniques available to prevent unplanned or unintended pregnancies. Countries in SSA are currently facing the challenges of high birth rates that may be attributable to inadequate access and use of contraceptive methods. In response to this, the governments of these countries are focusing on the need for mass communication campaigns to encourage the use of contraceptives. The empirical evidence documented in some studies although limited in scope, geographical boundaries, and in some cases statistical analysis rigor has shown that exposure to family planning information remains critical if we intend to increase contraceptive use among women of reproductive age [13][14][15] . In addition, the investment made in FP education over the years has been enormous. For instance, the direct and indirect annual cost of providing modern contraceptive services to 671 million users in developing regions was estimated to be US$6.3 billion 14 . Despite this huge investment in FP campaign messages, there is a paucity of evidence regarding the effectiveness of FP messages on contraceptive use in sub-Saharan Africa. In Africa, some studies have estimated the pooled prevalence of contraceptive use and determined associated factors among reproductive women using nationally stratified surveys 16,17 , however, none of them have assess the effectiveness of FP messages on contraceptive use. Using the most recent demographic and health survey data, we determine the prevalence trends and quantify the impact of exposure to FPM on CU among women of reproductive age in SSA. --- Methods This study followed the standard guidelines for reporting observational studies using the Strengthening the Reporting of Observational Studies in Epidemiology . Data source. This study utilized secondary data from the most recent and available Demographic and Health Survey conducted in 26 SSA countries between 2013 and 2019 . The DHS is a nationally representative household survey with similar data collection instruments and study designs conducted in LMICs with the primary goal of generating estimates for indicators that are comparable across the sub-region. The DHS provides data for a wide range of monitoring and impact evaluation indicators in the areas of population, health, and nutrition. Specifically, the DHS collects data on family planning , maternal health , household wealth, parity, education, place of residence, and demographics, amongst other variables with sample sizes and typically are conducted in every 5 years, to allow comparisons over time. The survey employs a multi-stage stratified cluster sampling design where the index country is stratified into distinct geographical regions or provinces during the first phase of the design. The first phase of sampling involves the random sampling of clusters or enumeration areas using probability proportional to the size of the EA and the subsequent sampling of a fixed number of households within each of the sampled enumeration areas using a systematic random sampling approach. A complete household listing was carried out to update the sampling frame before the random sampling of households. Trained field data collectors were assigned to these sampled enumeration areas for the household survey. Details on the study design and procedures for data collection have been published elsewhere 18 . The DHS data is publicly available upon reasonable written request at the DHS website . All standard DHS surveys have been reviewed and approved by ICF Institutional Review Board . Additionally, country-specific DHS survey protocols are reviewed by the ICF IRB and typically by an IRB in the host country . This study did not require country-specific ethical approval since we only analyze secondary data from the DHS Outcome variable. The primary outcome measure in this study was contraceptive use. Contraceptive use as defined by DHS was among women of reproductive age who currently use any standard method of contraceptive . Contraceptive use was classified as a binary variable that takes the value of 1 if the woman is currently using a traditional or modern contraception method and a value of 0 if otherwise. The modern methods include women who use female sterilization , male sterilization , the contraceptive pill , intrauterine contraceptive device , injectables , implant , female condom, the male condom , diaphragm, contraceptive foam and contraceptive jelly, lactational amenorrhea method , standard days method and country-specific modern methods. Respondents mentioned other modern contraceptive methods , but do not include abortions and menstrual regulation 19 . Primary exposure. Exposure to FPM was defined as individual women of reproductive age who heard or saw FPM on the radio, on television, in a newspaper or magazine, or on a mobile phone in the past few months 19 . --- Confounders. Variables considered as possible confounders were selected based on an extensive literature review of factors that could potentially influence access to FPM and contraceptive use among women of reproductive age. The following variables were accounted for in all the multivariable models: the age of the household head , sex of the household head , household wealth Index , place of residence , religion , respondent age , marital status , educational level , currently working , children ever born 20,21 . These variables have been found to either increase contraceptive use, exposure to family planning messages or both. --- Statistical analysis. Assessing trend and factors associated with contraceptive use. We explored the trend of FPM and CU between 2013 and 2019 using tools from time series line graphs and estimated the weighted prevalence of FPM and CU over the period by adjusting for sampling weight for all point and interval estimates including regression models. Factors contributing to CU and FPM were assessed using the Poisson regression model with a cluster-robust standard error that generates prevalence ratios and their respective confidence intervals. Sensitivity analysis of the point estimates and corresponding confidence interval was conducted using the multivariable binary logistic regression model that reports odds ratio and CI. The Poisson model was preferred to the logistic regression model as the odds ratio may overestimate the prevalence ratio, the measure of choice in cross-sectional studies 22 . Assessing impact of family planning messages on contraceptive use. Augmented inverse-probability weighting was used to estimate the average treatment effect of FPM from cross-sectional data. The AIPW estimator is classified among the estimators with the doubly-robust property as it combines aspects of regression adjustment and inverse-probability-weighted methods to reduce bias associated with the impact estimate. The model accounted for sampling weight and used cluster-robust standard errors to address the methodological challenges associated with complex survey design. Since different impact estimation procedures may lead to slightly different impact estimates especially when the data originates from crossectional studies instead of the more rigorous experimental design, sensitivity analysis of the impact estimate was conducted using endogenous treatment effect models, inverse probability weighting, propensity scores, and nearest-neighbor matching techniques. Estimating the impact of an intervention, program or policy becomes difficult due to endogeneity. For instance, genetic predisposition, personal values, conservative lifestyle, religious beliefs, and other unmeasured confounders may simultaneously affect exposure to family planning messages and utilization of contraception 13 . The standard regression models and exposure to FPM. This assumption is largely violated in the context of observational data where both the outcome and exposure are usually measured at the same time and may correlate with unobserved confounders. We anticipated these problems, and as part of the sensitivity analyses that were conducted, we used endogenous treatment regression models to address endogeneity. Having radio or television was used as the instrumental variable since it met the exclusion restriction criteria recommended for instrumental variable regression analysis . All statistical analyses were conducted using Stata version 17 and a p-value of less than 0.05 was considered statistically significant. [0.4% ]. The geospatial distribution of contraceptive use and exposure to FP messages can be found in Fig. 1. Trend analysis of contraceptive use and exposure to family planning messages. The trend analysis showed that the prevalence of contraceptive use among women of reproductive age and adolescence fluctuated between 2013 and 2019 but increased marginally between 2015 and 2017. Contraceptive use among women of reproductive age increased between 2013 and 2015 and declined between 2015 to 2017. Among adolescents, CU increased between 2013 and 2015 and remained fairly constant between 2015 and 2017 but declined between 2017 and 2018 before increasing marginally in 2019 . There was a positive correlation between exposure to FPM and CU as a higher prevalence of CU was associated with higher exposure to FPM and vice versa . --- Factors associated with access to family planning message. The following factors were found to be associated with access to family planning messages: age of the respondent, female household heads, higher socio-economic status measured via household wealth, living in urban areas, religion, marital status, higher education level, parity and women who were currently working at the time of the survey were found to be associated with a higher prevalence of access to FPM . --- Factors associated with contraceptive use. The results from the multivariable Poisson regression model showed that the age of the household head, sex of household head, higher socio-economic status measured via household wealth, living in urban areas, religion, marital status, higher education level, parity, and exposure to FPM were found to be associated with contraceptive use . --- Impact of exposure to family planning information messages on contraceptive use among women of reproductive age: evidence from DHS study. Table 2 shows the results from the augmented inverse probability weighting with regression adjustment and sensitivity analysis of the impact estimate among women of reproductive age. Exposure to FPM increased contraceptive use by 7.1 percentage points among women of reproductive age in SSA. The impact of FPM on contraceptive use was highest in Central Africa [6.7 pp; 95% CI: and lowest in Southern Africa ]. Cameroon recorded the highest impact of FPM on contraceptive use ] but exposure to FPM did not have a statistically significant effect on contraceptive use in Rwanda, Namibia, and Liberia. Impact of exposure to family planning information messages on contraceptive use among adolescents: evidence from DHS study. Table 3 shows the results from the augmented inverse probability weighting with regression adjustment and sensitivity analysis of the impact estimate of FPM among adolescents. Exposure to FPM increased CU by 6.0 percentage points among adolescents in SSA. The impact of FPM on CU was highest in Southern Africa [7.0 pp; 95% CI: ] and lowest in Eastern Africa [2.0 pp; 95% CI: ]. Cameroon recorded the highest impact of FPM on contraceptive use among adolescents [impact estimate 17.0 pp; 95% CI: ]. --- Discussion This study assessed the prevalence, trends, and impact of exposure to FPM on contraceptive use among women of reproductive age in SSA and further conducted a sub-group analysis among the adolescent class of women using augmented inverse probability to treatment weighting with regression adjustment. Different sensitivity analyses were performed as a robustness check to confirm the results of augmented inverse probability to treatment weighting with regression adjustment. The empirical evidence presented in this manuscript allows us to draw four important conclusions. First, the prevalence of contraceptive use among women of reproductive age and the adolescent sub-class largely varies among countries and geographic groupings in SSA and changes significantly over time. The high fluctuations in the prevalence of contraceptive use based on the trend analysis could be attributed to the variations in the level of intensity of family planning campaigns over the period, access and affordability of contraceptives in the sub-region. The marginal increase in CU among the general population coupled with the declining CU among adolescents despite their increased exposure to FPM would indicate that regardless of exposure to messages, barriers to use persists. Second, our final multivariable regression analyses showed that exposure to FPM does increase the likelihood of using contraceptive methods among women of reproductive age and adolescents sub-class in SSA although the effect size estimate varies by country and regional block. The regional and national diversity of SSA may play a key role in the diffusion of fertility regulating ideas and practices adopted by women 17 . Evidence of this is seen as contraceptive use among women of reproductive age and adolescents is higher in Southern Africa compared to Central, Eastern, and Western African countries. Eastern African countries are the most exposed to FPM this has been documented to be attributable to the government's investments in improving access to SRH Vol:. 26 . Among the general population of women and adolescents, the high prevalence of CU and exposure to FPM in SSA occurred among women in Southern Africa and specifically Zimbabwe. Zimbabwean women have benefited from the strong post-independence encouragement of contraceptive use by their government 27 . For CU, approximately less than and a little more than one-twentieth of adolescents utilize contraceptives among adolescents and the general population respectively. The lowest prevalence of CU in Chad has also been confirmed by Ahinkorah et al., 2021 28 . Chad is at a disadvantage in both the use of contraceptives and exposure to FPM which needs urgent attention for improvement This calls for the adoption of new strategies to include adolescents in exposure to FPM programs since non-exposure to FPM directly translates into a high unmet need for FP among adolescents 29 . Communication is a vital mechanism connecting social factors and health outcomes. Finally, we infer that exposure to FPM was found to be associated with CU among participants. Exposure to FPM was defined as hearing or seeing an FPM on the radio, television, in a newspaper or magazine, or on a mobile phone in the past few months. By using a counterfactual control group in this current study, the impact of exposure to FPM significantly increased the utilization of contraceptives in the SSA region. Findings envisage that the average conceptive use among women of reproductive age who are exposed to FPM significantly increased as compared with those who are not exposed. This finding corroborates the findings that exposure to FPM enhances the use of CU among reproductive-age women 30 . Our study has provided empirical evidence to support the incessant calls for policymakers, external donor funding agencies, Civil Society and NGOs to prioritize and increase the resources for implementing family planning communication interventions in low-and middle-income countries. We proposed diverse country-specific policies, programs, and interventions that incorporate the different dynamics of socio-political, cultural, and other contextual factors that hinder access to family planning messages and the use of contraceptives in SSA. Assessing the impact of health interventions poses a great challenge in situations where the data used for the analysis were from observational studies due to the problem of endogeneity . Although a more rigorous statistical technique and sensitivity analysis of the impact estimates were conducted to generate an unbiased estimate of the program impact that addresses the problems of endogeneity, we believe that other unmeasured covariates such as health-related conditions, genetic predisposition, socio-cultural factors and area-specific inherent traditions in some part of SSA and many other factors may contribute to the observed change in the contraceptive use. Notwithstanding these limitations, this impact evaluation study represents one of the few efforts to examine the effects of FPM on contraceptive use in SSA using data that originate from observation studies compared to the more preferred experimental study designs. It is the first study to assess the effect of family planning messages in SSA. In addition, the main outcomes were self-reported, which are subject to participants' recall bias or socially desirable responses because the DHS asked the participant to recall over the past 30 days. --- Conclusion Prevalence of CU and exposure to FPM varies significantly across countries in SSA and the exposure to family planning messages increased the use of contraceptive among women of reproductive age. Despite disparities observed, exposure to FPM has contributed to an increase in CU among women of reproductive age and the adolescent sub-class. We emphasized the need to implement policies that incorporate social-cultural and political support to encourage women to adopt contraceptive methods following exposure to messages. Funding for family planning education via print and electronic media should continue unabated. --- Data availability The datasets that were used in the study are publicly available on the DHS website . --- Ethics approval and consent to participants. This is a secondary data analysis of publicly available data with de-identified participants' information. --- Results Characteristics of the study participants. The study involved 328,386 women of reproductive age in SSA with an average of 30.5 years . Approximately 61% of the women lived in rural areas and 73% were married. About 30% of the women had no formal education. The sociodemographic characteristics of the women can be found in supplementary Table 2. --- Prevalence of contraceptive use and exposure to family planning messages. --- Competing interests The authors declare no competing interests.

Many women of reproductive age in sub Saharan Africa are not utilizing any contraceptive method which is contributing to the high burden of maternal mortality. This study determined the prevalence, trends, and the impact of exposure to family planning messages (FPM) on contraceptive use (CU) among women of reproductive age in sub-Saharan Africa (SSA). We utilized the most recent data from demographic and health surveys across 26 SSA countries between 2013 and 2019. We assessed the prevalence and trends and quantified the impact of exposure to FPM on contraceptive use using augmented inverse probability weighting with regression adjustment. Sensitivity analysis of the impact estimate was conducted using endogenous treatment effect models, inverse probability weighting, and propensity score with nearest-neighbor matching techniques. The study involved 328,386 women of reproductive age. The overall prevalence of CU and the percentage of women of reproductive age in SSA exposed to FPM were 31.1% (95% CI 30.6-31.5) and 38.9% (95% CI 38.8-39.4) respectively. Exposure to FPM increased CU by 7.1 percentage points (pp) (95% CI 6.7, 7.4; p < 0.001) among women of reproductive age in SSA. The impact of FPM on CU was highest in Central Africa (6.7 pp; 95% CI 5.7-7.7; p < 0.001) and lowest in Southern Africa (2.2 pp; 95% CI [1.3-3.0; p < 0.001). There was a marginal decline in the impact estimate among adolescents (estimate = 6.0 pp; 95% CI 5.0, 8.0; p < 0.001). Exposure to FPM has contributed to an increase in CU among women of reproductive age. Programs that are geared towards intensifying exposure to FPM through traditional media in addition to exploring avenues for promoting the appropriate use of family planning method using electronic media remain critical.AIPW Augmented inverse probability weighting CU Contraceptive use DHS Demographic and health survey ETE Endogenous treatment effect FPM Family planning messages IPW Inverse probability weighting MM Maternal mortality

Background The introduction of combined antiretroviral treatment has drastically reduced HIV-related mortality and morbidity, and has yielded improvements in the quality of life and prognosis of people living with HIV [1]. Antiretroviral treatment, together with programmed caesarean sections, have significantly reduced mother-tochild transmission to rates under 2% [2]. All this has led a growing number of HIV-infected persons to consider the possibility of having children. Studies in women diagnosed with HIV infection in Africa, Europe and North American show that many of these women decide to have children even after their HIV diagnosis [3][4][5]. However, to our knowledge, there are no studies assessing the reproductive desires of HIV-infected women in Spain. A large variety of individual, clinical and social factors influence reproductive desire in HIV-infected persons. Some studies show that HIV infection is not a dominant factor in the desire to have children, and that the factors influencing this desire are similar in both HIV-infected women and those in the general population [3,4]. HIVinfected women of reproductive age have reported high anxiety related to reproductive concerns, such as a great desire for pregnancy along with experiences of stigmatization for trying to become pregnant [6,7]. In Spain, women comprised 17% of new HIV infections and 21% of AIDS cases diagnosed in 2011. Most of them acquired their infection through sexual contact [8]. The objectives of this work were to evaluate the reproductive desires of HIV-positive women included in the CoRIS cohort, to analyze the different motivations and to identify the associated factors. Such knowledge will be important for developing programs to support these women in planning safe pregnancies. --- Methods --- Design, sample frame and subjects To achieve the study objectives, we designed a crosssectional observational study framed within the cohort of HIV-infected adults in the AIDS Research Network . CoRIS is a prospective, open and multicenter cohort of patients newly diagnosed with HIV who are over 13 years of age and naïve to antiretroviral treatment at cohort entry [9,10]. Of the 28 centers participating in CoRIS, nine hospitals and one sexually transmitted disease clinic in each of eight cities from six Autonomous Communities of Spain participated in this project . Each patient signed an informed consent form, and the study was approved by the Ethics Committee of the Carlos III Institute of Health. The first phase of the study included all women in the CoRIS cohort in the participating hospitals and centers in January 2010 who were between 18 and 49 years of age. Healthcare and/or clinical staff in each participating center contacted the women selected to explain the project and to request their participation and written informed consent. After receiving their consent, in the second phase a person from the research team contacted each woman by telephone to conduct the interview. Of the 494 women selected, 240 could not be located or were no longer being followed up in the hospital or healthcare center. Of the remaining 254, 155 agreed to participate, 63 refused, and 36 were excluded for various reasons . One of the hospitals included in the study began to recruit women when the field work was almost completed, therefore the number of women included was very small . A total of 161 women were finally interviewed. The interviews were conducted between November 2011 and December 2012. The main outcome of this analysis was reproductive desire, defined as having a desire to be pregnant at present OR having unprotected sex with the purpose of having children OR wanting to have children in the near future. Thus, the sample was limited to women of reproductive age , a total of 134 women. --- Definition of variables A questionnaire specific to this study was designed and validated in a pilot test with 10 women. The final questionnaire included a mixed qualitative-quantitative approach . Quantitative data were collected on sociodemographic, epidemiological and clinical variables, as well as sexual and reproductive history. To collect information on the reasons underlying reproductive desire, issues pertaining to reproductive health and disclosure of HIV infection, we chose a qualitative approach using openended questions. Two experienced interviewers conducted the interviews by telephone. Country of origin was categorized as "Spain" or "Other countries". Women from other countries were subsequently categorized according to their region of origin into three groups: "Europe", "Latin America" and "Africa". Educational level was defined as "low" if women had no formal education or primary education, "medium" if women had completed secondary education, and "high" if women had a university degree. Current occupational status was defined as "employed", "unemployed", or "other"which included students, homemakers and pensioners. Current partner and cohabitation status was grouped into "has a partner and they live together", "has a stable partner but they do not live together" and "no stable partner". HIV transmission category was classified as "injecting drug user" , "heterosexual contact" or "other". Hepatitis C Virus serological status was classified as positive or negative antibodies, AIDS defining conditions as "Yes" or "No", and Time since HIV diagnosis as "<3", "3-6" and ">6" years. In accordance with the previous answers, women were then asked the open-ended question "Why do you want to have children in the near future?" or "Why do you not want to have children/more children in the near future?". Answers were recorded and later grouped into the following categories for those who "Desired to have children": To have a normal life, To form a family, They like children and want to have them, To experience maternity, To have someone to take care of them in the future, Their partners want to have children, and Religious beliefs. The categories for those who did "Not wish to have children" were: Age , They already have children, HIV infection, Personal reasons, Economic reasons, Drug abuse, Do not like or want to have children, Health problems other than HIV infection. Women were also asked if they had looked for or received information about sexual and reproductive health, and whether they had disclosed their HIV infection to their closest relatives and friends. Social support was evaluated through a 4-item MOS-SSS version [11], each representing a social dimension: Someone to help you with daily chores if you were sick [Tangible], Someone to turn to for suggestions about how to deal with a personal problem [Emotional/Informational]; Someone to do something enjoyable with [Positive social interaction] and Someone to love and make you feel wanted [Affectionate]. Each item ranged from 1 to 5 , so a higher score meant better social support. An overall score was calculated and transformed into a 0-100 scale using the formula: 100 × [/]. --- Statistical analysis Women's characteristics were described according to reproductive desire using frequency tables for categorical variables and median and interquartile range for continuous ones. The chi-squared test for independence was used for comparison of categorical variables and the non-parametric Kruskal-Wallis test for continuous variables. Multivariate logistic regression models were fitted to evaluate the association of women's characteristics with reproductive desire. Crude and adjusted odds ratios with their 95% confidence intervals were obtained as the measure of association. Predictive modeling was performed, and variables with a P-value lower than 0.05 were retained in the multivariable model. All statistical analyses were performed using Stata software . --- Results --- Description of women A total of 134 women were interviewed, with a median age of 36 years . Some 54% were of Spanish origin and, of the remaining 62, 44% were from Latin America, 32% from Europe, and 24% from Africa. The foreign women had been in Spain a median of 8 years . About 77% of women had medium educational level and a similar proportion had low and high education: 11% and 12% , respectively. With regard to employment status, 50% were currently employed, 35% were unemployed and 15% had other occupational status . A total of 67% had a stable partner at the moment and, of these, 77% lived with their husband or current partner. At the time of the interview, 21% reported having a health-related circumstance which complicated or impeded procreation. Among them, 43% had undergone tubal ligation. Only 2% of their male partners had fertility problems. Regarding their HIV infection, 51% had been diagnosed with HIV before 30 years of age, 37% between 30 and 39 years, and 11% at age 40 or over. The majority had acquired the infection though heterosexual exposure and 13% through injection of drugs. About 26% had been diagnosed with HIV for less than 3 years, 39% for between 3 and 6 years, and 35% for more than 6 years. Most women reported being asymptomatic, while only 7% were on AIDS stage. In all, 85% were on antiretroviral treatment at the time of the interview. HCV coinfection was present in 15% of women. Table 1 shows the descriptive analysis of sociodemographic, reproductive and clinical characteristics of women included in the study, stratified by reproductive desire. --- Reproductive history Out of the 134 women included in the study, 116 had been pregnant in their lifetime and a total of 280 pregnancies were reported, 5 of them in process at the time of the interview . Over half of the women had been pregnant once or twice, 22% three or four times, and 9% more than four times. About 60% had one or two children, 8% had three or more children, and the rest did not have any children . Among the 116 women who had been pregnant, 54 had only been so before their HIV diagnosis, 24 only after the diagnosis, and 38 had been pregnant both before and after the HIV diagnosis. Regarding the 280 episodes of pregnancy, 65% occurred before the HIV diagnosis, and the remaining 99 when the subject was already HIV-positive. Seven of the total pregnancies were achieved though assisted reproductive technology . --- Analysis of reproductive desire Of the 134 women, 49% wanted to have children or more children in the near future. The median age of women with reproductive desire was 32 years compared to 39.5 years in those with no reproductive desire . Reproductive desire was more frequent among women of non-Spanish origin than in those of Spanish origin . No statistically significant differences were found regarding educational level, occupational status or partner and cohabitation status. Reproductive desire was more frequent in women without any children , but no differences were found by number of pregnancies. With regard to clinical variables, significant differences in reproductive desire were found in women who were coinfected with HCV vs. those who were not coinfected , among those currently receiving cART vs. those not on cART , and among women who had been diagnosed at a younger age vs. medium or older age . In the multivariate analysis , the factors associated with reproductive desire were: 1) Age. Considering women between 30 and 39 as the reference category, the OR for women under 30 was 4.5 , while it was 0.4 for women 40 or over; 2) Being from a country other than Spain ; 3) Not receiving cART ; and 4) Not having any children . --- Reasons for reproductive desire Of the 66 women who wanted to have children or more children in the near future, 6 did not state any specific reason or motivation. For the remaining 60 women, the most frequent reasons were liking children and wanting to form a family, with 23% each. Eighteen percent of these women wanted to procreate because their partner wished to have children of their own, although the women themselves already had children. Fifteen of the 60 women expressed more than one reason for wanting children . Regarding the 68 women who did not wish to have children or more children in the future, 3 did not express any specific reason or motivation. For the other 66, the main reasons were their HIV infection followed by age . About 55% expressed more than one reason for not wanting to have children or more children . --- Information on reproductive health Women were asked if they had ever sought or received information about how to have a safe pregnancy since their HIV diagnosis; 49% answered affirmatively. Of them, 59% fulfilled the definition of reproductive desire used in this analysis., i.e., having a desire to be pregnant at present, having unprotected sex with the purpose of having children, or wanting to have children in the near future. For the majority of these women , information was provided by the doctor or other health-care professional involved in the routine clinical follow-up of their HIV infection. Only 6% had received reproductive-health information from non-governmental organizations, and the remaining 12% had looked it up in books, journals or through the internet. The recommendation to avoid pregnancy had been made to 19% of women at some point since their HIV diagnosis. In 60% of these cases, this recommendation had been made by health-care personnel: in 3 cases, clinical circumstances made it inadvisable to suspend or change antiretroviral treatment; in the rest of the cases, it was related to uncertainty about HIVinfection in the early stages of the epidemic in the 80's and the 90's, or had been made by health-care personnel not involved in HIV care, and thus with no specific knowledge about this infection. In 24% , the recommendation to not have children came from the family or close relatives and was motivated by fear of HIV. --- Disclosure and social support Only 10% of interviewed women reported that none of their closest family members knew they were HIV-infected. Of the 120 women who had disclosed their serostatus to their family circle, 33% had told it to all their closest family members, i.e., parents, brothers and sisters, and stable partner, as well as close friends; 25% had only revealed it to their partner; 28% had disclosed it to some family member besides their partner; and 13% had only told it to a very close family member . Overall, 39% reported having experienced discrimination due to HIV infection, with no differences according to disclosure of serostatus. The majority of discrimination situations reported by women occurred in the early moments after their HIV diagnosis. About 37% had experienced discrimination from health-care professionals not used to dealing with HIV-infected patients, in 27% discrimination came from friends or work colleagues, in 15% from a relative, and in 6% from their previous partner. For 15% of women, this discrimination was due to the lack of knowledge and information about HIV in the general population. Regarding social support evaluated using the 4-item MOS questionnaire, scores were generally high : 26.3; median: 87.5). The scores were also high for each of the evaluated dimensions: tangible support , emotional/ informational support , positive social interaction , and affectionate support . --- Discussion In our study, we found that half of the women from the Spanish AIDS Research Network Cohort would like to have children in the future and that 35% of their pregnancies were produced after they were diagnosed. The main predictors of women's reproductive desire were young age, not having children previously, being of foreign origin and not receiving antiretroviral treatment. The reported reasons for wanting to have children were mainly related to the ideas of liking children, wanting to form a family and pleasing their partner. Among the reasons for not wanting children, HIV infection played a major role, but was not a definitive factor [12]. Few of them had disclosed their HIV infection to their social circle and in most cases, only to the partner or very close familiar circle. No association was found between reproductive desire and HIV disclosure. However, a considerable number of these women had not received information on sexual and reproductive issues, and few of them had consulted or visited fertility clinics. The proportion of women in our study who wished to be pregnant was much higher than in studies carried out in American [13,14] or African women [15], but similar to what was observed in a Canadian study [16]. In our study, we defined reproductive desire as having a desire to be pregnant at present, or having unprotected sex with the purpose of having children, or wanting to have children in the near future. This means our results may not be fully comparable with the aforementioned studies, which analyzed reproductive desire as a secondary objective within a larger study, with the exception of the article by Loufty et al. [16]. Overall, we observed a high number of pregnancies in these women, both before and after HIV diagnosis; specifically, 35% of pregnancies were produced after the woman was diagnosed with HIV. About 31% of women had been pregnant more than three times. Other studies have observed that the number of HIV-infected women who have consecutive pregnancies has increased, a fact that seems to be associated with age, with not yet having children, and with being diagnosed with HIV after their first pregnancy [17]. The first two reasons are similar to those reported by HIV-negative women for wanting children. Being younger has been shown to be a predictor of procreation desire in HIV-positive women in all studies [4,5,[13][14][15][18][19][20][21][22]. Women from other countries, mainly from Latin America, have a higher reproductive desire, probably associated with the cultural importance of maternity in these countries [23]. Another factor associated with a higher reproductive desire is not receiving antiretroviral treatment , similar to what was observed by Zhang et al. [24] in a study in Canadian women, where the use of cART was associated with a very low desire for procreation. This result contrasts with the findings of studies in resource-limited settings [13,25,26], where reproductive desire increases when women receive cART, probably as a result of improved health status after beginning treatment. In our setting, where the health system grants universal access to cART, we do not observe such a direct association between receiving cART and reproductive desire, as was also seen in the Canadian study. Regarding their motivations, we found that even if women already have children from previous relationships, a considerable proportion of them want to have children to please their partners, who wish to have their own children. However, in the multivariate analysis we failed to find any association between reproductive desire and having a stable partner or living with him. Expectations from the family or the partner have a great influence in the procreation desire of persons who live with HIV [14,18,21,27]. As described in the literature, for HIV-infected women, having children can be a key element in feeling useful in society, in doing something meaningful for their partners, and in freeing themselves from the stigma and discrimination experienced by HIV-infected persons [28]. As opposed to other studies focused on African-American women, religious beliefs do not appear to have any influence in the procreation desires of women included in our study [29,30]. Regarding motivations for not having children, in addition to HIV infection, other reasons expressed by women are age and having children already, which are closer to the feelings and concerns of HIV-negative women. In a national survey carried out in 2006 in Spain, the reasons women gave for not wanting more children were that they already had as many as they wanted, that they were too old, that having children was expensive, that children could cause a lot of concern, and for health reasons [31]. In this same survey, the main reasons for wanting children/more children were because it was gratifying , because it was nice to see children grow and develop and because they enhanced a sense of responsibility and personal development [31]. In another general population survey conducted in 2003, about women's attitudes towards children and their upbringing, 52% reported having fewer children than they desired, mainly due to insufficient economic resources and the difficulty in combining maternity and work . These two reasons were considered even more important when asking respondents why they thought that couples in general had no children or fewer than they desired [32]. Compared to our study population, this survey recruited women who were younger , less frequently unemployed , and more frequently had either high or low educational level, and fewer women with medium education. As previously pointed out, approximately half of the women wanted to have children in the future, form a family and have a normal life, which contrasts with the fact that few of them disclose their HIV infection to their social circle and, in some cases, only the partner, parents or brother or sister knew about it. Our data did not show any association between reproductive desire and HIV disclosure . Social and family support not only improves the quality of life of persons living with HIV [33,34], but also helps them in making the decision to have children [26,35]. This hypothesis could not be tested in our study, since a reduced version of the MOS questionnaire was considered more suitable for a telephone interview. The lack of information on reproductive-related issues, both on the part of doctors and other health-care professionals [5] and the women themselves, has important implications for their reproductive-health care. It is essential to include sexual and reproductive health issues in the routine clinical follow-up of women with HIV. Few women consulted or visited fertility clinics , and few pregnancies were achieved using assisted reproduction techniques. Nevertheless, this proportion is higher than in the general population. In the previously mentioned 2006 survey, the proportion of women who had visited these types of clinics was 1.5% and 3.3%, for women with and without children, respectively [31]. In a Canadian study, the majority of HIV-positive participants reported they would like to have access to information on assisted reproduction treatments, but perceived this as beyond their reach due to the high costs [24]. In our setting, HIV-infected persons can access assisted reproduction treatments within the public health system [36,37]. However, some limitations regarding specific techniques such as sperm washing, and other general criteria such as women's age or long waiting times, can in the end act as barriers to access. Although the degree to which our sample is representative of HIV-infected women in Spain, our study is nested within a cohort of patients who visit the recruiting site for the first time and are naïve to antiretroviral therapy, so it is possible that women with a better clinical profile and shorter time since HIV diagnosis are overrepresented in our sample. On the other hand, the method for data collection -the telephone interview-might have deterred the participation of women with a lower educational level, who speak and/or understand Spanish poorly or who have less awareness or are less motivated to participate in research studies. Compared to the population from the municipal registries of women between 15 and 44 years of age, our study had a higher proportion of foreigners, 46% versus 28%, respectively [38]. This is reflected in our results, where we found that being a foreigner is one of the factors associated with reproductive desire. Our sample included a large number of Latin American women, who have a higher birth rate than Spaniards and in whom the desire for maternity is more likely to be influenced by cultural factors related to the traditional role of women in their home countries [19,39]. A limitation of our study is the small sample size analyzed. Almost half of the women selected in the first phase could not be located or were not being followed up when the study was conducted. This was especially true in the largest hospitals . In these hospitals, which are reference centers for HIV, it is easier for a patient to seek care in the initial phase after diagnosis and then be referred to another center closer to her home. Despite the limited sample, we believe this study provides relevant information to improve reproductive-health services for HIV-infected women. Studies previously carried out in Spain have analyzed the situation of these women once they were pregnant, or have studied changes in mother-to-child transmission over the years [40][41][42], but they have not evaluated the reasons for wanting or not wanting to have children in the future or the factors associated with this reproductive desire. Use of telephone interviews to conduct our study offers some advantages: it avoids unnecessary journeys, can be carried out at the most convenient time and place for the person interviewed, and makes it easier for women to answer sensitive questions with greater privacy. At the same time, however, it makes it more difficult to evaluate and study complex subjects like the desire for procreation and social support which could have been approached better in a face-to-face interview. Nevertheless, given the scarcity of studies on this subject in our setting, we think our results can contribute to the understanding of reproductive-health needs of HIV-infected women in Spain. As noted in the methods section, we used a mixed qualitative-quantitative approach. The qualitative part was constructed by using open-ended questions, which can be difficult to apply in a telephone interview. However, we believe this is the correct approach to elicit motivations underlying reproductive desire, and in the case of social support and disclosure of HIV infection, to explain women's reasons for wanting or not wanting to have children. --- Conclusion We can state that the CoRIS HIV-infected women we interviewed have a high reproductive desire, that factors associated with this desire are not fundamentally different from those in the general population of women, and that maternity can even help them to face a situation that they still consider stigmatized and prefer not to disclose. Health-care protocols for handling HIV-positive women should incorporate specific interventions on sexual and reproductive-health, to help them fulfill their procreation desire and experience safe pregnancies. --- Additional files Additional file 1: Centers and researchers involved in CoRIS. Additional file 2: Questionnaire. --- Competing interests The authors declare that they have no competing interests. Authors' contributions VH, BA, and ALL were involved in the study design and participated in the data collection and analysis. VH and ALL wrote the first draft of the manuscript. All authors contributed to data collection, reviewed the draft of the manuscript and approved the final version.

Background: Antiretroviral therapy has created new expectations in the possibilities of procreation for persons living with HIV. Our objectives were to evaluate reproductive desire and to analyze the associated sociodemographic and clinical factors in HIV-infected women in the Spanish AIDS Research Network Cohort (CoRIS). Methods: A mixed qualitative-quantitative approach was designed. Women of reproductive age (18-45) included in CoRIS were interviewed by phone, and data were collected between November 2010 and June 2012 using a specifically designed questionnaire. Reproductive desire was defined as having a desire to be pregnant at present or having unprotected sex with the purpose of having children or wanting to have children in the near future. Results: Overall, 134 women were interviewed. Median age was 36 years (IQR 31-41), 55% were Spanish, and 35% were unemployed. 84% had been infected with HIV through unprotected sex, with a median time since diagnosis of 4.5 years (IQR 2.9-6.9). Reproductive desire was found in 49% of women and was associated with: 1) Age (women under 30 had higher reproductive desire than those aged 30-39; OR = 4.5, 95% CI 1.4-14.3); 2) having no children vs. already having children (OR = 3.2; 1.3-7.7 3); Being an immigrant (OR = 2.2; 1.0-5.0); and 4) Not receiving antiretroviral treatment (OR = 3.6; 1.1-12.1). The main reasons for wanting children were related to liking children and wanting to form a family. Reasons for not having children were HIV infection, older age and having children already. Half of the women had sought or received information about how to have a safe pregnancy, 87% had disclosed their serostatus to their family circle, and 39% reported having experienced discrimination due to HIV infection.The HIV-infected women interviewed in CoRIS have a high desire for children, and the factors associated with this desire are not fundamentally different from those of women in the general population. Maternity may even help them face a situation they still consider stigmatized and prefer not to disclose. Health-care protocols for handling HIV-positive women should incorporate specific interventions on sexual and reproductive health to help them fulfill their procreation desire and experience safe pregnancies.

Introduction Different studies showed that during the economic recession that started in 2008, some health indicators deteriorate and others remained stable [1][2][3]. Analyses of the effects of financial crises on inequalities in health and mortality led to heterogeneous results, with inequalities increasing in some countries and remaining stable in others [1,4,5]. Cirrhosis and other chronic liver diseases are a major and preventable cause of death worldwide; the Global Burden of Disease Study 2017 described these diseases as the 13th leading cause of years of life lost globally [6,7]. Variations in cirrhosis mortality between countries reflect differences in risk factors, such as alcohol use and hepatitis B and C infection [8]. Several studies have described inequalities in mortality due to cirrhosis [9][10][11][12] and alcohol use [13][14][15]. However, few studies have analysed trends in liver-related and cirrhosis mortality after the start of the economic recession of 2008 [7,[16][17][18]. Risk factors of cirrhosis mortality may change during economic recessions and influence the trends and socioeconomic inequalities in chronic liver diseases and cirrhosis mortality. For alcohol consumption, there is a contradictory evidence on the impact of economic crises on this consumption and alcohol-related-health problems. Possible mechanisms related to how these crisis affect alcohol consumption have been described as psychological distress, triggered by unemployment and income reductions, that increases alcohol consumption and secondly tighter budget constraints than imply less money spent in alcoholic beverages and reduce consumption [19]. Hepatitis B and C affect vulnerable populations such as for example persons who inject drugs or migrants of low income countries, populations who have suffered the financial crisis [20]. Therefore, the objective of this study was to analyse the trends in chronic liver diseases and cirrhosis mortality, and in socioeconomic inequalities in this mortality, in nine European urban areas before and after the onset of the 2008 financial crisis. --- Methods --- Design, Units of Analysis, and Study Population This is an ecological study of trends in three periods of time: two periods before the 2008 economic crisis and one during the crisis . This study is part of the EURO-HEALTHY project , through which other causes of death have been described [21,22]. The units of analysis were the geographical areas of nine European cities or metropolitan areas . The study population consisted of the individual residents in these areas during the three periods mentioned above. --- Data and Measures The data and measures used were the following: 1. Mortality, corresponding to the number of deaths due to chronic liver diseases and cirrhosis -9: 571; ICD-10: K70 and K74) by area, age group, and sex. Data were obtained from registries provided by local and national statistics institutions. The indicator of chronic liver diseases and cirrhosis mortality used for the analysis was the standardised mortality ratio , calculated by dividing the observed number of deaths in the study areas by the expected number of deaths . The latter was calculated considering the age-specific mortality rates in the standard population of the European Union -28 in the year 2007, and the population of the city/area by age groups. For descriptive purposes, we calculated the crude mortality rate , and the indirectly standardised mortality rate . The latter was calculated by multiplying the SMR in the city/metropolitan area by the crude rate in the standard population. --- 2. Population, corresponding to the number of inhabitants living in the study areas. Most urban areas had data for the entire period studied or for at least 2 years. These data were stratified by age and sex and were obtained from census or population registries. The median population of the areas ranged from 271 to 193,630 . --- 3. Socioeconomic data, obtained from the census, consisting of several indicators selected to identify the level of deprivation of the areas studied. We built a composite socioeconomic deprivation index using principal component analysis within each city. The variables included in the analysis for 2001 were the percentages of: unemployed population ; manual workers ; population with primary education as the highest attainment ; and population with university education . The percentage of manual workers was not available for Stockholm and the index was calculated with the other three indicators. The index of deprivation was the first component of the principal component analysis, which was performed separately for each city; therefore, it was possible to show the pattern of deprivation in each setting. The proportion of variance explained by the first component ranged from 0.578 in Lisbon to 0.918 in the Athens metropolitan area. The principal component analysis was performed using the prcomp function of the R statistical software package. The calculation was performed by a singular value decomposition of the data matrix, which is generally the preferred method for numerical accuracy. --- Data Analysis Due to the excessively variable estimates for the small areas included in this study, we smoothed the SMR and we used the hierarchical Bayesian model proposed by Besag, York and Mollié [23]. This model takes two types of random effects into account: spatial and heterogeneous; the former takes into account of the spatial structure of the data, while the latter deals with non-structural variability. We estimated smoothed SMR for both sexes and periods using the following model: O i ∼ Poisson log ∼ α + S i + H i where, for each area i, O i is the number of observed cases, E i the number of expected cases, θ i , the sSMR with respect to the European population, S i the spatial effect, and H i the heterogeneous effect. Let α be the intercept of the model. Expected cases were calculated by indirect standardisation, taking the liver cirrhosis mortality rates of the EU-28 in 2007 as reference by age . Table 1. Description of the nine European urban areas: number and type of small areas, period years, and total population . Moreover, to analyse the trend in socioeconomic inequalities, we fitted an ecological regression model, which included the socioeconomic deprivation index , the period , and their interaction: O it ∼ Poisson log ∼ α + β 1 D i + β 2 P 2t + β 3 P 3t + β 4 P 2t D i + β 5 P 3t D i + S it + H it where, for each area i and period t , O it is the number of observed cases, E it the number of expected cases, θ it the sSMR with respect to the European population, D i the value of the deprivation index; S it the spatial effect, and H it the heterogeneous effect. Finally, P 2t and P 3t took the following values: P jt = 1 if j = t, and P jt = 0 if j = t. Let α be the intercept of the model and β 1 , β 2, β 3, β 4 , and β 5 the parameters or coefficients associated with the different variables and their interactions. The expected cases were calculated as in the previous model. Changes between periods in the relationship between socioeconomic deprivation index and mortality were evaluated through the interactions included in model 2. Specifically, we studied the change between the first and second period and the second and third period . In the two models , an intrinsic conditional autoregressive prior distribution was assigned to the spatial effect, which assumed that the expected value of each area coincided with the mean of the spatial effect of the adjacent areas and had a variance of σ s 2 , while the heterogeneous effect was represented using independent normal distributions with mean 0 and variance σ h 2 . A uniform distribution U was assigned to the standard deviations σ s and σ h . A normal vague prior distribution was assigned to the parameters α, β 1 , β 2 , β 3 , β 4, and β 5 . Given that the socioeconomic deprivation index scale is dimensionless and arbitrarily fixed, we calculated the relative risk for each city/metropolitan area, which compares liver cirrhosis mortality of the 95 th percentile value of socioeconomic deprivation to its 5 th percentile value . RR estimates were obtained based on the mean of their subsequent distribution, along with the corresponding 95% credible intervals . All analyses were performed using the Integrated Nested Laplace Approximation library of the R statistical package. --- Results Table 1 shows the characteristics of the nine areas studied. Greater London is the area with the largest population size, and Turin is the smallest. In the majority of areas, the median population by sex was under 6000, except districts in Berlin and municipalities in Athens . Of note, the first period included 4 years , and the second and third period, 5 years for most cities. The description of mortality data is presented in Table 2. Chronic liver diseases and cirrhosis mortality rates were higher among men than among women in the three periods across all cities. Rates were lower in Athens, Stockholm, and London , and higher in Berlin. During the third period, death rates decreased in all settings except in Athens among men, where the ISMR increased from 8.50 per 100,000 inhabitants during the second period to 9.42 during the third. Figures 1 and2 show the RR of the association between socioeconomic deprivation index and mortality due to chronic liver diseases and cirrhosis, for men and women during the three periods in each city or metropolitan area. Among men, RR was always higher than 1 in all cities, except for Prague, Brussels, and Berlin. The highest RR in the first period was found in Stockholm ; however, it decreased over time . Turin's RR showed an increasing trend, with the highest value observed in the last period . Barcelona had a very similar increasing trend to that of Turin. However, these differences in RR did not have statistical significance. Among women, a positive significant association between deprivation and mortality was identified in Turin, Stockholm, London, and Barcelona. During the third period, the RR only increased in London, although without statistical significance . In the other cities, instead, the RR decreased during this period although without statistical significance. The decrease in RR in Lisbon was noteworthy . --- Discussion This study found that mortality rates of chronic liver diseases and cirrhosis mortality decreased in all cities during the financial crisis period, except among men in Athens. Mortality rates in men were higher than those in women. Moreover, socioeconomic inequalities in liver cirrhosis mortality were found in 6 cities/metropolitan areas among men, and in 4 among women. Although changes in inequalities were not statistically significant among men, these increased in Turin and Barcelona. Among women, several cities had lower RR in the third period. Previous studies on the long-term trends in cirrhosis mortality showed it decreased in the last four decades in the majority of countries in Western Europe, except for the UK, Ireland, and Finland [8,[24][25][26][27]. This decline was mainly due to a decrease in alcohol consumption and an improvement in alcohol quality. Indeed, in EU member states, around 60-80% of deaths from liver disease are due to excessive alcohol consumption [2][3][4]. However, some countries in Eastern Europe had higher rates of cirrhosis-related mortality. Such mortality mainly increased during the 1980s and early 1990s as a result of the dissolution of the Union of Soviet Socialist Republics and the use of low-quality alcohol; more recently, rates decreased in the majority of these countries [25,26]. In Central Europe, cirrhosis-related mortality rates were high with a stable trend [8]. In our analysis, the only city where the rates did not decrease was Athens, where the financial crisis was particularly impactful. On this note, an increase in cirrhosis mortality and alcohol consumption in the Greek population aged 15-49 has been presented by the Global Burden of Diseases project [28]. Bosque-Prous et al. carried out a study in 2006-2013 on alcohol consumption among adults aged 50-64 and found the prevalence of hazardous drinking decreased and abstention from alcohol consumption increased in the majority of European countries, except for the Czech Republic, where the opposite trends were observed among men [29]. However, the review carried out by Dom et al. [30] found mixed results, mainly related to a general decrease in alcohol consumption and an increase in harmful use within specific vulnerable social subgroups. As described above a systematic review found that, during economic crises, increased psychological distress and tighter budget constraints due to income reductions are the main behavioural mechanisms behind alcohol consumption and alcohol-related health problems. The former was found to lead to higher and more frequent alcohol consumption, and to a higher prevalence of drinking problems to cope with the distress; the latter can lead instead to lower alcohol consumption [19]. Although in Europe alcohol is the main contributor to chronic liver disease and cirrhosis mortality, other causes include hepatitis B or C infection and non-alcohol fatty liver disease, associated to obesity and type 2 diabetes. Pimpim reported that in this century viral hepatitis mortality increased in most Northern European countries, Hungary, Italy, Croatia and Portugal, but remained stable or decreased in others [26]. Ireland et al. observed a stable trend in mortality among individuals diagnosed with hepatitis C in England [31]. The fact that in our study chronic liver diseases and cirrhosis mortality was observed to be higher among men than women might be due to the higher alcohol consumption among men. Of note, a study of Bosque-Prous et al. [32] found that European countries with higher gender equality presented lower gender differences in hazardous drinking, mainly due to higher consumption among women. Due to this, the authors recommended policies aimed at improving gender equality to be accompanied by specific alcohol control policies. We found socioeconomic inequalities in chronic liver diseases and cirrhosis mortality in men in the majority of cities, and in fourcities among women. These inequalities are related to the different factors associated with chronic liver diseases and cirrhosis, namely alcohol consumption, and hepatitis B and C. It is important to take into account that deprived areas have a higher proportion of vulnerable populations, such as drug users and marginalised groups, who can be at a greater risk of contracting hepatitis B and C [20], and exhibit higher alcohol consumption [33]. However, some reviews that have analysed arealevel factors on alcohol use [33][34][35] did not find clear evidence that area-level disadvantage is associated with increased alcohol use. Previous studies have shown also socioeconomic inequalities in cirrhosis mortality in urban areas of European cities [15,36]. Moreover, studies that have analysed socioeconomic inequalities in alcohol-attributable mortality have found it is higher in lower socioeconomic groups in the majority of countries, and is similar in men and women [14,33,37,38]. Ford et al. described the increase of hepatitis C mortality in New York, from 2006 to 2014, and the highest rates occurred in neighbourhoods comprising socioeconomically disadvantaged groups [39]. Our study found that, during the 2008 economic crises, changes in the trends of socioeconomic inequalities in chronic liver diseases and cirrhosis mortality did not have statistical significance. Nevertheless, an increase was observed among men in Barcelona and Turin in the third period. Other studies analysing the trends in inequalities in cirrhosis mortality before the economic recession found different results: no increase for Barcelona [9] and an increase in Australia [12] The authors suggest that the increase in socioeconomic inequalities in Australia was due to the increased harmful alcohol consumption among lower socioeconomic groups, which might be attributed to a relative increased affordability of alcohol over time. Crombie and Precious described a reversal in the social class gradient of cirrhosis mortality during the 20th century . These findings indicate a major change in risk factor distributions across social classes, possibly explained by differential changes in alcohol consumption [40]. The study by Mackenback et al. [14] found a rise in inequality in alcohol-related mortality through the years in several countries in Eastern Europe and Northern Europe . These findings can be explained by a rapid rise in alcohol-related mortality in lower socioeconomic groups. These trends should be monitored in the near future to observe their progress. --- Strengths and Limitations This study reported, for the first time, chronic liver diseases and cirrhosis mortality and socioeconomic inequalities in 9 cities/metropolitan areas of Europe before and during the financial crisis. It is worth mentioning that vital statistics are a good source of information to study cirrhosis mortality in Europe [8] This study has some limitations. First, data used for socioeconomic indicators were dated from 2001, and the deprivation of areas could have changed over time; however, the rank of areas by socioeconomic deprivation has probably remained unchanged. Second, the number of areas included per city varied considerably, which may have affected the robustness of estimates ; we addressed this by using Bayesian methods to smooth the SMR of small areas. Third, the size of the small areas differed between the urban areas studied: smaller areas are more homogeneous and the possibility of observing stronger effects is higher, which may partly explain some of the higher RRs for Barcelona, Stockholm and Turin mainly among men. --- Conclusions This study shows a decrease in chronic liver diseases and cirrhosis mortality in men and women in European urban areas ; the existence of geographical socioeconomic inequalities in this mortality; and the lack of significant change in these inequalities over time, after the onset of the financial crisis of 2008. However, these results should be monitored in the long term in order to study future changes. Our findings hint at implementing policies to decrease risk factors for chronic liver diseases and cirrhosis, mainly among vulnerable populations. For alcohol consumption, we suggest policies such as: tax increase, the establishment of a minimum price for alcohol [24], marketing restrictions, and screening and behavioural interventions. Additionally, vaccination for hepatitis B for all population groups and the implementation of harm reduction programs for drug users should be prioritized [26]. --- Data Availability Statement: Data can be shared upon request to the author. Acknowledgments: The authors thank Sani Dimitroulopoulou and Christina Mitsakou from the Department of Environmental Hazards and Emergencies, Public Health England, UK, for providing the UK data. Claudia Costa received support from a research fellowship from the Portuguese Science and Technology Foundation . Paula Santana received support from CEGOT, funded by national funds through the Foundation for Science and Technology under the reference UIDB/04084/2020. ---

Objective: To analyse the trends in chronic liver diseases and cirrhosis mortality, and the associated socioeconomic inequalities, in nine European cities and urban areas before and after the onset of the 2008 financial crisis. Methods: This is an ecological study of trends in three periods of time: two before (2000-2003 and 2004-2008), and one after (2009-2014) the onset of the economic crisis. The units of analysis were the geographical areas of nine cities or urban areas in Europe. We analysed chronic liver diseases and cirrhosis standardised mortality ratios, smoothing them with a hierarchical Bayesian model by each city, area, and sex. An ecological regression model was fitted to analyse the trends in socioeconomic inequalities, and included the socioeconomic deprivation index, the period, and their interaction. Results: In general, chronic liver diseases and cirrhosis mortality rates were higher in men than in women. These rates decreased in all cities during the financial crisis, except among men in Athens (rates increased from 8.50 per 100,000 inhabitants during the second period to 9.42 during the third). Socioeconomic inequalities in chronic liver diseases and cirrhosis mortality were found in six cities/metropolitan areas among men, and in four among women. Finally, in the periods studied, such inequalities did not significantly change. However, among men they increased in Turin and Barcelona and among women, several cities had lower inequalities in the third period. Conclusions: There are geographical socioeconomic inequalities in chronic liver diseases and cirrhosis mortality, mainly among men, that did not change during the 2008 financial crisis. These results should be monitored in the long term.

Introduction Due to the ongoing synthesis of civilizations, multiculturalism and its background philosophical paradigm have aroused intense academic and professional interest as key notions in vast domains of society and culture. Philosophy and linguistics have been analyzing, usually, multicultural and paradigmatic aspects of modern worldview and language; economics, social and media studies, politology have been studying this duet as a set of fundamental trends that regulate the socio-economic environment. However, the majority of reflections on global contemporary society and its culture have still adhered to these transdisciplinary buzzwords: postmodernism and multiculturalism, in latest years academic studies more often turn the spotlight on metamodernism -a paradigm-to-be of the new era of pluralism and relativism. Therefore the article discusses relevant views on a contemporary philosophical paradigm and multiculturalism; analyzes how these social and cultural fundamentals are reflected in US multicultural literature with the help of linguistic means and thematic content; considers the possibility of placing much of multicultural literature in a framework of metamodern perspective. --- Methods and materials The diversity of accepted thoughts causes the absence of a definitive agreement on the topic among modern philosophers, and thus itself stands out as a hallmark of contemporary worldview standards. A critical comparative review of different academic ideas helps to come up with an overall appraisal of the complex interplay between the research notions. Textual and linguistic analyses are used in our survey to examine in detail American multicultural fiction. The applied materials are selected from a wide variety of works by contemporary US authors written, mostly, in English, and meet the following criteria: prose, officially or conventionally considered as multicultural without 'cultural appropriation' , shortlisted for awards, positively reviewed by public critics. The applied focus of the study is on the combined expression of metamodern philosophy and multicultural discourse in multicultural literature. full awareness of the phenomenon and the absence of new definitions mean the end of the period. The above leads us to the second group of thinkers so-called 'chief ideologists' of postmodernism as an independent philosophical stance responding to challenges of the post-industrial civilization. J.Baudrillard, P.Bourdieu, J.Butler, J.Derrida, M.Foucault, J.-F.Lyotard, I.Hassan, JR Welch, and many others determined the direction of the social and cultural movement since the 1950s under the banner of postmodernism. Although the first symptoms of a new mood and world order or more precisely, 'post-order', were already observed at the beginning of the 20th century. They were due to economic, and socio-political events , which gradually changed the Western system of thinking and formed subsequently a modern consumer society with mass production of everything . Further events of the century contributed to the final agony of the modern era. Society lost faith that man is naturally good, endowed with innate inalienable rights to freedom, equality, and happiness, or that the Reason is the only way to the realization of these aspirations, as the 'reasonable' conquest and transformation of nature led to an ecological crisis . There was a need to revise the guidelines for explaining human existence, justifying the objective reality. So postmodern ideology emerged as an inevitable special form of cognition of the complex world, it started to build a system of principles of being and its cognition without obligations to give unambiguous answers to the ultimate questions of classical philosophies, such as 'What is truth?', 'What is good?', 'What is Man?'. Typically J. Lyotard's 'The Postmodern Condition' from 1979 is considered to be the starting point of postmodernism as a philosophical and cultural trend of the XX century. Later, mature postmodernism once again awoke the natural human need for ideological asylum which resulted in the extremes of religious fanaticism and ultra-right nationalism adjacent to absolute sexual freedom and the ideology of consumerism expressed by Barbara Kruger's conceptual artwork 'I shop, therefore I am'. At the end of the 20th century, Western society, counting the sacred beads of simulacrums and discursive practices, copied, recycled, and collaged second-hand materials gathered, yet, on the ruins of modernism to restructure consciousness and ideology. Accordingly, the traditional study of the universal laws of social development, its moral categories, and guidelines became impossible. Attempts to defragment and systematize the worldview and structure of society lead to an artificial separation of elements and the loss of their meaning outside of unsystematic unity. Philosophy became a game of discourses that started to shape an epistemic reality . Postmodernism as an attribute gained intellectually elite status generations ago and it led paradoxically to the extensive use of the word and its derivatives in a variety of contexts often without a proper philosophical, social, and cultural understanding of the notion. As a result, there is still a generally biased idea about postmodernism, particularly its manifestations in philosophy and arts. In case of insufficient academic background all deviations from the classical canon, even modernistic techniques, have been automatically ranked as postmodern indicators. Purposefully we referred to Wikipedia, the collective pool of information, to describe postmodernism as understood by an average creator of contemporary knowledge sources. A word cloud was generated from keywords chosen as contextually meaningful elements in the articles about postmodernism [https://en.wikipedia.org/wiki/Postmodernism] and postmodernity [https://en.wikipedia.org/wiki/Postmodernity]. These words and phrases may serve as an information retrieval system to show the general idea of the concept. Yet the attention must be drawn to the point that although the entries had been edited three days before the date of access, on analyzing the references , we see that the latest reference was published in 2018, and the latest recommended further reading to 2016, besides both are explanatory dictionaries on postmodern-related topics. The majority of referred works date back to the 1980 and 1990s. No up-to-date academic articles and philosophical essays on postmodernism and postmodernity are cited. From this simple observation, it follows that postmodernism is not a relevant issue anymore, it had been dissected and described in detail even before the XXI century started. Again it follows that for some time we have been already looking at postmodernism from the position of Bachtin's outsideness and the majority has voted to declare the death of postmodernism. --- Metamodern and Multiculturalism Yet, wordlviews thus arts and literature still mirror relatively much of postmodern philosophy, undoubtedly it is an outdated mode of discourse for the current mode of thinking as its conceptual apparatus does not fit the complex diversity of a truly globalized and multicentered community. 'A Greta Thunberg' has already redirected the pessimistic narrative of postmodernism toward the brand-new era. 'Hypermodern' , 'digimodernism and/or pseudomodernis' 'automodernism' , 'altermodernism' 'remodernism/reconstructivism', 'renewalism', 'stuckism' , 'the New Sincerity' , 'The New Weird Generation' , and, of course, post-postmodernism and metamodernism are the terms that have been used in philosophical and artistic discourse of the recent decades. Based on the brief comparison of the above theories and a wide use of the term, we have chosen 'metamodernism' as the key word of the research. Metamodernism, which undoubtedly should be considered the current worldview paradigm expressed through a variety of mindsets, practices, and forms in philosophy, arts, politics, etc. Therefore, we are currently experiencing Metamodernity and its definition is an ongoing discussion. The social, cultural, and political backgrounds affect the context observability and perspectives of onlookers. Summarizing a variety of ideas on metamodern given by thinkers and artists like A.Vermeulen, R. van den Akker, D. Görtz, T.Björkman, J. Rowson, L. R. Anderson, T. Amarque, S. Knudsen, L.Turner, A. Kardash, O.Mitroshenkov we see that they agree that the three derivatives of 'modern' are closely linked to each other. This connection or collaboration can be illustrated by a construction site where rational modernism built a skyscraper from concrete bricks; later postmodernism mockingly build a classical full-sized mansion from plastic blocks; and now metamodern is simultaneously reading building plans of previous engineers to selectively use ideas for erecting a tall-to-be structure from colorful concrete Lego blocks. In other words, metamodernism is an active crisis manager solving problems caused by globalization and technological progress. Rational reasoning is on the trend again but emotions are taken into account as well; depression is diagnosed and accepted but medication is being taken and there is hope and eagerness to gain mental health; disappointment and frustration from previous errors motivate the creation of a new not-so-brave but conscious world; information is available but it is delivered by a discorded orchestra of Wikipedia-like authors but one can always choose a suitable source; identity is important but one is allowed to express multiple identities. So other notions that fuel the metamodenistic philosophical engine are pragmatic idealism, glocalization, transsentimentalism, neoromantisism, collective intelligence, space virtualization, digitalization of society, informed naivety, moderate fanaticism, cultural sensibility. The real keywords of this oxymoron epoch are pluralism, diversity, and thus all the multi-derivatives. A metamodern plurality of realities is only possible with a special doctrine carefully crafted to keep multiple elements in balance. Globalization thus mass migration directly resulted in the development of an actively discussed policy of multiculturalism that started as a must-have public policy focused on the social structure and culture of Western countries. Interpretations of the policy oscillate metamodernisticly between radically diverse ideas from "multiculturalism leads to undeniable ghettoization and national disintegration" to arguments that "equal inclusion, access to education and power help cultural minorities to positively contribute to their new homeland" . Maintaining the balance between diversity and assimilation resulted in attempts to measure and calculate the effective figures for multicultural policies in projects like Multiculturalism Policy Index or Migrant Integration Policy Index . For obvious reasons, social diversity, and consequently, multiculturalism mostly have been fundamental for economically developed and thus attractive to immigrant countries of Europe or North America. The modern US national model relied on the principal significance of cultural diversity along with politically preferred integration into the American community . As the community includes a vast variety of races and ethnic groups, it is natural that equality is assumed to be related to cultures as well as to individuals. On this basis, multiculturalism seems to be that political philosophy that handles the challenges of effective social interaction in a culturally-diverse metamodern community. This political philosophy in action upholds the immutability of cultural differentialism and reflects collaboration and the creative paradox of meta-modernity in the interdisciplinary discourse of social relations. --- Metamodernism in Multicultural Literature Art has always been a basic platform to express worldviews and cultural or social identities. Adjusting Parech's ideas on multiculturalism to related literature we consider that multicultural literature contributes to the integral unity and consolidation of a culturally mosaic metamodern community . The interpretation of both the content and form of multicultural works can help to illustrate the metamodern principles in general. To frame metamodern social and cultural fundamentals we refer to already definitive studies by Vermeulen, van den Akker, and Luke Turner's The Metamodernist Manifesto because most of the on-topic discussions are based, support, or criticize the above-mentioned works. It is important to note that we analyze works of contemporary American authors who are first or secondgeneration immigrants from around the world. Despite some standard categorizations of multicultural authors according to their origin like Latino Americans, East Asian Americans, etc given, for example, in the report of The Bureau of International Information Programs of the U.S. Department of State , we study a wide range of contemporary prose fiction without focusing on formal grouping. Maxine Hong Kingston says "... I have had to translate a whole Eastern culture and bring it to the West, then bring the two cultures together seamlessly. That is how one makes the Asian American culture" . We believe that distinctive characteristics of multicultural works are universal and present in most culturally blend works regardless of authors' origins or topics they write on. Multicultural authors drag and drop ornaments from background cultures into American literature making it truly multicultural, and thus metamodern. According to Ulinich, context and imagination play a greater role in understanding the real meaning of some culturallybound words and phrases. As a real follower of metamodernism, she reminds us that there is always a wider narration behind simple forms . So Anya Ulinich, Chimamanda Ngozi Adichie, Jhumpa Lahiri, Gary Shteyngart, Lisa See, Jamie Ford, Junot Díaz, Jean Kwok, Gish Jen, Cristina Henríquez, Lisa See, Nicole Dennis Benn and many more works have been used in our research to analyze and interpret the form and the content while drawing a parallel with declarations of metamodern philosophy. The most distinctive accent of multicultural literature that reflects the metamodern condition of pluralism and diversity is the use of language means. Bilingualism, codeswitching, misspeaking, extra cultural insertions and intertextuality are common definite techniques in the majority of multicultural works. For example, the Spanish-English duo in the short-stories collection "Come Together, Fall Apart" where Maria is rehearsing her speech with her grandfather who embodies her cultural roots and native tongue: "Hola, Abuelo. ¿Como te sientes? Me excitan para verte" . Here bilingualism functions as a selective diglossia to solve exact communicative aims, as a password to enter a cultural community, and it reveals characters' borderline identities. Code-mixing and all variants of language alternations like pidginization, hybridization, translanguating, crossing, etc are occasional language mixings that multilingual characters use their languages as an integrated system of communication. It creates a specific local and ethnic context of the characters' community. Every single paragraph of "In the Time of Butterflies" by Julia Alvares has a coded message for 'those who know': "But Manolo's warmth could thaw any coldness. He gave his old business partner un abrazo, addressing him as compadre even though neither one was godparent to the other's children. He invited himself in, ruffled the boys' hair, and called out, "Doña Leila! Where's my girl?" Obviously, the boys suspected nothing. They yielded reluctant kisses to their mother and aunt, their eyes all the while trained on the screen where el gato Tom and el ratoncito Jerry were engaged in yet another of their battles. Doña Leila came out from her bedroom, ready to entertain. She looked coquettish in a fresh dress, her white hair pinned up with combs. "¡Manolo, Minerva! ¡Qué placer!" But it was Dedé whom she kept hugging" . In this extract, non-English inclusions express special ceremonial refinements and manners accepted in Latin American culture. Replacement of one syllable English "hug" to Spanish "un abrazo" converts neutral "hugging" into Latin American "a hug to the heart" -holding someone tightly and amorously in one's arms. Addressing a business partner as 'Compadre' -"godfather" shows respect and heartfelt sympathy and highlights a significant connotation of a godparent in Dominican culture. It is always a linguistic puzzle as authors do not translate: "Oai deki te …" She paused. "Ureshii desu," Henry said, softly." Only the context and tone can give a hint: "They stood there, smiling at each other, like they had done all those years ago, standing on either side of that fence." . In the same way, we need to read between lines to understand the current metamodern diversity of life. A motto of metamodern philosophy, "Making mistakes is OK!", should be noted separately. Characters misstep, misinterpret, mishear, and speak no perfect English but it is a driving force of a storyline and characters' development. Dynamic literary technique based on misspeaking is used by Kwok in "Girl in Translation". A-Kim's broken English improves from chapter to chapter, telling the story of her transformation from A-Kim to Kimberly: "Why are you late ?" This, I understood. "I sorry, sir," I said. "We not find school". versus the last page lines: "I worked four jobs at a time while I was a student, but I still graduated with honors and then moved on to Harvard Medical School. In those debt-ridden years before I finished medical school, I called upon any and every talent I had to become the best surgeon I could" . Jen Gish also plays with grammar in "Who's Irish?": "I am work hard my whole life, and fierce besides. My husband always used to say he is afraid of me, and in our restaurant, busboys and cooks all afraid of me too". Moreover, Gish puts noninclusive phrases into the mouth of her character violating language or behavior norms thus making her sound even more foreign: "They say they cannot find work, this is not the economy of the fifties, but I say, Even the black people doing better these days, some of them live so fancy, you'd be surprised. Why the Shea family have so much trouble? They are white people, they speak English. When I come to this country, I have no money and do not speak English" . According to the metamodern attitude, constancy and perfection are illusory while mistakes make sense and inspire future changes. In multicultural literature intentional incorrectness, direct transcription of accents, and misapplication of words provide ground for a deeper understanding of the context. However, this language game is not just fine literary techniques, it is an actual echo of borderline personalities who are naturally incorporating alien blocks into full English flow thus, gradually expanding the norms of General American. Globalization, digitalization and mass migration have significantly increased the multicultural and multilingual population -these people are always going back and forth between their languages and cultural practices intentionally or intuitively picking the best option for a particular context. So metamodern diversity and pluralism mean multilingualism of expression. The above-described language shift leads to the first classical postulate of metamodernism that "oscillation is the natural order of the world" , which widely manifests itself in the vast majority of multicultural works through oscillation between two dimensions: actual citizenship and ethnocultural background and each dimension may dominate in a certain context. Duality of cultural consciousness leads to the discordance of social perception and self-image of cultural 'hybrids' and explains the core conflicts of multicultural literature -cultural and national identity search in the ultra version of "the identity is a task" demand of the metamodern world [Two opposite opinions on cultural diffusion coexist in the multicultural community and thus in multicultural literature: some agree with Amy Tan's character who is lamenting "I wanted my children to have the best combination: American circumstances and Chinese character. How could I know these two things do not mix?" , while others join cosmopolitan Elif Shafac who states the opposite "East and West is no water and oil. They do mix" . But let us consider a common stereotypical situation illustrated in multicultural literature: a young girl, having come to America recently, feels she is "too Chinese" in society and "not enough Chinese" at home. The problem escalates until she accepts belonging to both cultures and fully acknowledges her fusion identity . This is a common lifelong journey for most newly American real and fictional characters who form fluid identities that fully fit in the "liquid modernity of the metamodern community" . In multicultural communities, metamodern globalization blurs cultural borders and, simultaneously stimulates cultural identity search. The latter is impossible without returning to cultural roots or at least without trying on traditional rituals. This explains why metamodern is often understood as a mediator between premodern and modern codes. Characters in multicultural literature usually feel nostalgia and romanticize native codes. Recounting past lives in "The Hundred Secret Senses" by Amy Tan, return migration in "Americanah" Chimamanda Ngozi Adichie, and the immigrant feeling of "two homes" in "Grandfather's Journey" by Allen Say are common storylines: characters, whether physically or mentally, make journeys of identity, meaning of life or family history back to their homeland or to the past just to get aware of their current foreignness to their cultural roots; at the same time, the bond is unbreakable, it forces migrants to repeat journeys again and again. So Vermeulen's visual image of a swinging pendulum is perfectly applicable to describe cultural nomadism, identity search, and idealization of cultural practices in a multicultural context. The consideration of constant oscillation between innumerable cultures as adequate and acceptable is a cross-cutting theme of multicultural literature and "a sort of open source" metamodern philosophy in general. According to Vermeulen and van den Akker, Hegel's positive idealism concluded, a metamodern movement "inspired by a modern naïveté yet informed by postmodern skepticism … consciously commits itself to an impossible possibility" . Characters of multicultural literature fully fit the related definition as they or their ancestors were desperate and hopeful enough to proclaim the end of one history and start from zero. Metamodern optimism and doubt made many immigrant stories possible as they are told in "Girl in Translation" by Jean Kwok, "Esperanza Rising" by Pam Munoz Ryan, "Still Here" by Lara Vapnyar, "Unaccustomed Earth" by Dzumpa Lahiri, etc. So these works describe a metacocktail mixed from postmodern misery and pragmatic neo-romanticism that encourages a metamodern person to chase happiness over "a plurality of disparate and elusive horizons" "that are forever receding" . A model of a metamodern individual is reflected in multicultural literature through approaches and personality traits. Moreover, metamodern reconstruction, sensibility, metaxy, sincerity, hope, praising mistakes, and pluralism are key principles of a multicultural personality. The condition of a constant identification journey makes these characters "not smug or deconstructive, but reconstructive" . By reconstructing fragments from motherlands left behind, they enter the internal and external dialogues with new homelands and co-create the diversity of metamodern globalized perception. In addition, "immigrants on their way to ultimate assimilation… present themselves as partially alienated strangers" by dragging and dropping into US English language and cultural elements. Anya Ulinich fills her award-winning "Petropolis" with Soviet realias and calques of Russian idioms. "District 7 is all the way up the devil's horns," replied Sasha, trying hard to hide her relief. "Are you sure the place is fit for the intelligentsia?" -"Don't sneer, detkа", sighed Mrs Goldberg "You don't need another tic" -the short extract includes: a) direct translation of Russian idiom "up the devil's horns" meaning "in the middle of nowhere"; b) the Soviet term "intelligentsia"; c) transliterated concept "detka". The text looks like a linguacultural puzzle fully open only to readers with Soviet backgrounds. Ulinich does not deconstruct English, she intentionally and artistically reconstructs her ex-language on the American ground. We could say that multicultural literature with its collaborative reconstructions and "interconnection and continuous revision" of the past provides "the possibility of grasping the nature of contemporary cultural and literary phenomena" of metamodern plurality. Often borderline sensitivity in multicultural literature mixes reality and spiritual worlds into a single stage for the live performance of "real" characters, memories, gods, spirits, and previous lives. Characters coexist in two or more worlds of native and American culture, religious and civil life, etc; for every situational context, they have an exact personality with definite practices or languages. In J.Ford's "Hotel on the Corner of Bitter and Sweet", young Henry Lee demonstrates a personality change when switching Cantonese and English, his Chinese and American versions take center stage in turns . This frequent artistic approach in multicultural literature accurately mirrors the objective metaverse of "digimodernism" where virtual personalities are already on equal terms with humans. The metamodern community lives between the real and virtual worlds -an average person may have several self-presenting online images and life has become a performance. Metamodernism has normalized being multicultural and switching multiple personalities. Moreover, the dual sensibility of a multicultural person allows them to see "several points of view simultaneously, perceive opposite ideas, and, as a consequence, more holistically, nonlinearly perceive the series of events and phenomena" , namely to be an open-minded, tolerant and truly metamodern personality. One more metamodern approach to life reflected in multicultural literature is a rebirth of early modern "desires for change and heartfelt "yes we can' calls" . Multicultural, usually immigrant literature, is a complex narrative about a long and challenging path to rebuilding a new life and social connections. It is not always a story of success but of finding peace with oneself and harmony in a world of contradictions. Multicultural characters illustrate the current human condition -they almost died of postmodern depression but now they "express a hopefulness and sincerity" toward the future. "Do not ever be afraid to start over" , "But if some things are fated and some people are luckier than others, then I also have to believe that I still haven't found my destiny. Because somehow, some way, I'm going to find Joy" , "Then I took a deep breath, got off the bed and opened the door." , "Everything glitters in the new sunlight, just like Margot had always thought it would". -this random compilation of the last sentences from some multicultural novels expresses a typical metamodern personality's informed, sentimental but optimistic attitude. --- Conclusion Be it repeated that arts always pioneer noticing and expanding changes in humanity's mood. And, of course, the attitudes and principles of the metamodern paradigm have long ago found their way to the artistic content of contemporary literature. Since the buzzword of the present academic, political, and cultural discourses are multicultural, multi-approach, multilingual, and other multi+ derivatives, it is natural enough that multicultural literature is on the rise. Compared with the prose of previous periods, multicultural literature is not rationally prescriptive and it is not created to 'make it new or difficult'; it is not pessimistic, mockingly ironic, and it does not intend to 'cross borders and close gaps'. Being a real metamodernistic project, multicultural literature feels neither borders nor gaps "as a space-time is both -neither ordered and disordered" ; its content is a descriptive, narrative and multilayered phenomenon of spiritual and material fusion of experiences that go far beyond one linguistic or cultural code. Overall, multiculturalism, with its multi-perspectival approach in political/academic/cultural discourse, has already started to be employed as an effective construal of metamodern time -'something altogether weird' but "ultimately a complex philosophy that will take academics and artists years to work through -but it's new, it's not jaded, and it puts real people's real emotions and real lives first" . Mixing up David Foster Wallace and Zadie Smith, it may be summarized that multicultural art fully reflects this metamodern "shift away from cynicism towards human connection" with "the useful employment of head and heart" ; so it highly contributes to the intercultural understanding and trust, giving hope for the future of metamodern humanism. Abramson, S. . Metamodernism: The Basics. . The Huffington Post, December 2014, URL: www.huffingtonpost.com/seth-abramson/metamodernism-thebasics_b_5973184 Alvarez, J. . In the Time of the Butterflies. Reprint edition. Algonquin Books. ISBN-10 1565129768 Bakhtin, M., . Speech genres and other late essays. U.S.A. University of Texas Press. ISBN-10. 0292775601. --- Bibliographic references

The article focuses on manifestations of the metamodern philosophical paradigm in contemporary multicultural art, particularly in an American immigrant novel. In recent years, there has been a wide discussion on naming and defining the ongoing reality, so the related theories have been reviewed to identify the current period characterized by globalization and, thus, cross-culture diversification of the communities. According to our research, multiculturalism can be defined as a social policy of metamodern diversity. Analyzing multicultural literature, we have illustrated how the distinguishing features of the metamodern worldview, like constant oscillation, reconstruction, metaxy, pluralism, etc., are expressed in the content and form of multicultural literature. The research findings may contribute to the further understanding the multicultural nature of metamodern principles in philosophy, cultural studies, and social relations.

Background According to the World Health Organization [1], obesity is now a global public health concern. It is estimated that about 340 million children and adolescents are either obese or overweight. From as low as 4% in 1975, pediatric obesity and overweight rose to 18% by 2016, with 18% of girls and 19% of boys being overweight, and 6% of girls and 8% of boys being obese worldwide [1]. Adolescents are of concern because obesity at this stage could lead to premature death, and disabilities in adulthood. Furthermore, as they grow into adults, health conditions associated with overweight including chronic cardiovascular conditions like hypertension will be more severe [2,3]. There are also existing psychosocial impacts like stigma and discrimination [4,5] associated with adolescent obesity. Obesity was previously seen as a problem of highincome countries but, data from the past decade show that low-and middle-income countries , especially in sub-Saharan Africa, bear the double burden of underweight and overweight [6][7][8]. Hence, with the shift in the global burden of adolescent overweight and obesity, current prevalence and trends are required for South Africa, a middle-income, sub-Saharan African country that is experiencing sweeping industrialization and urbanization. Furthermore, the roles of sex and family socioeconomic status are not clearly defined in South African adolescent obesity and overweight trends. As LMICs become more industrialized, they are likely to acquire both the benefits and problems associated with industrialization which include obesity [9]. South Africa as a LMIC is undergoing a socioeconomic transition related to urbanization and industrialization in most of its cities with a corresponding energy imbalance due to consumption of highly processed and caloriedense diets, lack of built structures that encourage being physically active, etc. This nutritional transition is closely linked with obesity transition. The obesity transition model [10] describes global trends in relation to population parameters and determinants that predict the trajectory of obesity and future outcomes based on extensive tracking of available data. Most Sub-Saharan African and South Asian countries are believed to be in the first stage of the transition model, characterised by higher obesity prevalence in women and those with a higher SES among adults [10]. Furthermore, European countries are said to be in the third stage, characterized by higher obesity prevalence among those in lower SES, and a plateau in prevalence among women in high SES and children [10]. A comparison between SA and European adolescent obesity data will facilitate an understanding of the main local and contextual drivers of these trends and prevalence and furthermore, in developing policies for its mitigation. Surveillance data on adolescents overweight and obesity are not regularly collected in SA as done in HICs like in Europe. Although there are several nationally representative surveys conducted at different times over the past 2 decades in South Africa, which reported on the prevalence among South Africans of overweight and obesity, none of them were conducted regularly. Hence, these surveys need to be collectively analysed to investigate prevalence and determine trends to date. In addition, the association of sex and family SES with these trends and prevalence over the past 2 decades requires full investigation. Hence, this study aims to investigate South African adolescents' obesity and overweight trends in prevalence by collating, analysing, and generating trend patterns with data from national surveys. In addition, it aims to investigate and report the sub-populations at greater risk of adolescent overweight and obesity based on sex and family socioeconomic measures. Furthermore, to compare South African and European trends and identify contextual and local factors related to the obesity transition conceptual model [10]. These will facilitate developing context-relevant strategies and policies for mitigation and prevention of adolescent obesity and overweight in SA and sub-Saharan countries. --- Methods The study is an exploratory and quantitative analysis of primary data collected from three national South African surveys and compared to a continental European survey report to investigate trends in prevalence of overweight and obesity. SA Survey data were independently collected, and analyses were performed on the pooled samples to draw trend-related inferences. --- Data sources The South African surveys and the year of data collection are as follows: SA DHS was conducted 3 times -1998, 2003 and 2016 but 2003 data was removed from the public domain because of concerns about its quality, hence this study extracted and analysed data only from 1998 and 2016 for which the data qualities are acceptable and are in the public domain. We further reported findings from the European survey report - The HBSC survey is conducted every fourth year and survey years included are the most recent and comparable years with South African surveys. These survey years applied similar World Health Organization growth chart as used in the South African cohort for the classification of overweight and obesity, to aid in the comparison of the two cohorts [13,14]. Primary data of participants aged 15-19 years were extracted from SA surveys and analysed. The total sample size of SA participants across the 3 surveys used for the study is 27 884 ) age: 16.83 ± 1 years). In this study, we present findings based on the HBSC survey published report for participants aged 15 years old . Two of the SA surveys' sampling methods applied the Statistics South Africa's census enumeration area as the primary sampling unit -SANHANES and SADHS -while the YRBS used schools as the PSU. All three surveys applied a stratified cluster sampling design as follows; in the YRBS, classes within selected schools were sampled, and all individuals in each selected class were eligible to participate. In SANHANES and SADHS, households within selected EAs were randomly sampled and individuals within selected households were eligible to participate. The findings from the European survey were based on the published summary report [14] hence HBSC data were not analysed in the current study. The key measures of interest for this study are as follows: --- Overweight and obesity The SA overweight and obesity data are based on measured weight and height collected according to a standardized protocol for all three surveys [8,11,12,15]. The European report is based on young people's self-reported height and weight . In this study, we computed BMI as weight divided by height squared. BMI was categorised according to the World Health Organization BMI-for age Z-score classification for age and gender [16]. Cut-offs for overweight and obesity were allocated based on the WHO growth reference chart for age and gender. Similar WHO growth reference for age and gender was also used for the classification of the HBSC survey as reported in the protocol [13]. --- Socio-economic status For the SA data, different SES variables related to adolescents' family wealth as measured in the SA setting and as recommended in previous studies were used [17,18]. In the SANHANES and SADHS surveys, variables of interest are household possessions owned by adolescents' families, specifically refrigerator and washing machine, source of drinking water, type of residential house and type of toilet, type of cooking fuel used by the family and presence or absence of electricity supply for the family. These variables were used to compute wealth indices of study participants' families using the principal components analyses method [19,20]. Further, the wealth index values were categorised into one of the following 3 groups -lower, medium and upper to make it comparable to HBSC data [13] and increase statistical power of each of the wealth index groups following the international DHS format [19,20]. For the YRBS data of 2002, 2008 and 2011, SES was assessed using parents' employment status as well as the amount received by adolescents as monthly allowance, in SA Rand . A dichotomous variable indicating if their father and mother are "employed" or "unemployed" was created for parents' employment status as well as a variable indicating if "none", "one" or "both" parents are employed. These indicated the cumulative earning ability of the parents . For the monthly allowance received by adolescents, participants were placed in one of 3 groups as follows lower Father: "Does your father have a paid job? " --- Mother: "Does your mother have a paid job? " Monthly allowance: "In a normal / usual month , how much spending money do you get?" The empirical results about SA participants' families' socioeconomic measures were compared to narrative findings presented in the HBSC report --- Statistical data analyses To investigate overweight and obesity proportion and trends in prevalence in the SA adolescents' cohort, extracted weight and height of participants were imported to STATA software with adolescents' sex and age variables. Body Mass Index -for-age Z-score was computed using the WHO STATA macros algorithm for all surveys. Participants whose BAZ met WHO's recommended category for Overweight and Obesity in all the surveys were included in computing the proportion of adolescents' overweight and obesity in the subsequent analyses for each year and to generate trend plots, frequency distribution and bar plots. We report proportion and trend in prevalence results with time. The significance of trends was assessed using the chi-square test for trend . All data were reported at a significance level of p < 0.05. In addition, trends in prevalence for overweight and obesity were assessed based on adolescents' sex and socioeconomic status measures. Trend plots of overweight and obesity were also compared by adolescents' sex between South African and European data. We also assessed the effects and significance of effects for adolescents' sex and family SES measures on overweight and obesity using odds ratio and Wald test statistics, and reported odds ratio by year. --- Results --- Sample characteristics Table 1 shows sample demographics for adolescent study participants from the three SA surveys extracted for this study. For all surveys, the number of boys and girls were comparable as indicated by similar percentage proportion. The mean BMIs for all survey years are within 21.45 -22.40 kg/m 2 . Adolescents' families in the lower socioeconomic group classified based on the wealth index are in greater proportion than medium and upper groups for survey year 2012 . The proportions of fathers and mothers who are employed are nearly equal in proportion to those who are not, while the disparity between both parents being employed and not, was lowest for survey year 2008 . --- Overweight and obesity prevalence Within the SA cohort, inspite of an initial reduction in proportion by 2002, there was a gradual and consistent increase in the proportion of adolescents who were overweight or obese until 2012 . This appears to be driven by an increase in the proportion of the obese group. By 2016, the proportion of adolescents who were overweight or obese was 21.56% and nearly equal to the Europe adolescents' proportion who were either overweight or obese in 2018. All changes in trends were significant with time. --- Overweight / obesity by sex Across the survey years, our results showed that boys consistently had lower prevalence of overweight and obesity than girls and this difference is statistically significant . From 1998 to 2011, SA boys and girls followed a similar trend in overweight-obesity proportion. Subsequently, a reduction in the percentage of boys' who are overweight or obese was observed until 2016 , while the trend for girls increased further to 33.88% . Wald Chi-square showed a significant association between sex and adolescents' overweight and for all the years assessed -1998, 2002, 2008, 2011 and 2012 -except in 2016 . Odds ratio result showed that by 2016, there was 1.13 higher odds of a girl becoming obese or overweight than a boy in South Africa. In HBSC findings, a greater proportion of boys were overweight or obese from 2010 to 2018, when compared to girls, in contrast with SA where a greater proportion of the cohort who were overweight or obese was girls . Furthermore, there was almost no change with time in Europe's percentage of adolescents who were overweight or obese compared with their SA counterparts, where there was a considerable change with time. --- Obesity / overweight by socioeconomic groups SA wealth index data from 1998, 2012 and 2016 showed that obesity prevalence among adolescents from families with a medium SES was the highest at all three-time points . However, the trends indicated that obesity prevalence in adolescents from families with higher SES increased with time, in contrast with medium and lower SES groups, and peaked at 5.90% in 2016 . Furthermore, adolescents from medium and lower SES families followed a similar trajectory that showed decrease in overweight and obesity proportion recently, which is different from adolescents from upper SES whose proportions have increased recently . The differences in the proportion of adolescents who were overweight or obese with time from the three SES groups were statistically significant . Data from the 2002, 2008 and 2011 YRBS showed that the proportion of adolescents who were overweight or obese and with employed parents was slightly higher than those whose parents were unemployed . The difference in proportion of adolescents who are overweight or obese with either employed or unemployed fathers was significant but not for mothers. Figure 4b, also showed that the number of overweight or obese adolescents who received higher monthly allowance was greater compared to those in the medium and lower groups. The difference in proportion between the three groups was significant with time. Table 3 showed that family wealth indices for lower and medium SES families were significantly associated with adolescents' overweight or obesity only in 2016, but monthly allowance of medium and upper groups of adolescents was significantly associated with either being overweight or obese in 2008 and 2011 . Furthermore, fathers' employment status was not associated with being overweight or obese, but mothers' employment status had a significant association with adolescents being either overweight or obese in 2002, 2008, 2011 . --- Discussion The current study uses the most recent, available, and nationally representative data to investigate the trends in the prevalence of adolescents overweight and obesity in South Africa. The study compared SA adolescents' overweight and obesity prevalence to a European cohort from a reliable European adolescent survey report and further applied the obesity transition model to predict the sub-populations at greater risk with respect to gender and socioeconomic status. Our findings showed statistically significant increase in the proportion of overweight and obese adolescents from 1998 to 2016 in SA. Furthermore, a similar percentage in the SA cohort was observed in in 2016 when compared to that of Europe in 2018. We also found a significantly greater proportion of girls being either overweight or obese in South Africa in contrast to Europe and further among adolescents from high SES families. --- SA trend in prevalence of adolescent overweight and obesity Our findings of increasing overweight and obesity with time among SA adolescents confirmed our initial hypothesis and findings from previous studies which showed that obesity prevalence among South African adolescents increased with time [21,22]. This, as reported by other studies is not unrelated to current urbanization and industrialization sweeping through most LMIC, including South Africa, and the associated nutritional transition from more traditional healthy diets to westernized, refined and highly processed foods [10]. Furthermore, environmental factors encouraging a sedentary lifestyle and other unhealthy dietary behaviours are implicated as major drivers of the increasing prevalence [22,23]. Our findings also showed that girls and adolescents from high socioeconomic families were at greater risk of being overweight or obese. For example, in 2016, girls were more than 10% more likely to be obese than boys in South Africa. Similarly, Grobbelaar et al. [24] and Otitoola et al. [21] have previously identified girls as having a higher prevalence of overweight and obesity than boys in a child welfare home in Durban, Kwazulu-natal and at Comfivaba, Eastern Cape respectively, both in South Africa. It is also evident from our results that among girls, overweight and obesity prevalence have increased over time but decreased since 2011 among boys. There are suggestions that the risk of overweight and obesity in adolescent girls may be related to the onset of menstruation, its associated hormonal changes and psychological impact on physical activeness. In group sessions, adolescent girls shared their perceptions that teenage pregnancies and contraception use contributed to weight gain [25]. In contrast, adolescent boys at the same age, mark an increase in energy needs and physical activities leading to "burning of fats" and reduced energy storage [22]. Other factors also linked to the boys-girls disparity in overweight are cultural and behavioural factors. For example, in some traditional African cultures being overweight is seen as an indication of being healthy, happy, and not suffering from debilitating conditions like HIV/AIDS [26]. We propose that multiple sociocultural, physiological and behavioural factors increase the odds of overweight and obesity among adolescent girls in the South African setting, hence, to address these factors With regards to adolescents from high SES families within the context of our study sample, the overweight and obesity observed in higher proportion may be because of the tendency for increased consumption of processed and fast-food products among those who have the financial means. Our findings are similar to studies collated in a sub-Saharan African review [27] which showed that children from the highest SES are about 5 times at greater risk of overweight and obesity than those from the lowest SES household. Aounallah-Skhiri et al. [28] showed in a North African study that one of the major outcomes of urbanization and high income is the tendency for consumption of processed food including white bread, dairy products, high-fat content, and high-energy products. We hypothesize that adolescents from higher income bracket in this study are more likely to be able to "purchase" comfort including buying electronic gadgets like smartphones and video games which encourage a sedentary lifestyle and reduce being physically active [29]. Furthermore, they are more likely to have means of affording unhealthy food compared to adolescents from medium and low-income families. This hypothesis is supported by our results that indicated that adolescents with higher monthly allowance had higher overweight obesity prevalence. McLaren [29] and Pavela et al. [30] also indicated that the relationship between obesity and socioeconomic status is nuanced by several other factors including ethnicity, sex and geographic location. For example, while obesity is associated with high SES in low-income countries, the reverse is the case in high-income countries [29]. We found that mothers' employment status was significantly associated with the proportion of adolescents who were overweight or obese in each survey year in contrast to fathers' employment status. This is similar to previous findings in UK and USA [31,32], both judged as fully industrialized nations, where mothers' employment was associated with children gaining weight. The pathway for the increase in weight is that mothers' availability after school is seen to reduce sedentary behaviour among adolescents [32], while mothers' employment led to poorer eating habits and an increase in sedentary behaviour among children [31]. In the South African context, the impact of traditional gender roles is an important factor to consider. Women are expected to look after the house and ensure that children live a healthy life including preparing meals for the family. We hypothesize that employed mothers are likely to be unavailable and ensure that adolescents engage in activities and behaviours that mitigate overweight and obesity, which could be the reason for the association between mothers' employment status and adolescents' overweight and obesity. --- South Africa's trend in prevalence compared to Europe findings Contrary to our hypothesis, our findings showed that SA and Europe have almost equal proportion of adolescents with obesity and overweight, although the steeper slope in the gender plot indicates that the rate of progression of the pandemic may be greater in SA. South Africa is an upper -middle-income economy and is seen as one of Africa's most industrialized and civilized economy, hence we propose that the similarities in most economic indices between South Africa and European nations create a nearly similar sociocultural context including factors affecting adolescents overweight An important finding in this comparative study is the gender disparity in overweight and obesity burden. While South African girls are at greater risk of being overweight or obese compared to boys, the European report presents the exact opposite. These findings may be influenced by disparity and nuanced interaction of cultural emphasis and current social environment in both locations. Awareness of cultural influence and social pressure is important in understanding the South African finding. Most African culture link health and living well to being plump and fat [26,34]. Females are more desirable and attractive in the African setting when they are not slim and lean, especially in a context where being lean or slim is attributed erroneously to debilitating diseases like HIV and AIDS [22,26]. Being plump and fat is mostly associated with fertility and therefore, there is social pressure even in the early years of adolescence which could contribute significantly to the proportion of girls who are overweight or obese. In contrast to Africa, most developed countries' cultures like in Europe link beauty and attraction to being slim and lean [21,26]. There are indications that social media and peer effects have increased the pressure among girls in developed nations to ensure lean and slim body structure [35,36]. Some extreme weight loss behaviours associated with this social pressure among girls include irregular fasting and skipping of meals, self-induced vomiting, laxative, and diet pill use [14,37]. Furthermore, cultural factors may also explain the contrasting findings among boys. While the traditional African culture emphasizes fattening girls in preparation for marriage [22,26], boys are socialized to be active and to work hard in preparation to be the breadwinner of the family. Hence, most boys are influenced by this culture to be physically more active than their female counterparts, unlike in high-income countries like Europe where there is less social pressure on males to be the sole breadwinner, work harder or engage in more physically taxing work than their female counterparts. Findings show that in Europe, proportion of adolescents engaging in moderate-to-vigorous physical activity decreases from age 11 to 15 years between 2006 and 2018 [14,38]. Hence fewer boys engaging in physical activity at the age of 15 years would have played an important role in the increasing proportion of European boys being overweight and obese in contrast to the SA cohort. --- South Africa's overweight and obesity trend in prevalence, and the obesity transition model Our study showed an increase in the percentage of SA adolescents who were overweight or obese between 1998 and 2016. Although SA adolescent population is generally at risk of overweight and obesity, girls and adolescents from high SES families may be at greater risk. According to the obesity transition model [10], there are four stages in obesity transition. Specifically, and going by our findings it will appear that the SA adolescent population is going through the first stage of the model. At stage one, there is a higher prevalence of obesity in females compared to males, and higher in those from higher SES compared to lower SES [10]. Many countries in South Asia and sub-Saharan Africa are also at this stage [10], which appears to be plausible within the SA adolescent cohort given our findings in this study. According to Jaack et al. [10], the importance of using the obesity transition model to predict its prevalence is to enable stakeholders including researchers and policymakers to gain guidance in identifying the subpopulation at greater risk and the stage within specific communities in order to take proactive steps to prevent severe prevalence while considering all local contextual influences. The final aim of this study is to shine the spotlight on the at-risk subpopulation for overweight and obesity within the adolescent cohort, specifically girls and adolescents from high SES especially those living with employed mothers. Further study will hopefully elucidate the environmental and behavioural factors that put these subpopulations at risk of overweight and obesity, and possibly raise points of discussion among stakeholders on how to prevent further increase, among the adolescent population as the future workforce in South Africa. --- Study limitation A limitation of the current study is the imbalance in the proportion of adolescents' sample size based on wealth index grouping for 2012 survey year. Specifically, the higher SES participants were smaller in number compared to middle and lower SES groups. This could be because the adolescent cohort from the original survey was skewed towards the lower SES families, although all SES groupings were matched equally when other age groups, not relevant in this study, were included. Regardless of this discrepancy, our findings regarding SES link with adolescent overweight and obesity are consistent with other SES measures that didn't include data from 2012 sample. Furthermore, socioeconomic data for all 3 SA surveys were collected differently, yet they are all nationally representative sample. Finally, it is important to note that all European data used in the study came from the survey report, hence data analyses was not performed on HBSC data in this study. --- Conclusions This study is one of the first aimed at using both the obesity transition model and comparative approach to understand the trends in the prevalence of adolescent overweight and obesity in the South African context over a period of about two decades. Then to further identify the at-risk subpopulation in order to provide guidance to stakeholders in mitigating the pandemic. Our findings showed that the prevalence of adolescent overweight and obesity has increased with time until 2016 based on data collected in the national surveys. Furthermore, girls and adolescents from higher socio-economic families, especially those whose mothers are employed, may be at a greater risk for overweight and obesity in our local context. Further investigation on this cohort will highlight behaviours of the at-risk subpopulation that mediate the increased prevalence. This will assist researchers and policymakers in developing policies and strategies to mitigate and prevent further increases in overweight and obesity among the adolescent cohort. Further followup and regular adolescent-focused surveys are needed to track adolescents' health and behaviours. --- Abbreviations WHO: World Health Organisation; HIC: High-income country; LMIC: Low-and Middle-income country; SES: Socioeconomic Status; SA: South Africa; YRBS: National Youth Risk Behaviour Survey; SANHANES: SA National Health and Nutrition Examination Survey; SA DHS: SA Demographic and Health Surveys; HBSC: Health Behaviour in School-aged Children; EA: Enumeration Area; PSU: Primary Sampling Unit; BMI: Body Mass Index; BAZ: BMI-for age Z-score; PCA: Principal Components Analyses; ZAR: SA Rand; X 2Trend : Chi-square test for trend; OR: Odds Ratio; HIV: Human Immunodeficiency Virus; AIDS: Acquired Immune Deficiency Syndrome; UK: United Kingdom; USA: United States of America. --- --- Funding The study received financial support from the "Confronting obesity: Co-creating policies with youth " project that received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 774210. The funding body did not contribute to the design of the study and collection, analysis, and interpretation of data and in writing the manuscript. The study is rather the output on completion of Work Package 3 of the project in South Africa and Europe. --- --- --- --- Competing interests The authors declare that they have no competing interests. ---

Background: South Africa has several national surveys with body weight-related data, but they are not conducted regularly. Hence, data on longitudinal trends and the recent prevalence of adolescent obesity are not readily available for both national and international reporting and use. This study collectively analysed nationally representative surveys over nearly 2 decades to investigate trends in prevalence of adolescent obesity in South Africa. Furthermore, it compared these data with similar continental report for 45 countries across Europe and North America including United Kingdom, Norway, Netherland, Sweden, Azerbaijan, etc. to identify at-risk sub-population for overweight and obesity among adolescents.The study included primary data of adolescents (15 -19 years) from South African national surveys (N = 27, 884; girls = 51.42%) conducted between 1998 and 2016. Adolescents' data extracted include measured weight, height, sex, parent employment status, monthly allowance received, and family socioeconomic-related variables. Data were statistically analysed and visualized using chi-square of trends, Wald statistics, odds ratio and trend plots, and compared to findings from European survey report (N = 71, 942; girls = 51.23%). South African adolescents' obesity and overweight data were categorized based on World Health Organization (WHO)'s growth chart and compared by sex to European cohort and by family socioeconomic status.: By 2016, 21.56% of South African adolescents were either obese or overweight, similar to the 21% prevalence reported in 2018 among European adolescents. Girls in South Africa showed higher trends for obesity and overweight compared to boys, different from Europe where, higher trends were reported among boys. South African Adolescents from upper socioeconomic families showed greater trends in prevalence of overweight and obesity than adolescents from medium and lower socioeconomic families. Mothers' employment status was significantly associated with adolescents' overweight and obesity.Our study shows that by 2016, the prevalence of adolescent obesity was high in South Africa -more than 1 in 5 adolescents -which is nearly similar to that in Europe, yet South African girls may be at a greater odd for overweight and obesity in contrast to Europe, as well as adolescents from high earning families. South African local

Introduction With the rapid development of AI technology, great threats to society, some of them even existential, can be expected to materialize and society has to be prepared to react to those changes . This cautionary sentiment is shared by other scholars, albeit with a focus on distinct detrimental consequences . Thus, fairness, accountability, and transparency are increasingly discussed in the scientific community as are other related concepts that aim for Artificial Intelligence development for the Common Good. The Common Good encapsulates the ambition that AI should benefit everyone or "be good for all", not only the ones who invest in or deploy the technology . While normatively building ethical AI systems in the interest of everybodywhich is often formulated as a central aim, for example by the EU -research has shown that there exist considerable obstacles and costs on the way to achievement. For instance, according to Elzayn and Fish market interests can be in opposition to developing fair systems as developing such systems requires additional resources and is cost-intensive. Aside from economic interests, political interests may not necessarily prioritize ethical AI systems either . Political actors often rather argue that AI development must be pushed forward for the sake of international competition, aiming at certain economical or scientific goals . Concerning the primary aim of politics to achieve and maintain power or the primary goal of economics to optimize businesses and grow, these actors are not necessarily aiming at a broad and society-including discussion about ethical AI which may oppose their primary goals . This paper starts from the assumption that neither academic debates nor political promises or declarations of intent alone will lead to shaping AI in the public interest. On the contrary, given the inherent logic of politics and economics in capitalist society, the claim for ethical AI will have to be asserted against the interests of economic and political elites. Regardless of the national economic and political contexts in which these interests operate, this can only succeed if a significant mass of citizens actively requests ethical principles in their roles as users and consumers of these technologies, or if they demand credible policies for ethical AI as voters. Hence, as a proposed safety measure against detrimental consequences of the emerging technology, one may demand the implementation of AI systems that a priori fulfill certain ethical criteria. Though, as argued above, the fulfillment of respective calls for ethical AI does not follow automatically but must be-at least to some extent-expressed by societal groups or the broader public in the form of articulated claims and requests to the political decision-making bodies. Thus, civil society has to issue convincing incentives for economic or political actors to invest in AI systems that benefit the Common Good . Consequently, there must be public awareness of social problems for them to lead to political actions. While we have some knowledge of how citizens react to the ethical design of AI , we only have little to no knowledge about the salience of ethical AI among the public in the first place. Hence, we ask, what issues are top of citizens' minds and, in particular, how prominent ethical issues feature among these. We subsequently ask whether the salience of ethical AI leads to different behavioral intentions. To shed light on these questions, we first reflect on the current state of approaches to achieving ethical AI, which is primarily connected to ethical guidelines. We highlight the shortcomings of the current approaches to ethical AI, emphasizing their conceptual weaknesses as well as the non-binding character of proposed ethical guidelines. We infer that public pressure must be exerted to achieve ethical AI for the Common Good. As the issues of emerging technologies like AI are mostly raised by the media, we then discuss the current literature on published opinion on AI. We especially reflect on the topical structure of the media reporting, i.e. the prevalence of ethical topics, as well as the presence of different actors. Afterward, we review studies about public opinion on AI and highlight that no study so far has researched citizens' salience of ethical AI without confronting respondents with ethical problems of AI beforehand. In the empirical part of the paper, we first explore the extent to which German citizens are concerned with ethical issues of AI and compare that to the salience of other phenomena and consequences of AI, for instance, technical functionalities or specific applications of AI. Second, we show which factors are linked to a higher likelihood of being concerned with ethical issues. Thereby, we especially look for differences in societal groups and explore whether groups of lower social status-those groups, that are arguably most endangered by contemporary AI technologies-are aware of those issues. Or, in other words, whether salience of ethical issues with AI is a discourse, which primarily bears fruit with people of higher societal status. In a third step, we investigate whether the issue salience of ethical AI has an impact on behavioral intentions towards AI, namely the intention to avoid AI and the intention to engage in discussions about AI. We point out theoretical and practical implications in light of the low salience of ethical AI with the German public and the differences among societal groups in their respective salience levels. We further discuss the potential of the salience of ethical AI for the articulation of ethical demands to those, who are in charge of the development and deployment of AI systems. In this paper, we show that ethical AI discourse needs to be taken into all levels of society. --- Pathways to Ethical AI --- Ethical AI Guidelines Against the backdrop of ethical problems in AI development, some approaches were developed to tackle those issues that emerge with further development and implementation of AI in society. Jobin, Ienca, and Vayena collected and analyzed corresponding ethical AI guidelines from around the globe. Interestingly, many ethical guidelines were proposed by private companies or political institutions ), but also by academia and research institutions such as the Association for Computing Machinery . However, many of these guidelines do not address the broad public or civil society and instead target specific stakeholder groups. Thus, they focus only on the good of a subset of society or speak to those who are in direct contact with the technology. Guidelines that address the Common Good rarely occur. AI for the Common Good can be defined as an approach to developing AI systems that take many different social groups into account and not only those, who "directly pay for the development or use of this AI" . Consequently, ethical AI guidelines can not necessarily be equated with AI that benefits the whole society or that may recognize and include multiple societal perspectives at all. Looking at the sources and target groups of the ethical guidelines summarized by Jobin et al. , only a small subset aims for AI for the Common Good or the broad public, respectively. However, that does not imply that the other ethical guidelines exclude these aims, but they simply do not mention or prioritize them as the main goal to satisfy respective stakeholders. But what about the effectiveness of ethical guidelines? Given the different aims of stakeholder groups, it is debatable to what extent these actors want a broad discussion about ethical AI or only treat guidelines primarily for communicative purposes with other stakeholders. Out of their logic of achieving or maintaining power or increasing monetary profit , one can argue that most political or economic actors are not aiming for a broad and society-including discussion about ethical AI, as this can be opposed to their primary aims. Hagendorff analyzed 22 ethical guidelines from companies as well as research institutes and political institutions and attested the current state of AI ethics a bad condition: "Currently, AI ethics is failing in many cases. Ethics lacks a reinforcement mechanism. Deviations from the various codes of ethics have no consequences." . He argues that current ethical guidelines mainly serve as a "marketing strategy" and that ethics are seen as an add-on rather than a crucial step integrated into AI development. This is especially true in economic terms that oftentimes are in contrast to-and in the end override-commitment to ethical principles in AI design leading to a violation of ethical principles. Further, he highlights the non-binding character of these guidelines, which have the aim to avoid state regulation. In addition, focusing on the political approaches of the US, the EU, and the UK for AI for a "Good Society", Cath, Wachter, Mittelstadt, Taddeo, and Floridi report that, although issues are adequately addressed, the reports fall short in providing clear political perspectives on how to tackle these issues. Thus, also in the context of political guidelines, the effectiveness of those guidelines is questionable. Turning towards the practical side of developers' adherence to ethical guidelines, McNamara, Smith, and Murphy-Hill surveyed software engineers and report that reading ethical guidelines did not change their reported working habits. Thus, the prevalent top-down approach to ethical guidelines appears to have only a limited impact on the ethical AI development. --- The Role of the Public in Implementing Ethical AI Accordingly, the expression of AI ethics guidelines is not a sufficient requirement to achieve the goal of an AI for the Common Good. From a civil society perspective, therefore, the question is how to commit relevant stakeholders to the goals of an AI for the Common Good. How can 'bottom-up' political pressure be created that encourages those responsible to commit themselves and each other to the observance of corresponding guidelines and to stick to their commitments? Here, especially in pluralistically organized democratic societies, the focus must inevitably turn to the public sphere as the place where the electorate informs itself about political issues and forms an opinion. Irrespective of which specific conception of democracy and its public sphere one follows-for example, deliberative, participatory, and representative conceptions of democracy -the public and the media have important roles to play in the processes of political negotiation and legitimization . From a normative perspective, the media is often expected to serve as a watchdog , while the public-represented in part by advocacy groups-voices support for or disapproves of the implementation of technology. This results in the widespread continuous monitoring and analysis of public opinion that assesses citizens' attitudes and understanding of technology . Media coverage is especially relevant in the case of emerging technologies, since most people do not have personal experience with such technology or are not even aware of it, respectively. Thus, media present those topics to a broad public and have the potential to shape public opinion in setting a frame of reference for what and how to think about technology, or, at least, what relevant and powerful actors think about AI . With that, the audience learns about potential opportunities and risks of a technology as well as future pathways for dealing with the emerging sociotechnical systems. Thus, citizens' opinion toward emerging technology is affected by the way the media reports on it. Accordingly, it is important to understand how the media report on AI, especially concerning the topical structure and the presence of different social groups and actors . Taken together, public attention concerning ethical issues of AI thus may result in substantial pressure on the actual implementation of ethical AI, as affected groups and concerned citizens, reject sociotechnical AI systems altogether or protest their implementation . For instance, in the UK, students protested against an algorithmic decision-making system that autonomously graded exams, which lead to a withdrawal of the respective system . In France, a public debate about a university admission system also ended in abolition of the system . Concerning the use of an algorithm to allocate Covid-19 vaccination in Stanford hospital, medical staff heavily protested against the decisions of the algorithm, as front workers were not given priority . Additionally, journalists and activists also raised concerns about the implementation of unethical AI systems, for example in Sweden in the use case of a social benefit application and in Germany in the use case of face recognition in public places . Additional studies suggest that the perceptions of unfair treatment by ADM systems can lead to the rejection of such technology and of those who apply it, respectively ; furthermore, Lünich and Kieslich noted that a lack of trust in ADM systems leads to them being perceived as illegitimate. However, Algorithm Watch estimates the awareness of ethical problems with AI as not being overly high: "Protests in the UK and elsewhere, together with high-profile scandals based on ADM systems, have certainly raised awareness of both the risks and opportunities of automating society. But while on the rise, this awareness is still in its early stages in many countries." When analyzing the given examples, two reasons for reluctance towards AI systems are apparent. First, reactance towards AI is tied to a specific use case and articulated by those who are at risk of being treated negatively . These stakeholder groups organized themselves and fueled widespread protests against the use of the specific AI application. Second, media reporting led to public attention about ethical issues with some-again specific-AI systems. Media reporting may not lead to immediate protest behavior in the public, yet, it can damage the public image of developers or suppliers of AI technology. Hence, they may proactively work on ways to resolve ethical issues before widespread protests can emerge. Thus, analyzing public perceptions of ethical issues of AI can indicate how strong the public sphere is concerned with AI implementation and what public demands and objections may follow from social debates around the ethics of AI. In this study, we do not focus on a specific application but on the general attitudes towards AI and its ethics. --- Published and Public Opinion on AI Focusing on the inclusion of public perceptions in the debate, we first need to address the extent of AI's ethical issues that are discussed in the sphere of the media. We then turn to the analysis of public opinion concerning AI and AI ethics to document whether issues from media discourse are also reflected within public opinion. --- Published Opinion on AI In recent years, a considerable amount of research has focused on the analysis of news reporting on AI . According to the studies of Fast and Horvitz , Sun et al. and Chuan et al. for the US, Fischer and Puschmann for the German and Vergeer for the Dutch press, media reporting about AI sharply increased in recent years, leading to a steady presence of AI in news reporting. Thereby, several studies pointed out that media reporting about AI was more frequent by right-leaning, respectively conservative outlets . Further, Vergeer found for the Dutch press that national newspapers and tabloid newspapers report more on AI than the regional press. Focusing on the topical structure of news reporting about AI, several studies across different countries found that economic topics dominate the content of the news . Accordingly, it is economic actors that dominate the media coverage of AI across the globe; oftentimes with a special focus on so-called 'tech giants' . On the other hand, civil society actors or NGOs are rarely featured in news reporting about AI . For example, in German and UK media reporting only four percent of the articles analyzed cited actors from civil society. Moreover, several studies indicated a positive or ambivalent sentiment regarding AI in news reporting , while none of the studies reported that AI is portrayed as particularly negative. Positive sentiment is often connected to an optimistic turn towards AI or based on a positive wording about AI . Additionally, two studies explicitly focused on the occurrence of opportunities and risks in media reporting. Fischer and Puschmann report for the German context that the number of articles analyzed that outlined opportunities of AI was more than double as high as the number of articles that reported on risks of AI. Opportunities were mostly associated with economic progress and efficiency and risks, in contrast, with a lack of AI competence within the general population . For the US media coverage, Chuan et al. report that there is an emphasis on opportunities compared to risks about AI as well. Looking at news coverage about ethical issues of AI, Chuan et al. reported that the topic of AI ethics received increased attention by the US-American press from 2018 on. However, in comparison with other perspectives on AI, it is still a niche topic. Ouchchy et al. specifically analyzed news articles about ethical issues of AI using an English-language sample of various media outlets. They found that prejudices reflected in data, privacy, and transparency were the most common issues in the news articles. Furthermore, they report that when specific applications were discussed in terms of AI ethics, autonomous driving was the most mentioned application followed by the use of AI in the military and healthcare domain. Overall, the discussion about AI ethics was neither particularly critical nor enthusiastic. For the German media landscape, Fischer and Puschmann report that, if risks of AI were mentioned in the news, especially accountability and transparency issues were featured. Summing up the literature on news coverage of AI, some similarities emerge on a global level. The media discourse is led by economic and political actors, and topics that highlight the positive impact of AI. Ethics is a niche topic in media reporting about AI. These findings also apply to Germany, where the present study was conducted. The empirical results on published opinion on AI may be partly explained by the dominance of conservative and business-friendly publishers of news stories about AI among the press. This explains the observation that especially those actors are most present in the mass media within the public sphere who utilize ethical guidelines as a marketing device or have a vested interest in an unregulated approach to AI, while actors who actively strive for implementation of ethical AI for the Common Good are fairly underrepresented. In absence of personal experience with AI, from the perspective of communication science, the underlying assumption is "that the media crucially influence audience attitudes towards emerging technologies" . For the salience of AI issues among the public this may mean the following: On the one hand, news media weigh some topics over others and the public gets to know a rather one-sided picture of the technology with a strong emphasis on economic opportunities. On the other hand, when there is no media interest in reporting on ethical AI, then salience of AI-related ethical issues among the public seems to be highly unlikely, too. --- Public Opinion on AI Following increasing academic interest in media reporting on AI, several studies have recently explored public opinion on AI. Research on public opinion on AI found that, while many people have heard about the term, usually, only some of the surveyed people suggest to have an idea of what AI is and what AI entails, or claim to be somewhat knowledgeable about AI, respectively . Thereby, AI is mostly associated with technical terms, foremost robots or computers . All in all, when surveyed, current AI technology is perceived rather positively in various countries . In addition many studies conducted around the globe, highlight sociodemographic differences in the evaluation of AI: Male, younger, and well-educated respondents, people with higher income, and people, who consider themselves to possess much knowledge about AI are more likely to expect that opportunities materialize with the advent of AI . Moreover, Kelley et al. report country differences in the strength of positive sentiment towards AI, whereby residents of non-Western countries are more enthusiastic about AI than residents of Western countries . Some studies investigated public opinion towards selected ethical aspects of AI using closed-ended questions. Utilizing a representative German sample, Kieslich, Starke, Dosenovic, Keller, and Marcinkowski asked for risk perceptions of some particular ethical challenges that AI might pose in terms of injustice and discrimination. On the one hand, the authors report that uncertain accountability, as well as loss of control, were perceived as major ethical challenges of AI implementation by the public. On the other hand, potential systematic discrimination of social groups or injustice via AI were only articulated by a subset of the German population when inquired. The results also suggest that a significant portion of respondents did not answer or indicated that they did not know whether the latter issues pose a social risk in contrast to 5%-9% missing answers for other ethical issues. This indicates that fairness and discrimination issues may not be salient or even conceivable for a fair share of the German population. Another study by Kieslich et al. researched preferences for the ethical design of an AI system in the use case of tax-fraud detection. They report that accountability was perceived as the most important ethical criterion among the respondents, while machine autonomy was perceived as the least important. However, the researchers also report that the different ethical AI principles were perceived as roughly equally important, even though the respondents show different preference patterns among each other. Nearly one-quarter of the respondents cared only marginally about the ethical design of AI systems, while 32% were highly concerned about AI ethics. Moreover, the results suggest that those that were not concerned with ethical AI were predominantly older, less educated, and less interested in AI than respondents that were concerned about the ethical design of AI systems. In summary, the German public expresses some concerns about AI ethics, but the ethical concerns vary between different segments of the population ). Researching the perceived importance of ethical requirements proposed by the EU, Choung, David, and Ross showed for a US sample that, again, accountability was perceived as most important. As in the study by Kieslich et al. , they also report that all ethical requirements are evaluated equally important. Ikkatai, Hartwig, Takanashi, and Yokoyama surveyed Japanese citizens to evaluate public attitudes towards AI ethics. Testing for four different scenarios, they show that public attitudes toward the importance of ethical AI design are context-dependent, with AI use in the military being the most ethically problematic one. AI use in a military context was also the scenario most citizens disagreed with. Ikkatai et al. also showed that age was significantly correlated with attitudes towards AI ethics in all scenarios; moreover, gender, interest, and AI literacy were significantly correlated to attitudes toward AI ethics in some scenarios. All in all, when it comes to public opinion concerning AI ethics, there is only limited evidence towards general salience of ethical issues connected with AI. On the one hand, the research at hand often addresses these issues as a rather broad category within the topic of AI and it is not clear what specific ideas citizens have when it comes to the ethics of AI. On the other hand, research that directly gathers opinions on specific ethical aspects often does not capture the general salience of ethical issues of AI with citizens. In asking directly about opinions of ethical aspects of AI, the question itself makes ethical aspects salient; also to those, who would otherwise not indicate having thoughts about ethical AI, to begin with. Therefore, we do not know much about the actual salience of ethical AI and its consequences on intended behaviors among citizens. Thus, in this study, we bridge the literature strands of broad associations with AI and concrete opinions on ethical AI in investigating the general salience of ethical AI with the public. --- Hypotheses and Research Questions In the present study, we build on the existing literature and dig deeper into public opinion on AI with a focus on ethical AI. Our study was conducted in Germany, which-politically-follows the Western, respectively European approach to AI. That is, the German government aims for broad funding of economical and scientific progress in AI while acknowledging ethical issues . Thus, many AI projects were funded by the German government in recent years. However, as a small inquiry by the Left Party in the German Bundestag revealed, that only a few AI systems were checked with an AI risk analysis before their implementation . Regarding the state of AI usage in the economic sector in Germany, 5.8 percent of all companies used AI methods in 2019 , while, according to a study conducted in 2021, 30 percent of German companies indicated that they are thinking about implementing AI . Hence, AI plays a continually growing role-at least in politics as well as in the economy. While former public opinion studies utilized open-answered questions with a focus on knowledge about AI, we are interested in the issues that are present in the minds of citizens-if any. Unlike other studies, we do not present respondents with a given set of different future perspectives or ethical issues but gather information if citizens link AI to ethical aspects on their own. As AI is still a relatively new technology and news reporting of AI is-albeit increasing -still a topic that is not on the top of the news, we first explore how many respondents are preoccupied with AI topics at all. In a second step, we are interested in the different topics citizens associate with AI. As discussed earlier, especially economic and political actors, which are most prominent in media reporting, are not interested in a wide discussion about the social impact of AI. Consequently, we suppose that issue salience of AI ethics is not very prominent among German citizens. Another reason for that is that in the German context, unlike in the UK, no major scandal concerning AI emerged in the past. Yet, we do not know the variety and the emphasis of different AI issues on behalf of citizens. Thus, we set up RQ1. RQ1: What topics are present in the German population with regard to AI? Further, prior studies suggest that several individual characteristics may influence public opinion on AI such as interest in or knowledge of AI as well as sociodemographic aspects like gender, age, educational level or socioeconomic status . Given the one-sided media reporting about AI and the focus on specific stakeholders by economic and political actors, we suppose that AI can be considered as a topic of the elite. Thus, we hypothesize: H1a: Salience of AI issues is higher among citizens with a higher level of education. H1b: Salience of AI issues is higher among citizens with a higher socioeconomic status. H1c: Salience of AI issues is higher among citizens with a higher general interest in AI. Additionally, we explore which socio-economic characteristics and interest in AI influence if someone is concerned with ethical AI in particular. In the context of ethical aspects of AI, the questions of which sociodemographic factors, as well as interest in AI, influence the issue salience of ethical AI, become particularly relevant, since AI implementation can have serious consequences for different societal groups. For example, it was found that AI systems treated women or persons with a low socioeconomic status worse than others. Additionally, Frey and Osborne assumed that the jobs of people with a lower educational degree are more at risk of being automated. Thus, we explore if those who are probably more endangered by the implementation of AI systems are aware of those ethical issues. Though, due to a lack of media reporting on ethical aspects and the dominance of economic topics, affected groups might not be concerned with ethical issues of AI. Hence, we ask: RQ2: Do educational level, socioeconomic status, and interest in AI have an effect on salience of ethical AI issues? Further, we investigate if different topics are associated in context with each other. We suppose that ethical issues are mostly connected to an anchor example. Thus, ethical issues emerge when discussed in light of a certain usage of AI. For example, Ouchchy et al. showed that ethical aspects of AI were repeatedly reported on in the context of specific applications like autonomous driving. Also, the cases were protest was articulated by the population or by civil society actors were tied to a specific example, e.g. education admission systems or vaccine allocation . As the literature does not suggest, which use cases are currently on the top of the head of citizens or especially suitable for a public debate about ethical issues of AI, we formulated H2 rather broad: H2: People who are concerned with ethical AI commonly connect AI to specific applications. Lastly, we investigate if the salience of ethical AI influences behavioral intentions. We have outlined that ethical AI has to be demanded by the public to be put into practice by companies or politicians. For companies, considering citizens' demands is necessary in the sense of economic profit. If citizens hesitate to use AI systems because of ethical considerations or raise protests against an AI tool, it would damage the profit companies make. Hence, public pressure can presumably force economic actors to follow ethical guidelines. Likewise, political actors can be influenced by public opinion as well. If citizens demand ethical AI systems, political actors could listen to those voices and set up binding guidelines for the development and implementation of AI systems. However, for that to happen, the salience of ethical AI needs to influence citizens' behavior in a way that leads them to reluctance toward AI. Given the use cases, where ethical flaws in AI systems lead to public outrage, we presume that the salience of ethical AI among the public affects AI avoidance intention, respectively the intention to engage in public discussions about AI. H3a: Salience of ethical issues of AI positively affects intended AI avoidance. H3b: Salience of ethical issues of AI positively affects the intention to engage in public discussions about AI. --- Method --- Data Collection & Sample The study is part of the long-term monitoring of public opinion on AI in Germany. In fifteen surveys, about 1,000 respondents were asked about interest, attitudes, and behavioral intentions regarding AI. The monthly surveys were conducted between May 2020 and April 2021 by the market research institute forsa GmbH as part of their Omninet omnibus panel. That is, the questions on AI were asked alongside other changing topics. The advantage of this procedure is to minimize self-selection, as respondents do not choose questionnaires by interest in specific subjects. This allows a more realistic and rather representative picture of the population. The Omninet panel consists of 75,000 panelists who are recruited via random sampling of phone numbers. The panel is representative of the German population with internet access aged 14 and older for the personal respondents characteristics age, gender, and regional place of residence. Our questions were only asked those panelists who are 18 years and older. All 15 samples included in this study were randomly selected from the Omninet panel. After data cleaning, the dataset contains 14,988 respondents. Of these, 51.6% are self-identifying female and the average age is 51 years. This corresponds very well to the distribution in the German population aged 18 and older with internet access . 60.2 percent of our sample have at least a high school degree, higher education being slightly overrepresented . --- Measurement Salience of AI-related issues. As for the dependent variable, we are interested in whether people are concerned with AI and, if so, what these issues are. Following agenda-setting research , we are looking for a measurement of the public agenda-but this time precisely on the subject of AI. Two conditions were important to us: First, respondents should be able to say that no issue concerns them at all. Second, we did not want to ask about problems by default, as was often done in agenda-setting research . Thus, we asked "If you think about recent times, which issues related to artificial intelligence have been of most concern to you personally?" Respondents were asked to provide up to three answers to this question in an open text box. After conducting the fourth survey, the codebook was developed in an iterative process. That is, the two coders and one of the authors randomly drew several samples of responses from the data set. They coded all the same data and discussed their results afterward. Each of the possible three responses was assigned to only one code. If two codes were deemed applicable, the coding that described AI in more detail or on which emphasis was placed was selected. For example, if an answer was "job loss due to automation", "job loss" was chosen as code as it depicts the consequence of automation. Analogously, "ethical issues due to autonomous driving" was coded as "ethics". After the first coding process, additional samples were drawn and previously formulated codes were used to categorize the new data until there was no new code assigned. This resulted in four major categories with a total of 49 subcodes. A new sample of 150 responses was then coded by the two coders for a reliability test. Krippendorf's alpha is rated as good at the value of α K = .82 [Bootstrap with 10,000 samples; CI 95%: .72; .90]. For the category ethical issues about AI, we adapted the principles outlined by Jobin et al. . We added the more specific ethical problems surveillance1 , manipulation as well as a general subcategory for ethics to the list of ethical criteria, as they emerged as subcategories during the coding process. The codebook with the occurrence of all subcodes in the sample can be found in Appendix A. Interest in AI. Interest in AI was measured with the following four items on a five-point Likert scale: "I follow AI processes with great curiosity.", "I am very interested in AI in general.", "I read articles about AI with great attention.", "I watch or listen to articles about AI with great interest." The index suggests high internal consistency . Behavioral Intentions. We were interested in how the salience of specific AI-related issues is associated with the intention to avoid AI or to engage in public discussions about AI, respectively. Both were measured as single items on a five-point Likert scale: "I will stay away from artificial intelligence wherever possible." and "I will express my opinion in public discussions about artificial intelligence." . --- Controls. In the results part, we utilized age , gender , educational level , and socioeconomic status as control variables. --- Results All in all, 6,221 respondents indicated that they were recently concerned with AI and, thus, gave at least one answer to the question at hand. 2,090 respondents gave at least two answers and 1,030 respondents gave three answers. In contrast, 7699 respondents were not concerned with AI, while 302 respondents stated that they do not know the term artificial intelligence, and further 766 preferred not to answer this question. First, we were interested in the specific AI issues the German public was concerned with . In general, specific application domains were the most frequent response. 4,122 named at least one application or application domain in the open-ended answers. In contrast to this, AI functionalities were only mentioned by 906 , AI issues by 1,274 , and AI ethics by 943 of the respondents. Thus, the German public is mostly concerned with specific use cases of AI rather than issues, functionalities, or ethical issues To dig deeper into the specific topics that the German public is concerned with, we also investigated the subcodes of the categories. Figure 1 shows the most common issues with AI and groups them according to their dimensions . For better readability of the figure, we only depicted those issues that at least 100 respondents mentioned. The result shows that the most salient issue was one specific application domain, namely autonomous driving or mobility. It was mentioned by nearly one-third of the respondents who were concerned with AI at all. Also, other application domains such as medicine or smart home applications were frequently mentioned. Concerning AI issues, the most frequent mentions were connected to workplace issues such as job opportunities or job losses due to AI. Regarding Figure 1: Concerns about AI among the German Population technical issues, AI was frequently connected to robotics, which is in line with the study of Selwyn and Gallo Cordoba . Only a small portion of respondents name ethical aspects as the most salient issues. Nevertheless, we were interested in the gradations of ethical AI issues. Thus, we also counted the mentions of all ethical issue subcodes. Figure 2 shows the result. Ethical issues are most connected with indications about control, i.e. if AI can or needs to be controlled. Another, more specific ethical concern is the possibility of surveillance through AI technology, which was the second most ethical issue raised by the respondents. Remarkably, terms that are connected with the FAccT dimensions are the ethical dimensions that are least mentioned by the respondents. Thus, there is practically almost no salience of fairness, accountability, and transparency issues among the German public. In the next step, we research the hypothesized effects of educational level, socioeconomic status, and interest in AI on the salience of AI in general . We included the age and gender of respondents as control variables in the model. By calculating a logistic regression model on the issue salience of AI, we find that higher educational level, being younger, and being more interested in AI in general, seeking information about AI, respectively, predict whether AI is on the personal agenda, as indicated by expressing at least one AI issue . No significant effect could be found for socioeconomic status. The model explains 37.2 percent of variance with interest in AI being the strongest predictor. Thus, H1a and H1c are supported, while H1b needs to be rejected. In the next step, we investigated the effects of educational level, socioeconomic status, and interest in AI on the salience of ethical issues. Again, we conducted a logistic regression on the dependent variable salience of ethical AI issues. Ever Heard of Ethical AI? Thereby, we only calculated the model for those respondents who were concerned about at least one AI issue. Thus, we explore if specific respondents' characteristics lead to a higher salience of ethical issues. We controlled the effects for respondents' age and gender . The model reaches significance, but the independent variables only explain 0.8% of the variance of the dependent variable. Even though we find significant effects for age, socioeconomic status, educational level, and interest in AI, these effects are not substantial. The low salience of ethical principles does not allow us to look for structural differences in the population. This could change should the topic become more prominent. Then it could be precisely these characteristics that account for greater attention. At this point, however, such an assumption does not hold up. Regarding RQ2, we found no notable effect of educational level, socioeconomic status, and interest in AI on the salience of ethical AI. To answer H2, we calculated the co-occurrences of the different categories of AI issue salience. Co-occurrences can only be detected if participants mentioned at least two issues they were concerned with AI. For each group we summed up the number of respondents who mentioned issues of both respective categories. AI ethics were mentioned 62 times together with AI functionalities, 218 times together with AI applications and 202 times together with other AI issues. However, 535 respondents mentioned AI ethics without any other category. In comparison, AI applications were oftentimes named without any other category. AI functionalities and AI issues , on the other hand, were mentioned to an equal amount with other categories. Though AI ethics are most commonly mentioned without any other categories, it is mostly connected to specific AI applications or other AI issues. Thereby, the huge number of single mentions is connected to the fact that a majority of the respondents only were concerned with AI in one issue and did not indicate up to three issues. Given that salience of ethical AI had the highest co-occurrence with AI applications, H2 can be supported. Lastly, we test if the different AI-related issues people were concerned with have an impact on 1) intended AI avoidance and 2) intention to engage in public discussions about AI . For that, we calculated two linear regression models with the dependent variables 1) intended AI avoidance and 2) intention to publicly speak about AI with the independent variable salience of ethical AI issues. We included the salience of AI applications, AI functionalities, and other AI issues as control variables. We further control the effects with the variables used in the models before-age, gender, educational level, socioeconomic status, and AI interest. We calculated the model for the respondents, who were concerned with AI omitting those, who are disengaged with AI. Thus, we can investigate in detail which effects different categories of issue salience of AI have. Tables 3 and4 show the results of the linear regression model. Concerning the OLS regression on avoidance intention, the independent variables explain 11.4% of the variance for the dependent variable avoidance intention. Among the issue salience variables, ethical salience had the strongest impact on avoidance intention. Avoidance of AI rather occurs if people have ethical issues of AI top of their mind, supporting H3a. Conversely, people who are concerned with specific applications show fewer intentions to avoid AI. The salience of AI functionalities and other AI issues do not have a significant effect on avoidance behavior. Concerning the control variables, AI interest is the strongest negative predictor of the dependent variable. We also find significant effects for age, socioeconomic status, and gender. Among people who are concerned with AI, those who are more interested in AI show fewer intentions to avoid AI. Additionally, older people, people with a socioeconomic status below average, and women report higher avoidance tendencies. Regarding the OLS regression on the intention to engage in public discussion about AI, the model explains 15.6% of the variance of the dependent variable. Hereby, only the salience of ethical aspects of AI has a significant, yet small, effect on the intention to speak about AI. Those who are concerned with ethical topics are more willing to talk about AI. Thus, H3b can be accepted. Respondents who have other AI topics in mind show no significant tendency to talk more or less about AI. Again, we find the strongest effect for interest in AI, followed by gender and age as significant predictors. Higher interest in AI leads to a higher intention to engage in public discussions about AI. This tendency is stronger for men and younger people. --- Discussion Our study sheds light on a rather understudied perspective on ethical AI-public opinion. We investigated this perspective by focusing on the salience of AI issues. For that, we collected data from 14,988 German respondents. --- Low Salience of Ethical Issues and its Implication for Implementing AI Ethics Our results show that AI does not play a major role in the minds of German citizens with about three-fifths not been concerned with AI recently. However, when thinking about AI it is foremost thinking about specific applications, especially autonomous driving. Ethical issues only play a subordinate role among the German public, and if so, the major share does not think about issues of fairness, accountability, and transparency to which many researchers in the field of Fair Machine Learning are devoted. These results indicate that the scientific-and partially political-discourse does not reach the broad public. Ethical issues of AI are mainly a topic, which is discussed by researchers. However, leaving out respectively not reaching the public may interfere with the normative goal of ethical development for the Common Good. Without a broad public discourse about these topics, it is very unlikely that public demands for the development of ethical AI will emerge . It is not plausible that political institutions or companies invest massive amounts of financial and personnel resources in the development of ethical AI that benefits the whole society and treats people equally if the public does not demand it. At this point, it seems more likely that political institutions set up guidelines but make compromises to hold their own in global competition . Likewise, as long as companies make vast profits and do not provoke massive scandals, economic actors presumably will develop AI systems in a way that satisfies their stakeholders . Importantly, only addressing stakeholders leaves out a significant portion of society. --- How Likely is Critical Engagement with AI? This leads to a reflection on the future of society-and who is affected by the ethical threats of AI. Like the study of Frey and Osborne show for the context of workplace development, those who have lower income and educational levels are most susceptible to negative consequences of automation. Alarmingly, we observe that it is exactly those people who do not care much about AI. Educational level is-besides interest in AI-a strong predictor of being concerned with AI. Thus, those who will be negatively affected the most are not concerned with it. However, we did not find effects for gender and higher socioeconomic status for being concerned with AI. As ethical issues with AI often relate to women being discriminated against or lower incomes resulting in deprivation by algorithms , it is also noteworthy that these groups of people are not particularly concerned with AI issues, which would play into the hands of economic and political actors as previously argued. Our data show that greater attention to ethical issues regarding AI is linked to both the intention to avoid AI and the intention to participate more intensively in debates. As a result, those who pay attention to ethical issues regarding AI-which in most cases are potential breaches of ethical standards-do not contribute to greater corporate profits and may even steer a discourse on AI toward problematic aspects. If people are more concerned with the ethical challenges AI systems pose for society, a more critical way of citizen engagement with AI will seem possible. Instead, the current discourse tends to result in a strongly product-and application-oriented way of thinking about AI. Thus, the public AI agenda reflects quite well the commerce-oriented tone in the media discourse about AI. Having the high amount of mentioned AI applications in contrast to ethical issues in mind, we describe the current state of critical engagement with AI in Germany as low. Yet, the results suggest that, if ethical issues are made prevalent, citizens are more reluctant to integrate AI in their lives or to publicly engage with AI. In summary, ethical issues of AI are not very prevalent among the German public. This may be due to the lack of a critical use case, which fueled public outcry against an AI system in Germany. Different from countries such as the UK or France , there has not been a major scandal in Germany concerning AI yet. Additionally, we attribute these results to the low level of media coverage on ethical aspects of AI and the dominance of actors from the economic or political sector in news reporting. Although there are some examples where specific AI applications were blocked due to public awareness , we are far away from a broad discourse about the ethical challenges of AI systems. Yet, we also detected in our data, that ethical issues may have an impact on citizens' behavior towards AI. We also found that ethical issues may have an impact on citizens' behavior towards AI when such issues are prevalent in the minds of citizens. Future studies should monitor the ongoing public debate about ethical AI and investigate, whether certain key events or a possible change towards more critical news reporting will affect the broad public opinion on AI. At least, for AI for the Common Good, it is relevant that the public is involved in an ethical debate about AI. --- Limitations Our study has several limitations that need to be acknowledged. First of all, we conducted our study in Germany. Citizens of other countries might have different concerns with AI, for instance, Kelley et al. showed that public opinion on AI differed largely between citizens of different countries. In the case of the focus on public concerns about AI ethics, it would be especially interesting to compare our results with those of UK or French citizens, where public debates about ethical issues of AI have taken place lately . It would be interesting to investigate, whether a scandal might have an influence on public concerns with ethical issues of AI in the long term. Accordingly, we encourage researchers to replicate our survey in other countries. We also point out some potential issues with our measurement of AI issues. We deliberately chose a broad approach with an open-ended question to grasp public concerns with AI. However, we had to reduce the complexity of our coding process. Thus, if two codes were applicable for one answer, we had to choose one code, which was in focus. However, some information might have gotten lost during this process. Moreover, not being concerned with AI, and especially ethical issues, does not necessarily have to mean that the public is not aware of these issues. It means, however, that these issues are not of immediate concern and arguably are not on the personal agenda. As we rely on agenda-setting theory, the measurement chosen most adequately grasps our concept. Finally, our measurement is not fine-grained enough to go into detail about the concerns the public has. For example, it might very well be that respondents, who answered "autonomous driving" in the open-ended question think of possible ethical problems . But if the answer was only "autonomous driving" it was coded as such . Though, if respondents answered "ethical problems with autonomous driving", it was coded as an "ethics" issue. As we aimed for an overview analyzing the public agenda of AI issues, we believe our measurement is satisfactory. Further research should nevertheless build on our measurement and extend it in a way that the tonality of several issues is measured as well. --- Implications We empirically showed that ethical AI plays a minor role in the minds of German citizens. Given the media reporting about AI in Western countries, which highlights economical aspects and oftentimes neglects ethical issues, this finding is not surprising. Though, media outlets oftentimes are self-committed to reporting in a multi-faceted way about issues that are of importance to citizens. Although Ouchchy et al. and Chuan et al. found that ethical issues are on the rise, news reporting on AI does not seem to be very balanced yet. Including ethical aspects in news reporting could lead to a higher salience of ethical issues on behalf of media consumers. This could, for example, be achieved by picking up ethical topics or including more perspectives of researchers or NGOs, who are actively engaging with the implementation of ethical AI. Currently, we see the dominance of economic and political goals mirrored in the public opinion towards AI in Germany. Vividly, German citizens are especially concerned with AI in the contexts of autonomous driving or advances in medicine, which are applications that the industry is vigorously pushing forward. Though, increasing attention on the media side is not the only way to fuel the salience of ethical AI issues. Long and Magerko define AI competencies that are needed for citizens to deal with AI reasonably. Beneath technical competencies, they argue that awareness of ethical AI should be taught to citizens. For that, literacy projects are needed that expand the current state of AI literacy programs that mostly focus on technical aspects of AI. However, for a curriculum in the sense of AI for the Common Good, societal and ethical issues should be included. Several programs already offer a wider scope on AI, for example, Elements of AI . In the long term, it also could be useful to integrate the topic of AI ethics into school or university curricula. Thus, students who study computer science would also learn about the ethical risks of the technology and implement this knowledge in their further development of AI systems. Additionally, strengthening civil society actors could be a promising way to fuel the public debate about ethical AI. Civil society actors could actively address journalists, so that media coverage includes positions of the civil society regularly. It was also shown that some civil society groups exerted public pressure in some use cases . However, the inclusion of civil society is rather an exception. Strengthening civil society can be achieved by investing in human and material resources. This is all the more important because civil society actors or academics advocating ethical AI contrast with economic or political actors who have many times more resources at their disposal. For a more balanced public discourse, there needs to be at least a somewhat more even balance between the different stakeholder groups. We conclude that the salience of ethical AI issues is very low. For AI to benefit the Common Good, the public debate about this topic needs to be strengthened and also carried out to those who are affected by negative effects of unethical AI. We find that there indeed is potential that citizens can critically engage with AI; however, it needs to be activated by media, civil society actors, researchers, and other stakeholders who aim for AI for the Common Good.

Building and implementing ethical AI systems that benefit the whole society is cost-intensive and a multi-faceted task fraught with potential problems. While computer science focuses mostly on the technical questions to mitigate social issues, social science addresses citizens' perceptions to elucidate social and political demands that influence the societal implementation of AI systems. Thus, in this study, we explore the salience of AI issues in the public with an emphasis on ethical criteria to investigate whether it is likely that ethical AI is actively requested by the population. Between May 2020 and April 2021, we conducted 15 surveys asking the German population about the most important AI-related issues (total of N=14,988 respondents). Our results show that the majority of respondents were not concerned with AI at all. However, it can be seen that general interest in AI and a higher educational level are predictive of some engagement with AI. Among those, who reported having thought about AI, specific applications (e.g., autonomous driving) were by far the most mentioned topics. Ethical issues are voiced only by a small subset of citizens with fairness, accountability, and transparency being the least mentioned ones. These have been identified in several ethical guidelines (including the EU Commission's proposal) as key elements for the development of ethical AI. The salience of ethical issues affects the behavioral intentions of citizens in the way that they 1) tend to avoid AI technology and 2) engage in public discussions about AI. We conclude that the low level of ethical implications may pose a serious problem for the actual implementation of ethical AI for the Common Good and emphasize that those who are presumably most affected by ethical issues of AI are especially unaware of ethical risks. Yet, once ethical AI is top of the mind, there is some potential for activism.

Introduction Software process and productivity improvement encompasses the activities which promise to increase the quality of a software product [5]. Predictably, these activities need to align process, tools and technologies with human factors and social considerations [11,1]. Software development is a form of social activity [8]. Therefore, it is commonly conducted by teams consisting of individuals identified by characteristics of "individualism, rationality, and mutual interdependence" [12]. In this particular viewpoint, one can argue that several factors affecting the software development process should arise from the complexity of individuals' interactions and social communication costs. Therefore, the investigation of social factors and corresponding interest in social side of software development has become a part of software engineering body of research [10]. Emerged in the 1930s, game theory is a field of mathematics, which has been frequently applied to social sciences for analyzing many different situations, e.g. variability of individual behaviors, and formation of coalition structures among the individuals and human networks [14]. Furthermore, the study of game theory has flourished over the last several decades and attracted many researchers. As a result, it has been applied to several diverse fields [15] including biology, linguistics, psychology, philosophy, and later in computer science [20]. There are two branches of game theory both are useful for investigating social interactions among software teams and individuals. Non-cooperative game theory deals with situations where a limited number of players interact and their choices will affect the overall outcome, essentially participants are inclined to benefit more from these situations individually. Cooperative game theory deals with arrangement of participants in different stable combinations [4] for value maximization, which should be suitable for software business practices. Because, it explores the cooperative solution concepts, to achieve a collective outcome, cooperative game theory should be useful for assessment of several software development activities and products. For example, it can support the collaborative features of software development where coalitions are formed for profit and performance. Software development organizations prefer to have their participants work collaboratively. Therefore, individuals need to be socially integrated so as to develop team cohesion. In a game theoretic sense, they need to correlate their activities to form teams of coalitions for better productivity. Recently, several studies have advanced our understanding of the possible use of game theory in software engineering research [2,21]. The evidence suggests that software develop-ment organizations rely heavily on several strategic nature of software management activities [7,9] , and therefore we suggest that software management teams should benefit from what game theory offers. --- Background The ultimate goal of software process engineering is to provide a roadmap for the production of high quality software products that meets the needs of its stakeholders within a balanced schedule and budget [24]. It concentrates on the creation and maintenance of tasks and activity structures rather than the output or the end product. A typical software process aims to solve the potential and future problems of software development with respect to planning and budgeting constraints. In the context of our approach, a process is considered as the coordination of structural social activities coupled and constrained with a set of people roles and skills for producing software artifacts in a predefined productivity level . In order to provide a comprehensive background to our approach, we first start by defining the concept of mechanism design and its potential impact on software practices. Secondly, we introduce the concept of software ecosystems to define a social viewpoint for software development organizations. Finally, we introduce a model to measure the social aspects that are affecting software productivity improvement. --- The concept of mechanism design A subfield of game theory, mechanism design, specifically deals with social decisions and their effects on outcomes. In this framework, a designer or a manager investigates how one designs the social structure of an organization so that the individual incentives of participants can be transformed into the organizational wide desired goals. In other words, mechanism design is an approach for thinking about the social structure of an organization. An organization can be modeled by depiction of social patterns and available actions for its participants. Further, we can make predictions about several organizational parameters with expected outcomes using game theoretical concepts. Based on selectable parameters for desired goals or given objectives, we define a mechanism as the model of an organization. It is built on several inputs from individuals in order to produce several types of outputs. The goal is to dynamically portray an organization by designing the structure of its teams for its defined objectives and hence to motivate individuals to act in the service of an organization. We aim to establish a structural improvement inside an organization where it is based on the fact that the quality of organizational production relies on the structure of the organization [6]. In software development practices, collaboration and communication is observed in several stages of the development life cycle. Consequently, information gathering and its distribution have become somehow decentralized. Although this decentralizion may facilitate an ability for self organization and improved evolvability, it could also increase the communication costs and information based complexity. The theory of mechanism design and its modeling implementation on software development organizations can provide a way to explore the effects of social structures on team composition, where we can use this information for better team and organizational structuring. An economic mechanism involves designing the rules for the economic activities that govern the social interactions of the participants. These rules, for example, can be designed to motivate individuals by stimulating them to behave in a certain manner, and to achieve certain economic or social outcomes. Finally, a mechanism establishes the fabric between the actions of individuals and social landscapes of software organizations. We suggest that, a mechanism enables us to maximize the economic and social outputs of the software development effort -through modeling the structure of software teams and further envisioning a software development organization. --- The software ecosystem In recent years, initial exploration of the importance of the interactions of an economic community, highlighted the fact that software development organizations should co-evolve their capabilities and roles together for maximizing the opportunities for project and business success. Therefore, the traditional viewpoint of software business; selling software to the mass market, has been replaced by the idea of interacting companies in a form similar to a biological ecosystem [17]. Based on the idea that interacting participants and organizations of the business world is considered similar to organisms of a biological ecosystem, Moore [19] from Harvard introduced the definition of a software ecosystem as an economic coating that forms around a software product. Concurrent to Moore's definition, Mitleton-Kelly from the London School of Economics investigates organizational complexity by applying the theory of complex social systems to the theory of organizations [18]. She suggests that complexity arises because of the interactions through the elements of a complex co-evolving social ecosystem, including all individuals and organizations based on their business, technical and organizational relations among suppliers, customers or competitors. Ultimately, we suggest that, a software ecosystem may be based on various information exchange networks. It could be considered as a set of several business entities working on collective outcomes in a shared market where several entities play distinctive roles . The relationship is based on the exchange of knowledge in terms of several forms, e.g. artifacts. Recognition of the software development organization as a social ecosystem brought the realization that the investigation of its social structure (e.g. --- Process Component --- Activity Deliverable Role Task Software Engineering Process Iteration Increment --- Software Development Life-cycle Figure 1: A meta model for software engineering process adapted from [23]. connectivity, cohesion or coupling of its members) may help to improve the human centric aspects of the business process. --- A model of productivity improvement Productivity improvement is one of the main concerns of a software organization. It starts very early in any development life-cycle. For example, previous research has indicated that size of a project, the development environment and the technologies has an impact on software productivity [22]. Although a generally accepted measurement model of productivity is lacking [13], it can essentially be considered as the production rate or capacity of a process -something that agile software development often terms as the project velocity. Productivity should be considered as the value creation activities in a specific time period, and according to Boehm, the best opportunities for improving productivity in the software development effort are to be found in the attributes of people and their interactions inside the software organization [3]. Consequently, in order to improve the productivity of a software organization, crafted methods and development strategies for software development should leverage the knowledge contained in well-established people-centered approaches. Such efforts can only boost the potential for success in software development companies. For example, a team of software practitioners is not only a good illustration of team oriented knowledge work but also a form of social network. A social structure can be defined in terms of social units, where they are considered as teams of interacting individuals gathered together for achieving a defined goal. In software development activities the intellectual workers continuously collect and process the collected information into knowledge that actualizes as software artifacts. Furthermore, the knowledge assets embedded inside the activities of a software organization are used for generating an economic value. This value should not only be determined by the outcome of the production process but as the human part of the capital which encompasses the value added to the workers during the process and, as the social capital , which is the capital captured by social interrelations. In order to bridge the gap between formal and the social world of software practices, any proposed valuation of a software development should not only be realized by financial form of the capital but also with its intellectual capital and especially in terms of the social capital. In summary, software productivity is heavily dependent on social aspects of productivity which can be achieved by better social alignment, i.e. setting the roles of people better regarding to their personality types for maximizing productivity. In addition, it is the skills of individuals and teams which transform the acquired knowledge into software artifacts and constantly increases the competitive advantage. --- Qualitative Simulation Paradigm The notion of personality types can be considered to be socially constructed entities. Based on behavioral response patterns, personality is like an individuals' mask with which one present herself or himself to others. We suggest that the outcome of several situations are shaped by individuals personality types when they interact. These interactions, however, can be defined in terms of game theoretical forms in a way to use personality types to promote goal attainment of participants. These types are created to use on detailing behaviors of distinctive personality types and their reactions to several situations. Therefore, we term them as game theoretic personality types which is a taxonomy of interactors that can be defined as a strategic situation . We propose an approach for providing a capability to induce the social environment: to reveal participants personality types from a game theoretical perspective, to as-sess how individuals interact on software development landscape, to investigate the cause and effect relationship of social aspects over productivity, to sequentially create personality profiles of organizations based on our gametheoretic classification approach, and to simulate and design team compositions based on several observed and hypothetical situations. Depending upon the complexity of tasks and human interactions, in our setting, a precise quantitative model is hard to construct. However, based on a finite set of qualitative knowledge and relations, we should be able to analyze and simulate several team formations and situations by using several techniques such as sociometric graphs, situational context cards, qualitative simulations for examining several aspects such as communication and social structure of a software organization. Here, we formulate a new approach to overcome the complexity of understanding social and economic activities . We have termed this approach to be a qualitative simulation: a scenario based information gathering, analyzing, and evaluation method relies on blending several qualitative research techniques. This method consists of three process cycles : initiation, generation, evaluation. The initiation cycle is used for creation of GTPTs. It starts with identification of a set of situations. Next, it continues with the observation of participants' types and we evaluate these observations by semi-structured interviews. The sec-ond process cycle is for enhancing the situational context cards creation. Several captured situation will be stored in this context, the cards however aim to help the identification of individuals and their reactions to situations. Finally, the revised context cards will be used to create several random or planned situations and they will be tested on individuals to determine their social coherence for creating optimal team compositions. As a summary, we define qualitative simulation as a methodology based on situational context cards to define several simulation scenarios where we create hypothetical work flow for reapplying observed events to other participants. It should also include focus groups and expert reviews and case study for better construction. For example, it uses semistructural interview techniques to validate the collected information and situational context cards , and to investigate participants reactions to several situations. Based on the data about the persons and the situations, our qualitative simulation model will be able to simulate scenarios and events that can happen during the software development life-cycle and ask several participants about their reactions to several events. Further, we will use this method to simulate and observe the changes regarding to social formation and conflicts among the structure of the software organization and share them with the participants to collect their responses. --- Conclusions and Future Work In this work, the primary outcome will be a modeling approach for team compositions on software development activities in terms of social interrelationships of participants, in particular by profiling the game theoretic personality types of participants. The research will extend the body of knowledge on software engineering based on the complexities several social issues such as team formations, interpersonal conflicts, social loafing that affect the group settings and structure software development teams. In addition to that, we have introduce several concepts as potential impacts and significant outcomes for software engineering research such as the notion of social productivity, and qualitative simulation which is a method formed by the combination of several qualitative techniques orchestrated to simulate several observed or hypothetical events. Furthermore, we will develop game theoretic personality types which represents a novel kind of personality types or traits, specifically designed for team composition suitable for using our game theoretic interaction model. Furthermore, by using the personality traits, a team composer will be designed for software companies not only suitable for team composition but also for the choosing and integration process of new personnel. In developing the concept of social productivity, a survey artifact is created, which will be used to determine the correlation among several factors of productivity and social aspects of productivity. Furthermore, by combining the notion of social network analysis and game theory concurrently, we will obtain a new viewpoint for software engineering research. In addition to the contributions identified above, we consider our work will benefit future software engineering researcher by drawing a road map that establishes a body of knowledge, specifically on the structures and formations of software development teams and organizations. Furthermore, this project may also useful for researchers in examining the experience of applying game theoretic concepts to software development organizations, in particular the practitioners who are seeking to improve the social productivity of their organizations and to assist in the complications of social issues related to software development productivity. For example, uncontrolled fluctuations in team velocity that could cause considerable complications.

Software development involves teams of interconnected individuals who are encouraged to work collectively in a knowledge and communication network to produce software artifacts. At the social level, the interactions of these participants and their ability to cooperate are important for improving the productivity of teams and organizations. Software development is a knowledge and human intensive activity. It is therefore not surprising to discover that recent contributions in software development have repeatedly asserted the critical role of people in software development efforts. However, existing approaches to software development fail to fully exploit the importance of social and intellectual capital that has been highlighted in the fields of economics and sociology. Leveraging the existing approaches from economics and sociology and applying to software development can assist software organizations in maximizing their return on investment. For example, by applying one such approach, mechanism design, we can improve and model the organization's total productivity based on social aspects affecting productivity (i.e. social productivity). Social productivity involves targeting the quality of social interactions in order to bring about productivity improvements. Furthermore, by optimizing the social structure and welfare of a software organization, we aim to improve software team collaboration, and maximize the team productivity.

Barnes, J.; Ball, M.; Niven, L. Providing the family-nurse partnership programme through interpreters in England Health and Social Care In The Community 19, pp.382-391 --- Introduction The Nurse-Family Partnership programme , a manualised nurse home-visiting service, is known in England as the Family-Nurse Partnership programme . Designed to support young, vulnerable, firsttime mothers, it draws on attachment , self efficacy and ecological theories and delivery is with a strength-based, motivational interviewing style . The detailed curriculum has three phases, pregnancy, infancy and toddlerhood . The concept of 'fidelity' is of paramount importance in the delivery of FNP with quantitative programme delivery objectives developed in the USA for each phase ). Its aims are to improve pregnancy outcomes, child health and development and the economic self-sufficiency of the family by helping parents develop a vision for their future . The effectiveness of the programme had been demonstrated in the USA with three randomised trials . In one it was provided by bilingual nurses in Spanish to mothers with Latino backgrounds but interpreters are not generally used in the USA . Since 2007 it has been tested in England, first in 10 pilot sites and at the time of writing in 50 sites around the country with interpreters involved where necessary. Sites have 4-6 nurses, each with a maximum of 25 clients. With the substantial and growing minority ethnic population in the United Kingdom programmes such as FNP, offered within the National Health Service , need to consider their applicability for non-English speaking mothers. It has been found that language barriers present a major obstacle to minority ethnic communities accessing healthcare . This mixed method study, one element of the ongoing national implementation evaluation of FNP ), compares quantitative aspects of FNP delivery with and without interpreters and examines emerging issues through qualitative interviews with nurses, clients, interpreters and their managers. Therapeutic work with an interpreter may be particularly adversely affected by omission of information, simplification, adding content, replacing concepts with those thought to be more understandable and interjecting opinions . It has also been suggested that development of empathetic therapeutic communication may be adversely affected by the interpreter's presence . District nurses in England reported that interpreters could detract from them developing relationships with their patients and reduce information exchange about emotional concerns . Research has shown that that nurses working with vulnerable refugees, concerned about the impact on their relationship with clients, often prefer to manage without interpreters . A Canadian review concluded that nurses, compared to physicians, may find working with interpreters particularly problematic in that their relationship with patients is more sustained and personal. This is also likely to be true for the FNP nurses who ideally work with clients for the duration of the programme, covering many sensitive and personal topics. A fundamental aspect of successful delivery of FNP is said to be the development and maintenance of a close relationship between the nurse and client, with a sense of common goals or purpose and feelings of safety and trust . It has also been pointed out that nurses working with an interpreter need to maintain not one but three dyadic relationships; nurseclient, client-interpreter and nurse-interpreter . In relation to providing the manualised FNP with interpreters the study aims to investigate whether the expected levels of delivery are attained and whether the nature of the crucial client-nurse relationship is affected. The study addressed four questions. 1. Was there any difference in delivery based on quantitative objectives when interpreters were involved? 2. Was there any perceived impact of the interpreter on delivery of the programme's content? 3. Was there any perceived impact of the interpreter on the nurse-client relationship? 4. What kinds of relationships developed between interpreters and both clients and nurses? --- Methods --- Data collection --- Programme delivery forms Standardised forms are completed by FNP nurses to collect demographic client information at intake and after each home visit recording delivery . All anonymised forms covering almost two years were available from 10 sites, from April 2007 to February 2009. --- Qualitative Interviews Semi-structured interviews were created for the study by the authors, based on the literature and on previous interviews with FNP nurses and clients about the programme's implementation ). They were conducted in April and May 2009 in three sites where the majority of the work with interpreters was conducted. Nurses were asked about their previous experience with interpreters, perceptions of programme delivery though an interpreter, modifications to delivery and issues raised in supervision regarding interpreters. Client interviews covered recruitment, introduction of the interpreter and their involvement in subsequent visits; and their relationships with the nurse and interpreter. All but one was conducted with interpreters . Interpreters and their managers were asked about training and background, their understanding of the FNP and perceptions of how well the interpreting of FNP visits was working. All interviews were recorded and transcribed. --- --- Ethical considerations Ethical approval was obtained from an NHS Research Ethics Committee for the analysis of anonymised data forms and then separately for the qualitative interviews. All interview participants were given information sheets describing the study and gave their written informed consent at the time when interviews were conducted. --- Data analysis Analysis of programme delivery comparing interpreter and non-interpreter clients was conducted. Continuous variables were compared using independent samples Student's t test, or Welch-Satterthwaite t-test where variances were unequal; categorical variables were compared using Pearson's chi-square test, or the Fisher's exact test when cell frequencies were low, with SPSS Release 16. Qualitative interview transcripts were initially analysed to identify themes by all three authors with validity ascertained though discussion and consensus, using thematic analysis . Then the interviews were re-visited by the first author for final coding. Formal reliability was not assessed. --- Results Programme delivery forms were available for 1304 clients, 43 non-English speaking for whom an interpreter had been present for some or all home visits and 1261 English-speaking. Clients requiring interpreting were predominantly located within three sites with 21, 12 and five respectively, while one site had two, three sites had only one and three sites had none. --- Demographic characteristics The 43 clients using an interpreter required translation from 14 different languages: Bengali or Sylheti ; Polish or Urdu , Albanian, Kurdish, or Punjabi ; and Chinese, Creole, Persian, Portuguese, Sign language, Somali or Spanish ; with four with no information. Compared to other clients, those who required an interpreter were unlikely to be of school age but more likely to be non-teen; more likely to be married and more often living with their partner and other family but not with their own mother, to be of Asian background and to have never been employed . --- Please insert Table 1 here --- Attrition According to nurse completed data forms a similar proportion of clients in each group left the programme during pregnancy . However, while deemed active according to nurses, fewer than that number received any infancy visits . --- Impact on quantitative delivery objectives Gestation at recruitment was similar for both groups, slightly later than the objective of 16 weeks . The objective is that 80% of the expected visits are delivered in pregnancy and 65% in infancy . The percentage was short of these objectives for both groups . Visits should last on average at least 60 minutes which was achieved for both groups with no significant difference during pregnancy or infancy . In both pregnancy and infancy the percent of the planned content covered was significantly lower for clients requiring an interpreter . Coverage of the domains was generally similar for both groups and mainly in line with the objectives except that in pregnancy more time was spent on maternal personal health for clients requiring interpretating compared to the remainder and less on environmental health. In infancy again less time was spent on environmental health for the interpreter group . Nurses' ratings of clients' involvement and understanding during visits were lower for those requiring an interpreter in both pregnancy and infancy . --- Please insert Table 2 here --- Perceived impact of interpreters on delivery of the programme Contrary to the quantitative analysis showing no significant difference between groups for the average visit length , nurses reported that sessions with an interpreter seemed longer: "I speak to the client and she looks me in the eye, then looks to the interpreter as she translates and I look at her to see her expression… They are longer than other visits." [N11] This suggests that time may hang heavily when the interpreter is involved: "It's quite boring at times….. the interpreter goes through it and you just sit there." [N3] One nurse, who also supervised, considered that the process of interpretating diminished the essence of the programme: "I don't think nurses feel that having a third person to interpret their words is a good way to deliver FNP; something is lost." [N14] It was also thought that "Motivational interviewing does not work because you have to be more direct." [N15] In addition to the essence or manner of the communication being changed there was also concern that the programme relied on written materials, which could not be used when clients or interpreters had literacy problems: "The written material is just impossible, not all interpreters can write the language they are speaking -and the girls definitely can't." [N16] However written materials were seen as useful in developing self-efficacy by some clients: " helps me through the interpreter to complete the form. I write on it during the visit. --- She gives it to me intentionally to see if I can practice my English." [C5] Other aspects of the FNP curriculum rely on nurses modelling parenting activities such as mother-infant play using dolls or stuffed toys, which can be complex when a third person is involved, but it appears that the nurses and interpreters generally managed: "I was talking about the five stages of play and getting the interpreter to relay that. Then I demonstrated it with a teddy on the floor. When it came to the mum's turn I had to stop her to say through the interpreter to let the child lead." [N9] This requires skills that interpreters may not use in other work; the role playing needs to be delivered in the same tone of voice as the nurse: "It has actually been quite easy because is quite animated, she says it how I say it." [N2] An interpreter manager estimated that 90 to 95 percent of what has been said is likely to be accurately translated [IM2]. However, uncertainty about whether wrong messages were being conveyed was described by some nurses, who were unsure how they should tackle it: "I was listening with all ears. I knew that she was saying things that I wasn't saying and I didn't tackle it…I didn't have the energy to be constantly diluting what the interpreter was saying." [N10] Others indicated that they had addressed this directly with interpreters. For example, after providing some information about weaning the child with family foods, one nurse discovered that the interpreter was recommending the use of purchased baby foods: "I asked her immediately afterwards not to express her own opinion. She was accepting, but she might do it again and I would not know." [N5] The extent to which interpreters may add their own opinions was linked to other work as a health advocate. Interpreter managers were inclined to dismiss this as a concern, one saying "An advocate can act as an interpreter but an interpreter cannot act as an advocate" [IM1]. Nurses considered that being an advocate may, in addition to adding their opinions, lessen the likelihood of the motivational approach being used correctly: "We were quite adamant that we did not want to have advocates as we're not an advice service, we are a motivating service" [N9]; "If somebody has been employed as a health advocate and thinks she knows everything, she would probably go off on her own tangent and start telling her things. That would not be appropriate when you are doing motivational style work." [N10] In addition to having the content of the programme presented incorrectly or without the relevant strength-based focus, there was also evidence that important issues may not be addressed in the interpreter's presence. For example a client whose visits were now without an interpreter since her English had improved reported that she had previously held back: "Sometimes I wanted to ask about something personal and I used to feel shy to do it through an interpreter. She wanted to know about the relationship between me and my husband. Now I feel free to speak." [C5] One nurse indicated that she avoided some physical health issues with male interpreters: "There are things I shelved like breast examination because I don't feel I can do that through a male interpreter" [N3] although a male interpreter considered that his gender was not relevant: "In our country there is no problem, when a lady is pregnant they talk about everything." [I1] However for women of some cultural backgrounds a male interpreter would not be acceptable: "We wouldn't be let into the house if we turned up with a male interpreter." [N12] The perceived impact of interpreters on nurse-client relationships Nevertheless some mentioned difficulty in initiating good relationships when using a third party: "I find it difficult to read between the lines when I've got an interpreter." [N12] "Having the interpreter there was a barrier to my relationship with the client." [N7] The time it takes to interpret between the client and FN can add a restrictive element to the dialogue and as a result one nurse commented: "They cannot be spontaneous, they cannot say everything they would like to." [N4] For one nurse it was only when the interpreter went on sick leave that she began to get an insight into the relationship between the young woman she was visiting and her partner who took on the interpreting role: "When she wasn't there he would talk to his wife and tell me what she'd said, they would giggle together and I saw another side to them." [N7] This nurse and client decided not to carry on with the interpreter as did other nurses, even if they thought their delivery was less accurate: "I'm probably not doing a lot of them as I should but I think they are good enough and she prefers not to have an interpreter."[N17] Clients also mentioned they were relieved when they received FNP without an interpreter. "I used to think 'Oh, I have to explain --- to someone else, I wish I could speak to instead of through an interpreter." [C2] " I feel free with , I don't feel hesitant." [C3] Lack of continuity of interpreters was also thought to impair the relationship: "Some of the clients take quite a long time to trust someone. I'd already gone in with and built a relationship, then she is gone and here is . I didn't want to feel that just any old person could come in." [N10] The general perception from the nurses could be summarised by this nurse's comment: "It will always be jolly difficult to provide a long-term programme based on your relationship with that person when you're doing it through a third party." [N3] --- Relationships between interpreters and both clients and nurses The clients tended to talk quite dispassionately about their interpreters, referring to them in terms of the quality of their language phrasing, in contrast to their expressed fondness for their nurses. Comments were made such as "She was good" [C6],"All the interpreters have been good, I have understood them" [C5], "She speaks really good {language} so she explains properly"[C8] and "I liked them, they were not bad." [C3] The relatively distant relationship is illustrated by this client talking about two interpreters: "I didn't prefer one to the other, they were not interesting, but they were friendly, I felt comfortable with them." [C1] One interpreter mirrored this view saying that "I am there just as the middle person, I am not there 'for' the nurse, I am not there 'for' the client…you must be neutral" [I1] an opinion confirmed by the other interpreter interviewed who stated "The relationship is between the nurse and the client." [I2] Most of the clients had experienced more than one interpreter and this was not said to be a problem. While they might prefer the same interpreter each time their feelings were not strong: "They were Bengali like me, they were all good. I would prefer if I had only one but it is not my matter to decide, if others come it doesn't make any difference to me." [C4] However one client who had experienced several interpreters indicated why she preferred her current one "Emotionally she is sentimental, we have a good bond, she has sympathy. She is 21, she understands me." [C5] Similarity between the client and the interpreter was thought by nurses to facilitate the development of interpreters' relationships with clients, something that helped the whole process to be successful: "The three of us just get along really well. I think the interpreter is getting something out of it herself, she's a mother herself." [N1] A Bengali interpreter also mentioned the benefits of her similarity to the clients; "The families they are like me, I got married at 17; I enjoyed it, the work is fascinating, it crosses over with my experience and my interests." [I2] Interest and appreciation of the programme approach itself was also useful: "She's really enthusiastic about the programme….. when she hears my voice sounding sensitive then she'll do the same." [N2] She went on to explain how she knew that this interpreter's sensitivity helped the development of a trusting relationship with the client: "The other day we did the 'Trust Petal' where you write who you trust and two of the people were and me. I asked her why and she said 'because helps me to communicate so my voice is heard. '" [N2] Nevertheless, nurses were not always positive about close client-interpreter relationships. One thought the attitude of the interpreter who "had a grandchild the same age as the client's baby" and "kept personalising the exchanges, bringing her own grandchild into it" was detrimental to her own relationship with the client. [N6] Another commented about an interpreter who was well known locally that: "She was lovely but things turned into a conversation with her and the client." [N8] When a close rapport builds between a client and interpreter the nurse could feel excluded: "Sometimes would be having a little aside conversation with and I would ask, 'so what is she saying?' and would say 'oh nothing'. It was quite difficult." [N7] Another felt so sidelined by the interpreter-client relationship that "I tried to get to the house early, to have some time alone with her." [N5] Nevertheless clients did not necessarily want the interpreters as friends, "She asked me a lot of questions, but I have my own friends" [C2] and one interpreter manager noted that clients may not wish to be close to others from their home country "for fear --- of being judged" [IM3] A number of factors were identified that contribute to a good working relationship between the nurse and interpreter. The first was practical, having time either before the session to prepare or afterwards to discuss the visit: "I always brief them before we visit, but there is never enough time" The nurse-interpreter relationship could be strengthened when the interpreter presented himself or herself as a professional, working in partnership with another professional, helping mutual trust to develop: "We have a rapport too now. It's almost like I'm delivering the programme with her …I have a lot of trust in , and has a lot of trust too, which makes me reassured. --- ""[N2] Caution was indicated however in that the nurses were aware they were really dependent on the interpreter; they were concerned about not being in control. Thus the nurse-interpreter relationship was not one that could be thought of as equal professionals but this was not always achieved "My mistake was to allow the interpreter to have intellectual input… I should have imposed tighter boundaries." [N4] There are always going to be two individuals rather than one for the nurse to be aware of "All the time you are trying to make sure that two people are OK, the girl and the interpreter, you need them both to understand exactly, it is your responsibility." [N16] Overall, balancing all three relationships could be stressful: "I have to say I don't dread going at all but when I've finished I think "phew!" and I feel lighter because I've done it." [N3] "You gain something from the help and input of the interpreters, but you lose the spontaneity of the client -there is always this third person there. --- " [N5] --- Discussion This study has some limitations. These nurses were the first in England to be trained in the delivery of the programme. Working in multi-cultural areas they all had some previous experience of interpreters but delivering the detailed FNP curriculum in combination with the strength-based motivational approach was unlike previous medical interactions. Being new to the materials they may have been particularly anxious that the programme content was translated appropriately. Nurses with more experience of the programme may find it less stressful to incorporate an interpreter into their work. In addition the study is limited by the predominance of Bengali and Sylheti speaking clients in these three sites so generalizability to clients with different cultural backgrounds may be limited until there has been more experience of FNP with a wider range of families. The study would also have been stronger if more clients and more interpreters had been interviewed; the focus of this study has been more on the nurses' perceptions of the experience. It must be kept in mind that, while statistical differences were identified in delivery this does not mean that the interpreter's presence caused them. A larger study could take confounding variables into account. Finally the statistical comparisons are based on groups of unequal size. This has been taken into account statistically but investigation in the future with a larger group of interpreter clients is important. With those provisos in mind some useful information has emerged about whether the FNP programme can be delivered effectively with interpreters. Quantitative data indicated that the FNP can be delivered with equivalent fidelity with or without an interpreter. The relatively small differences were: not as much of the planned material covered in each visit; slightly more focus on maternal health and less on environmental health, and clients' understanding was thought to be lower. Covering less is not surprising given the additional time it may take to go through materials twice, with the interpreter and then for the client. It is likely that the other differences in the focus of the content could be related to the more stable housing of most of the clients using interpreters, the majority of whom were older and married, and the nurses' awareness that pregnancy and infant outcomes are on average poorer for women of ethnic minority background . The judged lower understanding may reflect the pervading issue that the nurse was not getting their whole message across. However based on qualitative comments about the perceived impact of the interpreter, the conclusion of delivery with fidelity is tempered. Reflecting previous studies concerns were expressed by nurses that information was not conveyed as stated, or in a motivational style, or interpreters' opinions were added. In the interviews nurses indicated that interpreters did not always translate with a focus on clients strengths and motivation, but rather were didactic. This suggests that interpreters may benefit from time learning about the programme, not necessarily all the materials but the philosophy behind it, which would be more likely of there was consistency over time keeping the same interpreter for each relevant client. The third question was whether nurses could form the close trusting relationships with their clients, said to be important to programme delivery and setting FNP apart from other services for clients . In contrast to some previous research these nurses and clients confirmed that trusting and close relationships were being developed through an interpreter. This may be related to the carefully structured materials, all of which focus on the strengths and motivation of the client and the supportive role of the nurse. However, many nurse and clients would rather than the interpreter was not there. Considering the final question about the relationship interpreters developed with the nurse or the client, nurses could become concerned if they thought clients were close to interpreters. This led to some nurses feeling sidelined and less in control which is likely to have a negative impact on their relationship with the client. Nevertheless, clients did not express strong desires to form relationships with interpreters. It was their positive relationship with their nurse that they mentioned more often. The interpreter could enhance nurse-client relationships as long as their own relationship with both nurse and client was maintained as friendly but not too close, with an understanding that they were not expert in the programme's background or content. It is also important that the nurse believes her message is conveyed accurately and with the right emotional content. Comments made by nurses revealed that they would like more time to spend with interpreters, not all of whom could be fully briefed about the FNP materials or strength-based approach. However in the real world, as the research team also found, interpreters are in short supply and often can only be arranged at the last minute so this may be unrealistic. These nurses worked with their clients over an extended period of time, the kind of support that may be vulnerable to the impact of needing an interpreter . While the nurses considered that they had been able to develop and maintain the appropriate rapport with their clients, it was not always easy. The maintenance of professional, open and sharing relationships with the interpreters was sometimes difficult to achieve in parallel with establishing the necessary close nurseclient relationship and some nurses described the stress they felt during or after interpreter visits. They may benefit from specific supervision sessions that focus on presenting FNP through interpreters, enabling them to discuss any anxieties about the interpretation of programme content or stress related to maintaining multiple relationships over time. In the future a more substantial quantitative investigation, with a broader range of ethnic backgrounds represented, may illuminate in more detail what any differences in delivery might mean in relation to enhancing delivery or predicting programme impact.

This study looks at the delivery in England with interpreters of the Family-Nurse Partnership (FNP) home-visiting programme for vulnerable, young first-time mothers, known in the USA as the Nurse Family Partnership. FNP is manualised with a number of fidelity objectives. This study covers the first two phases, pregnancy and infancy (up to 12 months). The programme relies on the development of a close nurse-client relationship. Interpreters can be a barrier for therapeutic work with vulnerable groups. The aims are to determine from quantitative and qualitative data whether the FNP programme can be delivered with fidelity in the presence of an interpreter and to explore issues concerned with the impact of interpreters on relationships. Statistical comparisons were made of delivery objectives over two years, April 2007 and February 2009, in the 10 sites in England, spread across all nine Government Office Regions providing FNP. 43 clients had an interpreter at some point and 1261 did not. Qualitative interviews were conducted between April and May 2009 with 30 stakeholders (nurses, clients, interpreters). In relation to quantitative indicators the percentage of planned content covered in visits was lower with interpreters (pregnancy: 90% vs. 94%; infancy 88% vs. 93%) and both understanding and involvement of clients, as judged by nurses on 5-point scales were lower (understanding, pregnancy 4.3 vs. 4.6, infancy 3.8 vs. 4.5: involvement, pregnancy 4.4 vs. 4.7, infancy 3.7 vs. 4.5). The interpreter was thought by nurses not to impede the development of a collaborative client-nurse relationship unless the interpreter and client became too close but some nurses and clients reported that they would rather manage without an interpreter. Some stress was noted for nurses delivering the programme with an interpreter. More research is needed to determine the extent to which interpreters accurately convey the programme's strength-based approach.

Introduction China's population aging is accelerating; 18.7% of its population was 60 and older at the end of 2021. This proportion of the older population has increased by 5.44% compared to ten years ago [1]. In addition, approximately 350 million people worldwide suffer from depression, making it one of the most prevalent mental health disorders. In terms of economic impact, depression ranks among the top five disease categories globally, accounting for 4% of the overall disease burden and 10% of the economic burden associated with non-fatal diseases. The prevalence of depression among older adults in China varies from 11 to 57%, with higher rates observed in individuals over 80 [2]. Thus, preventing and alleviating older people's mental health disorders has become an urgent health crisis in promoting healthy aging. As people get older, individuals physical ability decreases while their vulnerability increases, and they become sensitive to the surrounding environment [3]. In addition, life events, such as widowhood and the death of friends, may make them more dependent on the neighborhood environment and community resources [4]. The neighborhood environment is important for shaping a healthy daily lifestyle and social interaction activities among older adults [5]. The neighborhood recreation facilities environment refers to the community places that provide daily activities for older adults to socialize and spend time [6,7]. According to the Ecological Systems Theory proposed by Bronfenbrenner in 1979 [8], the community environment is considered part of the mesosystem in the model. This involves the interaction between individuals and their living situations [9]. The neighborhood recreation facilities environment is closely connected to an individual's living environment and is considered a proximal factor that may directly influence an individual's mental health status [10]. Previous research suggests that older adults engage in social interactions through community recreational facilities [11]. Consequently, they may acquire social resources such as friendship and empathy, establish social networks, and receive social support. These factors can potentially mitigate negative emotions among older adults [12]. Yang found that community-based senior activity centers and open-air fitness equipment significantly predicted depression among older adults [13]. Liu et al. found that the number of recreational facilities reduced depressive symptoms and alleviated loneliness among older adults [14]. Existing studies have validated the correlation between community recreation environments and the mental health of older adults using cross-sectional data. Notably, the Ecological Systems Theory is categorized into microsystems, mesosystems, exosystems, macrosystems, and the dimension of time. However, few studies have focused on how time affects the interaction between people and the environment [15,16]. Therefore, this study will use longitudinal data to examine the effects of community recreational facility environments on the trajectory of depressive symptom changes in older adults. Besides, understanding age difference in health effect of neighborhood environment is also a policy entry point for addressing the challenges of growing and aging cities. Specifically, if the negative impact of the community environment on mental health becomes more pronounced as older adults age, then it is important to focus on making age-appropriate environmental changes to improve the mental health of older adults. Neighborhood Sense of Community among residents showed age related difference [17]. Some studies have found that there was significant age difference in the relationship between neighborhood environment and quality of life among older adults [18]. Previous empirical studies have not directly addressed the age difference in the association between the community environment and mental health among older people. Thus, it is important to identify the age difference in neighborhood recreational facilities and trajectory of depressive symptoms. Neighborhood recreation facilities are less time-consuming and more cost-effective to renovate than major road improvements and pollution treatment projects. The results of this study will assist public health departments in quickly improving older people's mental health. --- Data and method --- Sample We obtained the data from the China Longitudinal Aging Social Survey , designed by the Institute of Aging, Renmin University of China, and implemented by the China Survey and Data Center, Renmin University of China. This survey covered 29 provinces in China and targeted older adults aged 60 years and older. CLASS conducted a multi-stage probability sampling, dividing the district areas and community levels, with its first nationwide survey completed in 2014 and follow-up surveys every two years. During the baseline survey in 2014, the research project distributed 25 questionnaires in 476 randomly selected communities from 30 primary sampling units , with 11,511 samples obtained in 2014. In 2016, this survey received 11,471 study samples, including 4,553 follow-up samples; in 2018, it obtained 11,418 study samples, including 3,874 follow-up samples. Therefore, the initial sample size for the study can be interpreted as a total of 3,874. This includes 3,874 samples from 2014 , 3,874 samples from 2016 , and 3,874 samples from 2018 . This study used data from three rounds of follow-up surveys . Three waves of data were pooled together with ID. First, the study excluded samples that responded with "don't know" or "can't answer" in the depression symptom survey. Second, the data used in this study are in a three-level nested data format: "time point -individual -community", and it is necessary to specify the minimum sample size in each level. In a study on " Adequate Sample Sizes for a Three-Level Growth Model, " researchers found that a sample size of 30 groups, each consisting of 10 individuals, was sufficient for estimating unbiased fixed effects [19]. Most studies support this finding [20,21]. The sample size of the community in this study was 333 groups, which meets the requirement. Therefore, the study screened the number of individuals at the community level, and communities with fewer than 10 individuals were excluded. In the end, the sample for this study consisted of 1,268 respondents living in 333 communities. --- Measurements This study's dependent variable was depressive symptoms. In 1977, Radloff published the CES-D scale containing 20 items [22]. The present study utilizes a subset of the CES-D scale, developed and tested in a study of the living conditions and health status of older adults in Taiwan [23]. The scale consists of nine items in which older adults indicate the frequency of their depressive symptoms in the past week. Each item corresponds to three response options . The study reverse-scored the questions on the positive mood to obtain a total score for the nine items on a scale of 9-27. Higher scores indicated more depressive symptoms. The Cronbach′s α coefficient for the scale was 0.82. Neighborhood recreation facilities was the independent variable. In this study, neighborhood refers to several apartment complex under the jurisdiction of the same neighborhood committee. CLASS asked respondents about the facilities provided in the community to determine whether the community offers a recreational environment. Previous studies concluded that a community recreational environment provides a space for senior fitness and socialization and improves the quality of life for seniors by enhancing their perception of life and promoting emotional relaxation. The premise of these environmental effects is that older adults perceive the community as providing a recreational environment. Therefore, it is more convincing to use older adults' subjective responses to determine the health effects of the community environment. Neighborhood recreation facilities was divided into five sections: outdoor activity area, fitness room, library, activity room, chess and card room. Each activity facility was scored on two points . The control variables included age , gender , place of residence , education level , marital status , cohabit with children , income , self-rated health, and social network , developed by Lubben et al. [24]). --- Method Data analyses were performed in Stata 15.0 software. Firstly, descriptive analysis and correlation analysis was performed to describe the main characteristics of older adults. Second, the study performed a three-level linear growth model of "time point -individual -community" to fit the trajectory of depressive symptoms and focuses on examining the role of age and neighborhood recreation facilities on the trajectory of depressive symptoms. In the three-level linear growth model, individual's observation record is the primary level, the second level of the model include the basic characteristic of older individuals, and the community is the third level. With the above in mind, we divided three-level linear growth model analysis into four steps: We constructed a null model without any explanatory variables to test whether depressive symptom scores had significant variance across dimensions . We included the time point and its quadratic term as independent variables in the model to predict the trajectory of depressive symptoms over time . We included the neighborhood environmental variables and their interaction terms with time point variables in the model to investigate the impact of neighborhood environmental characteristics on individual depressive symptoms . We performed subgroup analysis to investigate potential differences in the effects of community environment variables on depressive symptoms across different age groups . --- Results --- Descriptive analysis Table 1 illustrated the basic information of all variables. Depressive symptoms increased among older adults from 2014 to 2016 and decreased from 2016 to 2018. The mean score of depressive symptoms was 14.35 in 2018. Each of the neighborhood recreation facilities showed an upward trend over the four-year period. Based on Table 1, the most common recreational facilities in the community are outdoor activity space and activity room . Most people belong to youngest-old group . 44.47% of older people are female, and 51.26% live in urban areas. fewer older people were highly educated, with only 8.28% participating in junior college and above. 20.82% of older adults were illiterate. Most older people had a spouse , this percentage decreased by 2.1% between 2014 and 2018. More people did not choose to live with their children . --- The effect of neighborhood recreation facilities on depressive symptoms In this study, a three-level linear growth model of "time point -individual -community" was constructed to fit the trajectory of depressive symptoms among older individuals. In model 1 , null model included no independent variables to examine whether there are significant time and community differences in the depressive symptoms. The intra-group correlation coefficient of time point was 0.218, indicating that 21.8% of time point variance in depressive symptoms. Besides, there is 4.5% of the community variance in depressive symptom. Model 2 included the time variable and a quadratic term of the time variable to reflect the nonlinear trend of depressive symptoms. The results revealed a significant increase in depressive symptoms over time. . Additionally, there was a pattern observed where depressive symptoms initially increased and then decreased as time progressed. . Model 3 incorporated community recreational facilities and its interaction term with time variables. The results showed 13.62 ± 6.28 14.44 ± 5.54 13.57 ± 5.09 * Categorical variables were described with the number of cases , and continuous variables were described with mean ± standard deviation that good Fitness facilities had a significant protective effect on initial levels of depressive symptoms in older adults, even after controlling for individual characteristics. Specifically, the presence of better fitness facilities in the community of residence is associated with a decrease in depressive symptoms . Notably, we also found the interaction term between community fitness facilities and the time variable to be significant . This suggests that although the older population experiences an increase in depressive symptoms over time , good fitness facilities can significantly slow down the rate of increase in depressive symptoms in old age. Model 4 and Model 5 perform HLM among two samples based on age difference . The results showed that fitness facilities were significantly associated with depressive symptoms among youngest old people . More fitness facilities in the community would decrease the depressive symptoms among youngest old adults. Youngest old people experience an increase in depressive symptoms over time , while good fitness facili- ties can significantly slow down the rate of increase in depressive symptoms among youngest old people . Among oldest old people, outdoor activ- ity space was significantly associated with depressive symptoms . More outdoor activity space in the community would decrease the depressive symptoms among oldest old adults. Although we found the significant association between time and depressive symptoms, the interaction term between community fitness facilities and the time variable was not significant. --- Discussion This study explored the relationship between neighborhood recreation facilities and depressive symptoms among Chinese older adults and identify the age difference in this relationship. Based on the three-level linear growth model of "time point -individual -community", it was found that fitness facilities significantly associated with depressive symptoms. Fitness facilities was the important predictor for people in youngest-old group and outdoor activity space was a protective predictor for oldest-old people's depressive symptoms. This study found that the increase in community outdoor activity space significantly reduced depression symptoms among oldest-old people, which was consistent with previous studies [25,26]. The open space in the community was the main venue for older adults to perform physical activities and social activities. Neighborhood outdoor activity space was beneficial to improve the daily activities and expand social networks so that it can reduce depressive symptoms [27,28]. In addition, neighborhood outdoor activity spaces created a sports culture atmosphere for the older adults and improved their psychological status [29]. In this study, the increase of fitness facilities significantly decreased depressive symptoms among youngestold people. In China, unlike the heavy equipment found in gyms, community public fitness equipment is both convenient and free. This equipment includes twisters, upper body traction machines, spacewalkers, backboards, and low back massagers. Youngest-old people are more capable to participate in exercise than those over 75 years old. Older adults have a narrower social circle and fewer opportunities to exercise, and neighborhood gyms were convenient for them. Previous studies have also found that exercise weekly helps prevent depression in older adults [30]. The findings in this study have some policy implications. Firstly, neighborhood recreation facilities can be included in health assessments of older adults and used to identify older adults with poorer mental health status. Second, the neighborhood age composition is an entry point for policies addressing aging [31]. Combined with our studies, there was age differences in the association between community activity facilities and depression among older adults. Further understanding of the age composition of different communities needs to be tailored to create healthy neighborhood living environments. Although we obtained some meaningful results, there were some limitations. First, limited by the questionnaire setup of CLASS, objective indicators of neighborhood recreational facilities, such as facility size or quality were not included in this study. Future research needs to measure neighborhood recreational facilities more comprehensively and explore their role in mental health. Second, there is a distinction between presence and use of recreation facilities. This study demonstrated relationship between the presence of recreation facilities and mental health. Future research is needed to compare the effects of frequency of recreational facility use on the mental health of older adults. Third, although we made speculations about the reasons why no neighborhood recreation facilities decreased oldest-old people's depressive symptoms, future studies can apply qualitative interviews to explore the influence mechanisms further. --- Conclusions It was found that community with fitness facilities was significantly associated with depressive symptoms. There was a significant age difference in the association between neighborhood recreation facilities and depressive symptoms. Among youngest-old respondents, neighborhood stressors stemmed mainly from fitness facilities, whereas among oldest-aged people, the stressors came from outdoor activity space. This paper has no Funds. We have no conflicts of interest to disclose. --- Data Availability On reasonable request, these data may be made available from the corresponding author. --- --- Funding Not applicable. --- Declarations Ethical approval and consent to participate All participants provided informed consent, and the protocol was approved by the Ethical Review Committee of Renmin University of China. Informed consent for all illiterate participants was obtained from their parents and/ or their legal guardian. All methods were carried out in accordance with relevant guidelines and regulations. --- Consent to publish Not applicable. --- ---

Objectives Neighborhood recreation facilities has been associated with depression that commonly tested with cross-sectional data. This study used longitudinal data to test the effect of neighborhood recreation facilities on the trajectory of depressive symptoms among Chinese older adults.

Background The association between socioeconomic level and reproductive factors has been widely studied. Several studies suggest that sociodemographic, socioeconomic characteristics and social class would be independent risk factors associated with neonatal morbidity [1][2][3][4]. Women with lower socioeconomic status tend to have more children, the age at the first born being earlier [5]. Other studied aspects have been the number of abortions, performing breastfeeding [6] or having a child small for the gestational age [7,8]. However, most studies did not differentiate between individual and contextual socioeconomic levels. The individual socioeconomic level refers to individual characteristics such as incomes, educational level or occupation, while contextual socioeconomic level would refer to characteristics in the area of residence, such as unemployment and illiterate rates or social facilities [9]. Area-level deprivation form a composite score. The higher the deprivation index value, the greater the level of the neighborhood deprivation. Regardless of individual socioeconomic circumstances, greater area deprivation is associated with an increased risk of premature mortality and chronic disease [10][11][12][13]. However, little is known about how the socioeconomic changes occurred in a country can modify women reproductive factors. Thus, as far as we know, the association between socioeconomic status and reproductive factors has never been studied in terms of the great social changes produced since 1970 in Spain. Therefore, in the field of study of social inequalities in health, it would be pertinent to delve into the knowledge about differentiated patterns of reproductive and neonatal health associated with social class in these periods. For instance, would we expect that university women some 60 years ago differed from the less educated ones in the same way that current university women do? The main goal in this study is to analyze the influence of both individual SES and area-level deprivation index on reproductive factors among Spanish women, and to explore whether this influence has changed in the last 50 years. For the purpose of this study, 2038 women recruited as population controls were used in a multiple case control study conducted between September 2008 and December 2013 in 12 Spanish provinces in Spain . --- Methods --- Study design and population The MCC-Spain is a case-control study on several types of cancers, namely breast, prostate, colorectal, gastric and chronic lymphocytic leukemia, carried out in 12 Spanish provinces. Its main characteristics have been described elsewhere [14]. For the purpose of this study, only women without cancer were considered. Selection of population controls being selected from general practitioners' representative sample of the Spanish women, given the almost universal coverage of the national health system in Spain. All procedures performed involving human participants were in accordance with the ethical standards of the institutional and / or national research committee, and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The protocol of MCC-Spain was approved by each of the ethics committees of the participating institutions. The specific study reported here was approved by the Ethical Committee of Clinical Research of Asturias, Barcelona, Cantabria, Girona, Gipuzkoa, Huelva, León, Madrid, Navarra and Valencia. Informed consent was obtained from all individual participants included in the study. Women were recruited between 2008 and 2013 ; they were selected by random from the General Practitioners roasters. First, they were contacted by phone; if they agreed to participate, they would be cited for a standardized face-to-face interview with trained interviewers. The questionnaire included information about anthropometric data, smoking habits, alcohol consumption. Body Mass Index was calculated taking into account self-reported weight and height 1 year before the interview. --- Reproductive factors The questionnaire also incorporated information about dichotomous reproductive factors as if ever suffering an abortion or a dead newborn, ever use of hormonal contraceptives or hormone replacement therapy, ever suffering fertility problems, ever receiving treatment for fertility problems, ever delivering a preterm or post-term newborn and also includes a reproductive quantitative factor as age at menarche, age at menopause, age at first birth, number of pregnancies, average time between pregnancies, cumulative months of breastfeeding, average breastfeeding time per child, number of abortions and number of newborns alive. We exclude records of women whose age at first birth is below 18 or above 45. --- Individual socioeconomic level In order to measure the individual socioeconomic level, three variables were recorded. Firstly, the participant's educational level . Secondly, the economic level of the parents . Thirdly, the longest occupation of the participant , medium and high . Our questionnaire used a series of comprehensive questions to collect self-reported information on the relative socioeconomic status of participants' parents in three categories . Detailed data on occupational history was collected through face-to face interviews performed by trained personnel. Work was assessed through lifetime occupational history consisting of all jobs held for at least 1 year and included information on age at beginning and end of the job, job title, and the main task of the job. For this article only include the longest occupation of the participant. Seventy participants did not report their occupation and their parents' economic level; four participants did not report the economic level of their parents and 298 did not report their occupation. --- Area-level socioeconomic index To measure the contextual socioeconomic level, we used the Urban Vulnerability Index as published by the Spanish Ministry of Foment [17] . It combines five indicators based on the proportion of: unemployed, unemployed aged 16-29 years old, nonfixed employed, employed without qualification, and people without studies. The UVI-SE scores range from 0 to 1 . We categorized UVI-SE by quintiles, Q1 indicating lower vulnerability and Q5, higher vulnerability. Each participant was assigned to the UVI-SE of her area of last residence. --- Statistical analysis The association of each SE indicator with dichotomic reproductive factors was tested using logistic regression analysis including as regressors all four SE indicators adjusting for age at enrollment and province of recruitment. Results from logistic regression are displayed as marginal probabilities of the event occurrence with 95% confidence intervals. The association between the SE indicators and each reproductive quantitative factor was analyzed using linear regression with all four SE indicators , adjusted for age and province of recruitment. The results are shown as marginal averages with 95% confidence intervals. All analyses were conducted for the whole sample and separately for women born before and after 1950It is noteworthy that women born in that year reached sexual maturity around 1970. This period was defined in Spain by political democratization and the corresponding social and economic changes. All statistical analyses were performed using Stata/SE-16. --- Results --- Sample characteristics Overall, 1031 women born before 1950 with an average age of 70 years and 1007 after 1950 with a mean age of 48 years were included in the analysis. Table 1 displays the main characteristics of the sample. Compared with women born before 1950, women born after 1950 had a higher proportion of overweight and obesity, which is in line with a greater consumption of energy and alcohol per day. The proportion of smoker women was also higher in this group. Regarding reproductive factors, the proportion of women who had their menarche after 12 years old, had a more advanced age for their first birth, had a smaller number of children, and shorter breastfeeding periods was higher among women born after 1950. The use of hormonal contraceptives was also higher in this group and they also reported more fertility problems. --- Association between SE indicators and reproductive factors Figure 1 shows the relationship between socioeconomic level and age at first delivery. All four SE indicators were related with age at first delivery. Higher parent's SE, education and occupational level and lower urban vulnerability were associated with higher age at first delivery. Although these differences are seen in all women, they are higher in women born after 1950 with a difference of about 5 years between the lowest and the highest educational and occupational levels. Age at first delivery was also associated with urban vulnerability, with little changes in women born before / after 1950. In this way, although in Spain, the tendency is to increase the age at first child , supplementary Figure 2 shows that there are changes in the tendency when stratifying by educational level: the age at first child goes down all the time in women with unfinished primary or primary study levels, while the tendency is U-shaped in women with secondary studies. Supplementary Figure 3 shows the percentages of women with university or secondary studies have largely increased in recent cohorts. Therefore, the average change in the age at first child is mostly due to changes in women's educational level rather than changes within each educational level. The association between socioeconomic level and the number of pregnancies is illustrated in Fig. 2. In women born after 1950, but not in women born before 1950, the educational and occupational levels were negatively associated with the number of pregnancies: the higher the educational and occupational level, the lower the number of pregnancies . The percentage of women who have taken hormonal contraceptives is nearly double in women born after The proportion of women who took the pill increased with higher educational and occupational levels and with lower urban vulnerability index in women born before 1950; halving the consumption in the lowest educational levels with respect to the highest ones . and 22.6% in urban vulnerability Q4 compared to 44.7% in Q1 . In women born after 1950, however, there was no association between socioeconomic indicators and having taken hormonal contraceptives except in higher parent's SE. Figure 4 shows the proportion of women who took hormone replacement therapy in the two periods studied. Statistical significance is only reached in the group of those born before 1950, with more educated women near doubling the use of hormone replacement therapy of less educated women and women living in the less vulnerable areas doubling the use of HRT of women living in more vulnerable areas . Education level and urban vulnerability index were the SE indicators associated with fertility problems. Educational level displayed 6% more probability in Fig. 3 Relationship between socioeconomic level and use of hormoral contraceptives. Fig. 4 Relationship between socioeconomic level and Hormonal Replacement Therapy university level for women born after 1950 compared to less educated women. Whereas, a lower urban vulnerability was associated with higher fertility problems, but only in the group of women born after 1950 . --- Association between socioeconomic scores and other quantitative variables Number of alive newborns and average time between pregnancies The relationship between number of alive newborns and SE level mirrors that of number of pregnancies and the average time between pregnancies. In women born after 1950, education and occupational level was negatively associated with the number of pregnancies: a higher educational and occupational level was associated with fewer pregnancies. --- Socioeconomic level and breastfeeding The occupational level was negatively associated with months of breastfeeding in the whole sample and in women born before 1950. The cumulative number of months of breastfeeding was also negatively associated with education level in the whole sample. The relationship between the months of breastfeeding and SE level mirrors that of the average time of breastfeeding per child. --- Association between individual and contextual socioeconomic scores with other dichotomic variables associated with reproduction Socioeconomic level and abortion The percentage of women who have suffered at least an abortion was associated with higher educational and parent's SE level in the whole sample and in women born before 1950, but all these associations disappeared in women born after 1950. --- Socioeconomic level and dead newborn Women with lower educational level were more prone to have a dead newborn, but this result was only reproduced in women born before 1950. No other SE indicator was associated with this event. --- Socioeconomic level and diagnosis of fertility problems The level of education was indicators of SE associated with fertility problems. The educational level showed a V-shaped pattern with higher probability at the university level for women born after of 1950. --- Preterm and post-term newborn The proportion of women with post-term newborns decreased with increasing educational level in women born after 1950. However, no SE factor was found associated with the preterm newborn. --- Fig. 5 Relationship between socioeconomic level and fertility problems No association could be found between SE level and age at menopause, age at menarche and average time between pregnancies. --- Discussion The most challenging result in our study is the emerging of socioeconomic inequalities in age at first delivery, number of pregnancies, number of alive newborns, and diagnosis of fertility problems in women born after 1950. Breastfeeding, suffering abortions or dead newborns, consumption of oral contraceptives or hormonal replacement therapy were associated with socioeconomic level in women born before 1950 but not in women born after that year. Stratifying our analysis according to being born before or after 1950 was not arbitrary. It is noteworthy that women born in that year reached their sexual maturity around 1970, which would be considered some kind of social milestone for Spanish women. Spanish women incorporated to labor market after 1970, the labor activity rate in 1976 being 55.1% for women aged 20-24 and only 29.7% for women aged 25-59 [18]. Secondly, sexual behavior began to change in most western countries in the 60s, involving usage of contraceptives, family planning and women taking more control of their sexual / reproductive lives. The Spanish society, however, had their sexual habits ruled by the dictatorship that ended in the 70s, including for instance late initiation of sexual relationships, one partner only, or inclusion of adultery in the penal laws until 1976 [19]. Therefore, sexual freedom and generalized accessibility to contraceptives was reached by Spanish women from 1976 on, with some delay respect to women in other western countries. Thirdly, a deep economic crisis affecting Western countries began in 1973. This eventually led to high unemployment rates and to deep falls in birth rates, which were more intense in Spain: the average number of children per women fell from 2.90 in 1970 to 2.22 in 1980 and 1.36 in 1990 . In the same 20-year period, the birth rate fell in Spain from 19.5 in 1970 to 15.3 in 1980 and 10.3 newborns per 1000 inhabitants in 1990 [20]. Fourthly, the public health service increased its coverage from 1970 to 1980, reaching universal coverage in 1988 [21]. Finally, the continued social trend toward reduced family size instead of that in the past, in rural areas above all, where children were needed and deemed to be suitable for farming or helping at home [22]. That social context could easily explain changes when the socioeconomic inequalities decreased in women born after 1950. For instance, hormonal contraceptives were marketed in Spain from 1964 on, but their indications initially included keeping the ovary in rest, controlling menstrual cycle, and treating dysmenorrhea and acne, while their contraceptive effect was considered as an undesirable side effect [23]. For years, they were more accessible for highly educated women living in urban areas [24]. When the public health service included contraception in its portfolio, women could access it without socioeconomic differences, leading to the results we have described for women born after 1950. Hormonal replacement therapy appeared much later than hormonal contraceptives, but its usage was amplified via private practicing doctors. This eventually resulted in higher consumption for women in higher socioeconomic levels; in 2001, when most women born after 1950 had not reached menopause, the Women's Health Initiative study found an association between hormonal replacement therapy and several cancers [25] and other chronic diseases [26], leading to a dramatic decrease in hormone replacement therapy in Spain [27]. The emergence of new socioeconomic disparities in age at first delivery and number of pregnancies in women born after 1950, however, is challenging. These inequalities appeared or were intensified in an era of universal coverage of the public health service, with free access to contraceptive methods and widely available information about them. In this regard, it is noteworthy that universal health coverage does not imply -by itselfequity in health assistance. For instance, the OECD have noticed that despite the fact that most OECD countries have achieved universal health coverage, people from the most socially disadvantaged groups tend to have worse access to health services. Possible reasons include lack of awareness of health services, poorer quality of care and co-payments for care [28][29][30]. In this regard, Spanish women aged 25-44 declared having unmet needs for health care in higher percentages if their income are in the lower quintile, although the gradient associated with income level was rather mild in women born before 1950, which could be associated with inequities in accessing health care. The main inequalities in women belonging to later generations, however, seem to be associated with their own decisions , not with access to medical care. Along the same lines, a recent review carried out in 2019 showed how social determinants play an important role in the stage of breast cancer in diagnosis and survival [32]. Selection of population controls is a main strength of this study. Women aged 20-85 years were enrolled in 12 Spanish provinces after being selected from general practitioners' roasters; they can provide a representative sample of the Spanish women, given the almost universal coverage of the national health system in Spain. Some limitations of the study should also be noted. Firstly, reproductive variables were self-reported, which could lead to recall bias; however, as women were not aware of the hypotheses of this study, we would expect that recall bias -if exists-could be non-differential. Secondly, one of the SE indicators we have use -the Urban Vulnerability Index-is ecological by nature, which makes it possible the occurrence of ecological bias. In this regard, aggregate deprivation indexes have been found good proxies of individual income but less efficient to measure education or occupational category [33]. --- Conclusions The way socioeconomic level influences reproductive behavior in Spanish women have changed throughout the time: socioeconomic inequalities in usage of hormonal contraceptives and hormone replacement therapy have disappeared in recent generations, while inequalities in number of children and age at first birth have arisen. Further research is needed to clarify whether this tendency continues in more recent generations of Spanish women. --- --- Supplementary information Supplementary information accompanies this paper at https://doi.org/10. 1186/s12905-020-00936-4. Additional file 1 Supplementary Figure 1. Age at first birth without stratifying by education level. Supplementary Figure 2. Age at first newborn by birth cohort obtained via regression spline. Supplementary Figure 3. Percentage of women with each education level, according to birth cohorts. Supplementary Table 1. Association between socioeconomic scores and quantitative variables. Marginal averages with 95% confidence intervals. Supplementary Table 2. Association between individual and contextual socioeconomic scores with dichotomic variables associated with reproduction. Marginal probabilities of the event occurrence with 95% confidence intervals. Abbreviations SES: Socioeconomic status; BMI: Body mass index; UVI-SE: Urban vulnerability index-socio-economic; HRT: Hormone replacement therapy Author' contributions IGA, TDS and JLL designed the study. IGA, TDS, CP, GCV, BPG, PA, TFV, EA, CSC, JA, AMB, JJJM, ARB, MDCH, JPFV, AM, NA, MSS, GB, VM, MP, MK and JLL collected the data. IGA and JLL analyzed the data. IGA, TDS and JLL interpreted the data. IGA and JLL drafted the manuscript. IGA, TDS, CP, GCV, BPG, PA, TFV, EA, CSC, JA, AMB, JJJM, ARB, MDCH, JPFV, AM, NA, MSS, GB, VM, MP, MK and JLL revised the manuscript. IGA, TDS and JLL accept responsibility for the integrity of the data analysis. All authors agree to be accountable for all the 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. The author read and approved the final manuscript. dación Marqués de Valdecilla ; the Junta de Castilla y León ; the Consejería de Salud of the Junta de Andalucía ; the Conselleria de Sanitat of the Generalitat Valenciana ; the Recercaixa ; the Regional Government of the Basque Country; the Consejería de Sanidad de la Región de Murcia; European Commission grants FOOD-CT-2006-036224-HIWATE; the Spanish Association Against Cancer Scientific Foundation; the Catalan Government DURSI ; the Fundación Caja de Ahorros de Asturias; the University of Oviedo; Societat Catalana de Digestologia; and COST action BM1206 Eucolongene. None of the funders played any role in conducting research or writing the paper. --- --- --- Competing interests All authors have read and approved the content of the article. The authors declare that they have no competing interests. ---

The association between socioeconomic level and reproductive factors has been widely studied. For example, it is well known that women with lower socioeconomic status (SES) tend to have more children, the age at first-born being earlier. However, less is known about to what extent the great socioeconomic changes occurred in a country (Spain) could modify women reproductive factors. The main purpose of this article is to analyze the influence of individual and contextual socioeconomic levels on reproductive factors in Spanish women, and to explore whether this influence has changed over the last decades. Methods: We performed a cross-sectional design using data from 2038 women recruited as population-based controls in an MCC-Spain case-control study. Results: Higher parent's economic level, education level, occupational level and lower urban vulnerability were associated with higher age at first delivery and lower number of pregnancies. These associations were stronger for women born after 1950: women with unfinished primary education had their first delivery 6 years before women with high education if they were born after 1950 (23.4 vs. 29.8 years) but only 3 years before if they were born before 1950 (25.7 vs. 28.0 years). For women born after 1950, the number of pregnancies dropped from 2.1 (unfinished primary school) to 1.7 (high education), whereas it remained almost unchanged in women born before 1950.

INTRODUCTION The COVID-19 pandemic is a major public health and social issue due to its high contagiousness and its lethal potential, its impact on health systems and its massive societal impact due to lockdown measures, travel bans and economic hardship. 1 A considerable number of studies, including in West Africa, have described the health status of patients --- STRENGTHS AND LIMITATIONS OF THIS STUDY Open access and evaluated the effectiveness of new treatments and vaccines. 2 Studies are also aiming at assessing the longterm impacts of COVID-19 on physical and mental health, as well as its social consequences within communities. 2 One of the particularities and specificities of COVID-19 research is that it is carried out in an epidemic context. Epidemics are often challenging events, during which negative opinions and feelings are expressed, such as refusal to believe that the disease exists; circulation of rumours; expression of mysticoreligious beliefs; fear and anxiety; stigmatisation of those who are ill or cured; distrust of governments; or reluctance towards clinical trials. 3 4 Lack of knowledge about an emerging and epidemic-prone disease, fears about treatment and distrust of the research team can be barriers to acceptance of participation in a study. 5 Communities' positive perceptions and acceptability are key for individual participation in clinical trials, including participants' adherence to treatment. 6 Community sensitisation activities carried out before and during research are intended to improve the links between the researchers on the one hand and the beneficiary communities on the other hand, and thus the acceptability and feasibility of the research. 7 In West Africa, such strategies usually include the training of influential people, who then proceed to sensitise community members by informing them through informal channels and offering pragmatic arguments based on their realities, values and ways of functioning. 8 While 37 652 confirmed cases and 449 deaths associated with COVID-19 were reported in Guinea as of 30 September 2022, 1 it is estimated that 65.1% of Africans were exposed to the virus causing COVID-19. 9 Coverage-Africa , implemented in Burkina Faso and Guinea -and part of ANTICOV, an international multicentre platform clinical trial-is currently evaluating the efficacy of early administered treatments to prevent clinical worsening among patients with COVID-19. During the first year of the study, community information sessions were carried out in Burkina Faso and Guinea as part of the implementation of this trial. Studies have explored the perceptions, experiences and attitudes related to COVID-19 globally and in sub-Saharan Africa, highlighting a phenomenon of stigmatisation or fear of being stigmatised. [10][11][12] However, most studies have explored the perspectives of health professionals [12][13][14] and/or used quantitative survey-type methods 15 16 ; very few explored the perspectives of community members and used qualitative methods such as semistructured interviews. 10 11 Guinea has a recent history of country-wide epidemics, including the Ebola outbreak, with several studies documenting high levels of stigmatisation towards sick people and those who recovered from the infection. 17 However, to our knowledge, there are no published studies on the perception of Guinean communities regarding COVID-19, including the exploration of the stigma phenomenon. We therefore sought to explore the experiences, opinions and perceptions regarding COVID-19 of community members living in the urban context of Conakry, the capital of Guinea, to contextualise the implementation of the Coverage-Africa clinical trial. --- METHODS --- Study design To achieve this in-depth exploration, we conducted a descriptive qualitative study in conjunction with a community information session carried out prior to and during the implementation of the Coverage-Africa trial in Conakry, Guinea. The community information session in Conakry included information and advocacy meetings with key community stakeholders in the eight neighbourhoods most affected by COVID-19, that is, in four neighbourhoods in each of the Ratoma and Matoto communes . A total of 28 activities, led by 'health promoters', were organised between October 2020 and September 2021 . --- Selection and recruitment of participants We included community association leaders who acted as key informants providing their insights regarding opinions of the general population, as well as community members who participated in at least one Coverage-Africa information session. Participants were reached through the activity attendance lists. It should be noted Open access that the information provided during these sessions was not exhaustive. We purposively selected participants 18 to cover a diversity of profiles-by age, gender and education level-so that a variety of perspectives could be explored. We did not select participants on the basis of having had COVID-19 or of knowing someone with COVID-19. --- Data collection A Guinean interviewer trained in social sciences conducted in-depth semistructured individual interviews , exploring the following domains, among others: general opinions on COVID-19; behaviours and attitudes of patients with COVID-19 and of community members; and evolution of the local social context related to COVID-19 since the beginning of the epidemic. Interviews with community leaders focused on their perceptions of the opinions and behaviours of people in their communities, and interviews with community members focused on their own experiences and perspectives. Interviews were conducted between 17 May and 6 September 2021 in the participants' preferred language , and by telephone, in order to respect COVID-19 social distancing measures. Interviews lasted an average of 60 min and were digitally recorded. Interviews conducted in French were transcribed verbatim. Interviews conducted in Soussou were transcribed directly in French . All transcripts were anonymised. Verbatim excerpts included in this publication have been translated into English. Relevant information gathered by interviewers, project manager and researchers from informal discussions, that is, fortuitous conversations happening during fieldwork, 18 was also noted; these included discussions with the study's 'health promoters' and social scientists present in Conakry to work on other studies related to COVID-19. --- Data analysis and interpretation of results We analysed the data using the thematic analysis method, 18 which highlighted the main experiences, opinions and perceptions of the participants. This analytical process included the following steps: reading of interview transcripts; deductive and inductive coding of the transcripts ; identification of the main themes and subthemes; summary of each theme; and interpretation of findings using the themes summaries, which were triangulated with informal discussions, knowledge of the sociocultural and political situations in Conakry and literature. These analyses revealed likely data saturation. --- Trustworthiness Various measures were taken to ensure the rigour of the study. Data were triangulated by validating the results of the individual interviews with information collected in the field during informal conversations with key individuals. Thematic summaries were also verified by going back into the raw data . Excerpts from participants' interviews were included in the Results section of the paper, so the readers can have confidence in interpretations. 18 Finally, the interpretations were validated by the Guinean coresearchers and team members based on the field , acting as 'cultural brokers' . 19 20 Ethical considerations The study's 'health promoters' informed the participants of the objectives of the study and invited them to read the information leaflet. All participants freely signed the consent form to confirm their agreement to participate in the study and to have their interviews recorded. A code was assigned to the interviews in order to anonymise them. The data were stored on a secure server protected by a password with access restricted to study researchers. --- Patient and public involvement Patients or members of the public per se were not involved in the design or conduct of the research. However, the 'health promoters', who were involved in community information sessions and in seeking consent from potential participants, acted as intermediaries in reporting to the research team the views of the community regarding the conduct of the study. --- Open access --- Report The study is reported using the Standards for Reporting Qualitative Research guidelines. --- RESULTS --- Description of the sample A total of 10 people , aged between 29 and 70 years, agreed to participate in the individual interviews . Of these, four were key informants and six were community members. Six participants were from the Ratoma commune and four were from the Matoto commune. Some referred to COVID-19 as an infectious disease that exists just 'like malaria, polio, etc' . However, the majority of participants did not report an opinion about the factors that cause COVID-19-neither their own beliefs nor those of people around them. Anecdotally, some referred to the 'virus' , others to unsanitary environments and dirty clothes , 'dust raised by vehicles and motorcycles' or food contaminated by flies as vectors of the disease. --- A benign or foreign illness Because symptoms were reported as being often mild , COVID-19 was largely perceived as a 'bad cold'-a common, mild and transient condition-but not as a dangerous new illness per se. However, some people sometimes expressed a fear of aggravation that could lead to a need for medical assistance. The common cold often catches us. It's our disease. When you have a cold, as soon as you take Paracetamol, it heals. But, the cold that can catch you Additionally, it appears that many Guineans do not feel concerned about the disease, as several respondents mentioned that COVID-19 is considered a 'white man's disease', originating in the West and imported to Guinea by Westerners, affecting and killing Westerners more than Africans. --- A factitious disease You have to see to believe Following the same reasoning as above, many people do not believe in the existence of COVID-19 because they do not know anyone who has had COVID-19 and they do not see sick people around them. Furthermore, social distancing measures and barrier gestures are hardly respected in Conakry , which reinforces beliefs of an unreal disease. People don't respect barrier measures, so if the disease existed, they would get sick. In addition, according to participants, authorities have a policy of not reporting national and local statistics on the course of the epidemic to protect patients from stigmatisation: this, however, seems to contribute to making the disease virtually 'invisible'. It also emerged from the speeches of several participants-particularly among the most educated-that people with little education or who are illiterate have difficulty believing in the existence of the disease. Most of the population is not literate, many believe that [COVID-19] is an invention and that it doesn't exist. A conspiracy rather than a real disease Several participants mentioned a suspicion of manipulation or scam related to COVID-19 to the detriment of the Guinean people. In particular, a large section of the --- Open access Guinean population would have interpreted the concomitance in March 2020 of sociopolitical protests and the occurrence of COVID-19 in Guinea as an attempt by the government to muzzle them, thus deeming COVID-19 to be a fake disease. The Guinean political context doesn't allow us to believe 100%. Isn't this a way of barring people so that they stay [home], to impose that people do not go out when there is a protest while the elections are not over…? Or that it is a manipulation of Westerners to promote the use of their pharmaceuticals. All these problems that occur in our country are made by Westerners. For example, we are being vaccinated at all times. […] these are things that come to us from outside. Fortunately, they are the ones who have the means, as soon as they drop things on us [ie, a disease], they also create the counterpart, that is, The importance of community sensitisation Several participants mentioned the importance of implementing awareness-raising activities to inform communities about COVID-19. In their opinion, this must be done primarily within neighbourhoods, in person-for example, through 'door to door' or educational talk approaches-by trustworthy people , in local languages and with plain language messages. Community outreach seems particularly important for people who are illiterate or have little schooling. The disease exists, but what does it take for the people to believe? […] You have to go through awareness training workshops, otherwise people necessarily doubt… They don't have enough information, they say it's not something that is real. Theme 2: social behaviours/attitudes towards COVID+ people, the healed and their families Stigmatisation of patients Despite testimonies of non-belief in the existence of COVID-19, several participants reported stigmatisation of patients with COVID-19. However, no insult to patients with COVID-19 was mentioned. As soon as we know that a person has this disease, as soon as he passes by, people start indexing him or talking about his disease. Ostracism towards the sick, the healed and their families The fear of being infected by the person diagnosed with COVID-19 often generates mistrust among members of the entourage, who put the patient aside, sometimes for several weeks or months, even after recovery. In particular, they avoid visiting him/her, as well as sharing meals and utensils with him/her. People didn't even go to his yard to draw water, they didn't want to draw water from the same well anymore. They took him out of the social circuit completely for 2-3 months. Some participants mentioned that sidelining is usually temporary: face-to-face contact usually resumes after a while when suspicion of contamination has passed. Family members or people of the patients' entourage may also be targeted for discrimination. […] they move away from family members. They say: 'Oh be careful!! The sick person lives with these people in the same yard, so if you keep in touch with these people, it leads them to dangers…' […] they will segregate him, they hide from him, he is isolated. […] they did not hang out with him as they used to, they no longer come to his house. There is a bodily divorce between them, people are afraid. Consequences of stigmatisation for patients: psychological suffering Some participants mentioned a feeling of shame in patients or recovered people who have experienced stigma or social exclusion, which has consequences on their morale. Some recovered people even moved away because they felt they no longer had a place in their usual social environment. I called him to ask him why he moved? He replied: 'Hah, it's kind of hard… My friends that I thought were supporting me, nobody called me the whole time I was in the hospital, so really I can't stay in the neighbourhood, I'm going to go and live elsewhere'. […] even his best friends didn't call him anymore Strategies to avoid stigma and ostracism Discretion Several participants reported that patients with COVID-19 try to avoid being stigmatised by remaining discreet about their infection. Some patients chose to lie to avoid revealing their COVID-19 status: many make believe that they have 'gone on a trip' while in fact they are hospitalised in a treatment centre. --- Open access After they tested positive, they didn't tell people. They said they're out of town […] so they wouldn't get segregated or their family, so they don't get singled out, that people talk in the neighbourhood. Others conceal their status hoping people will forget they have been infected, or pretend they don't have COVID-19. [He] doesn't even show up [so he doesn't have] to say that he had COVID-19 and that he's cured. He does not engage in any behavior that can show you he has COVID-19. He pretends not to have COVID-19 […]. Presenting a certificate of recovery When they are discharged from the treatment centre, those cured from COVID-19 obtain a 'certificate of recovery' signed, stamped and transmitted by the National Agency for Health Security. This 'COVID-19 pass' allows them to resume working. Some present this certificate to avoid being sidelined by members of their entourage following their infection. There is the example of the boy who had COVID-19 and when he left the [treatment] center, he showed me his recovery paper. I was going out in the morning for work, he showed me his certificate and when I got back at 2:00 p.m. I found him, he was sitting around tea with some friends, people even took turns passing the paper to see if exactly he got sick or not. Others who have recovered from COVID-19 are also proactive, reaching out to people around them to let them know that they are cured from COVID-19. They feel that this facilitates their reintegration into their usual social environment. Protection from contamination and moral support Some participants reported observing empathetic behaviour towards patients with COVID-19, moral support for relatives during illnesses being a central element of Guinean culture. They do not provide material support but moral support, such as greeting them, as we are used to doing, as we are accustomed to. Moral support is a custom for us. If I am sick, those around me will be informed and will come to support me morally. Some reported ways in which they were personally able to provide support to a friend/relative with COVID-19 who was hospitalised in a treatment centre , or to have contact with him/her at home while preventing being infected, that is, either by phone or by video calls, by staying at a distance at the window to talk to each other or by wearing a mask and respecting physical distancing. Sometimes I spoke with him, I put him on a video call, together with friends we chat, we joke, sometimes we sent him money to his account, or credits. It reduced his stress a bit because he was there [and] had nothing. Evolution of stigmatising attitudes/behaviours since the beginning of the pandemic Several participants referred to the recent Ebola outbreak experience in Guinea, and mentioned that some initially thought that COVID-19 could be a similar disease, as in the early stages of the COVID-19 pandemic, the government and health authorities used similar infection control measures as those used to fight against Ebola. In particular, the homes of the sick were sprayed with disinfectant, which attracted attention in the neighbourhoods. These underlying fears and official health responses, combined with the lack of knowledge and information about COVID-19 in the community, are considered by our participants to have contributed to the stigmatisation and exclusion of people affected by this new virus. Don't forget that we came out of Ebola. There is this experience that is in some minds, COVID-19 is not that different from Ebola, the measures taken during Ebola and COVID-19 almost corroborate… The gradual dissemination of scientific data has led to a better understanding of the modes of transmission of the virus and of the means to prevent contamination, which in turn has led to a change in behaviour and attitudes towards sick and cured people-at least by part of the population. At the beginning it was a big problem. If it was known that someone had COVID in a family, no one went there, people avoided greeting them, neighbors avoided them. There was a stigma that was very, very serious. Over time, people have understood that we can come and go without problem, we can greet each other and visit each other without being contaminated by respecting the barrier measures. In addition, the growing perception that COVID-19 is not a severe and lethal disease has greatly reduced the fear of contamination, thereby reducing the stigma towards the sick, the cured and their families. The reluctance [ostracism towards COVID+ people] has decreased. Many are informed about the disease, it's not a fatality, you can have the disease and recover from it. However, despite interviewees' testimonies suggesting that stigmatisation has decreased since the beginning of the epidemic, the data seem to show that people are still discreet about their COVID+ status. It is therefore unclear whether the stigma associated with COVID-19 has really dissipated in Conakry. Distrust and resistance to COVID-19 in a complex historical, political, economic, sociocultural and health context The emergence of COVID-19 in Conakry has taken place in a context of successive historical events and overlapping contemporary events, which has influenced community experiences, reactions, behaviours, attitudes and perspectives on the epidemic, 21 as evidenced in this study. Participants reported that COVID-19 is perceived by some community members as a conspiracy orchestrated by the Guinean government, rather than an actual disease. The occurrence of COVID-19 in March 2020 22 coincided with sociopolitical movements and protests in Guinea in response to the presidential desire to modify the Guinean constitution. 23 Concomitantly, following the example of other countries, the Guinean government decreed a ban on gatherings in March 2020, as a means to contain the COVID-19 pandemic. This led a portion of the Guinean population to express significant mistrust about the existence of COVID-19, as well as distrust of state leaders. 21 As a side note, public dissatisfaction with the president let to the coup d'état of 5 September 2021, during which the president was arrested, the constitution suspended and the government dissolved. 24 The distrust in the government apparatus is not new in Guinea. The country's heavy historical-political past has, among others, influenced how communities reacted to the Ebola epidemic in 2014-2016, including acting violently towards Ebola response humanitarians. 3 25 This phenomenon of suspected conspiracy regarding COVID-19 is not unique to Guinea, as conspiracy theories have widely circulated in social media around the world. 26 Popular distrust is often a reaction to so-called 'top-down' approaches, especially when governments impose strict rules to contain the spread of an epidemic. 25 27 The mistrust of Guineans towards 'the Whites', mentioned by some participants, is not a new social phenomenon either. 25 Furthermore, because our participants reported a perception that the COVID-19 epidemic spread in the West before reaching the African continent, the disease might be seen as a danger to the survival of Guinean communities that came from outside. This could potentially foster prejudice against foreigners as they are perceived to be the importers of the virus 28 hence explaining the representation of COVID-19 as a 'white man's disease' and blame towards Westerners. Moreover, COVID-19 does not seem to be the highest priority in the minds of Guineans in terms of health risks. According to our data, it seems that several Guineans consider that they are not concerned by the serious nature of the disease. Guinea also had to deal with a resurgence of Ebola in early 2021, as well as epidemics caused by Lassa fever, measles, meningitis and yellow fever. 27 Malaria, which is ubiquitous and the leading cause of mortality in Guinea, causes more deaths annually than COVID-19 has caused so far nationwide since the beginning of the epidemic in March 2020 . While the prevalence of COVID-19 is low in Guinea , seroprevalence studies indicate that at least 25% of West African population have developed anti-SARS-CoV-2 antibodies, 30 which suggests that additional undetected cases would have been asymptomatic or paucisymptomatic, contributing to the 'invisibility' of the disease or to the perception that COVID-19 is a 'bad cold'. It is therefore not surprising in this context that a part of the Guinean population has trivialised the COVID-19 epidemic, relaxed barrier gestures and complained about social distancing measures such as the ban on gatherings, or even denied the existence of the disease. COVID-19 stigmatisation and the sociocultural context of Conakry: a collectivist society that stigmatises to protect the community The phenomenon of stigmatising patients with COVID-19 was widely reported by study participants, and was also previously observed around the world. 31 Stigmatisation appears to be particularly common and important in Conakry as per our data-because of the apparent persistence of COVID+ individuals wanting to be discreet about their status-as well as in Guinea overall 17 32 and more broadly in sub-Saharan Africa and low and middleincome countries. 17 The fight against stigmatisation of patients with COVID-19 has become a priority for the authorities of many African countries, such as Senegal and Kenya, as well as of international organisations such as WHO and UNICEF. [33][34][35] The concern of the Guinean government and health authorities to protect the population from stigmatisation was reported by several participants in our study. According to participants, stigma and ostracism associated with COVID-19 appear to be of limited duration, and thus would be a 'social interruption'-during the infectious phase and sometimes longer after recoveryrather than a 'social death'. 36 Stigma and avoidance of the person with an infectious disease are usually temporary, ceasing when the person is cured . 37 Stigmatisation of people with an emerging infectious disease is not specific to the current COVID-19 pandemic. Fear of Ebola has given rise to stigma, exclusion, humiliation and violence against many of those who have recovered. [38][39][40] According to Goffman, 41 42 stigma occurs when individuals are significantly different and this difference is unwanted by other members of their social context. 43 Because of the dangers that these persons represent, they may be subject to negative attitudes, judgements and various forms of discrimination such as ostracism-which limit their life opportunities and social relationships. 42 43 In Conakry, stigma seems a priori to serve a purpose of protection against contamination Open access by sidelining COVID+ people . Avoidance and social exclusion would therefore be adaptive responses serving to prevent the spread of the virus and to protect uninfected people from the disease. 28 However, this social avoidance creates tension for individuals living in collectivist societies like Guinea. The need for social ties is indeed particularly important in these interdependent societies, 44 especially in the face of illness, 37 as affected individuals expect to receive the solidarity support of their close ones. 45 In addition to the suffering of being socially isolated, hurtful social attitudes and behaviours add to the burden of those who are stigmatised. 35 46 47 Participants in our study spoke of shame and moral suffering experienced by those who are stigmatised and ostracised by other members of their social group. People living in collectivist cultures such as in Guinea indeed seem to be particularly sensitive to the reactions of people in their social environment 48 and thus are more vulnerable to criticism, negative looks leading to a strong sense of shame. 49 Distance/ deprivation from daily social interactions and lack of social support can therefore greatly affect the moral and psychological well-being of stigmatised and ostracised people. 42 50 The prevention of stigmatisation and its deleterious consequences for individuals should be anticipated in future epidemics. Providing moral support to contagious people while preventing the transmission of the virus, assessing the mental health of victims of discrimination and facilitating their access to psychological resources should also be key measures of pandemic responses. In addition to the actual traumatic experiences of discrimination and social exclusion, the anticipation of stigmatisation, that is, the fear of being stigmatised, 31 may lead individuals to modify their actions to avoid stigmatisation-such as the phenomenon of discretion mentioned by the participants. Discretion is a strategy used in other contexts in Conakry to avoid family and social sanctions in the event of non-compliance with certain sociocultural norms. 51 52 Furthermore, the strategy of concealment or lying is a 'ruse' to avoid stigmatisation, a phenomenon previously observed in Conakry in another context. 51 Circumventions to avoid disclosing the reality at all costs are strong evidence that the collectivist culture prevailing in Conakry imposes heavy burden on people who are not considered as being 'normal', and that the social price to pay is too high for many. In particular, they may seek to avoid being tested for COVID-19, 31 thereby posing a public health risk due to increased viral transmission. 28 31 35 The anticipation of stigmatisation can also have significant consequences on the physical health of the infected persons by increasing the risk of disease worsening if they delay seeking clinical consultation with health services. 28 31 35 Additionally, according to our data, individual discretion on COVID-19 status and public non-dissemination of local COVID-19 statistics as a strategy to avoid stigmatisation appear to have had a counterproductive consequence in making COVID-19 'invisible'. By possibly impacting the convictions of many people as to the reality of the epidemic, this might reduce people's willingness to comply with barrier gestures, to seek testing and treatment and their intention to participate in clinical trials. Anticipation and information are known to be the most effective responses to prevent disease denial and fear in the event of epidemics. 3 Our study highlights that avoiding misconceptions and preventing stigmatisation of infected people may rely on a clear, transparent and non-discriminatory communication about the disease and the anxiety-inducing potential that such situations can generate. 28 31 We also suggest that health response strategies include the psychosocial dimension of epidemics, including the risk of stigmatisation and its deleterious consequences, and the ways to avoid it. 28 31 Some participants pointed out a link between low levels of education and beliefs. The lack of education indeed limits the ability to be informed and to look critically at the information and rumours that circulate. 53 Stigmatisation is thus more common among subpopulations characterised by lower education levels. 17 Our data support the recommendations that community sensitisationespecially through face-to-face meetings with trusted individuals-is a strategy to be advocated in settings with low literacy rates, 3 25 such as Guinea. 54 Finally, more research should be carried out within collectivist societies to better understand and document the phenomenon of stigma, its consequences on mental health, and to highlight the best solutions for prevention and intervention. This recommendation is applicable both to the current COVID-19 pandemic and as a preventive measure for future infectious disease outbreaks. Preventing stigma associated with an emerging infectious disease would also contribute to promoting the acceptability of participation in clinical trials, such as the Coverage-Africa study. --- Strengths and limitations of the study This research is the first to explore community experiences, opinions and perceptions related to COVID-19 in Guinea. A key strength of our study is that it was designed and implemented by an international multidisciplinary team both from France and Guinea potentiating complementary strengths: the first suggested the interview questions and provided a more 'external' analytical approach to explain the sociocultural phenomena found in the data; and the latter provided access to the field, enhanced the richness of data collected and validated the sociocultural interpretation of the data through their in-depth knowledge of the local culture. Furthermore, in addition to deductive coding, we performed inductive coding of the data to allow for the emergence of unanticipated themes. Our study had some limitations. We faced recruitment difficulties, in particular due to the lack of motivation because participation was not financially compensated , logistical difficulty in going to the field to collect written consent Open access during the rainy season and confinement in the context of sociopolitical tensions and the coup d'état. However, this limitation does not seem to have biased the results since data saturation was likely reached, showing that our sample size was sufficient to comprehensively explore the perceptions of the members of the target communities. Additionally, despite efforts to diversify the profiles of potential participants, interviewees were mainly male and highly educated : we were therefore less able to explore in greater depth the perspectives of women and of people with little or no education. --- CONCLUSIONS Community viewpoints about COVID-19 in Conakry are heterogenous. Denial of the disease and negative perceptions about COVID-19 can be partly explained by the Guinean history of Ebola epidemics and sociopolitical tensions. Stigmatisation of people with COVID-19 could be intended to protect against contamination. However, social avoidance can greatly affect the morale of stigmatised individuals, especially in collectivist cultures like Guinea. Additional research on stigma-including its impact on individuals' health-seeking behaviours, and consequences on mental health among those affected/ exposed-would contribute to: identifying improved prevention and care interventions in anticipation of future health threats and promoting the acceptability of participation in clinical trials, such as ANRS COV33 Coverage-Africa. --- Disclaimer The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. --- 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 This study involves human participants and was approved by the National Health Research Ethics Committee of Guinea . Participants gave informed consent to participate in the study before taking part. Provenance and peer review Not commissioned; externally peer reviewed. Data availability statement Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information. The data sets generated and analysed during the current study are not publicly available. Data may be made available upon reasonable request in compliance with the sponsors' requirements, as stated in the informed consent form signed by the participants. However, sufficient verbatim extracts are presented in the paper to illustrate the results. 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.

Objectives To explore communities' perceptions about COVID-19 in the context of the ANRS COV33 Coverage-Africa clinical trial evaluating the efficacy of treatments in preventing clinical worsening of COVID-19. Design Descriptive qualitative study using semistructured in-depth individual interviews conducted by telephone in French and Soussou between May and September 2021. Data were transcribed, translated in French when applicable and analysed with the thematic analysis method. Setting The eight neighbourhoods most affected by COVID-19 in Conakry's urban context, capital of Guinea. Participants 4 community leaders acting as key informants-providing insights regarding population's opinions-and six community members, who were exposed to an information session conducted as part of Coverage-Africa. Results According to participants, community members have heterogeneous viewpoints about COVID-19: it exists and is dangerous; it is benign ('bad cold'); or it is fictitious (eg, government conspiracy). The fear of stigmatisation and social isolation of those sick or cured of COVID-19 was largely reported by participants, with illustrations of distressing situations for the victims. To avoid stigma, many patients seem to adopt strategies of discretion (eg, lying/hiding about the disease). Although community attitudes were reported to have evolved since the beginning of the epidemic, stigma remained a pervasive concern for many people. Conclusions Community perceptions about COVID-19 in Conakry may be partly explained by the Guinean context of Ebola history and of sociopolitical tensions. Stigmatisation of COVID+ people seems to be aimed at protecting others against contamination. However, social avoidance can greatly affect the morale of stigmatised people, especially in collectivist cultures like Guinea. Further investigating stigma, including its role on seeking COVID-19 screening and treatment services, and its consequences on mental health among affected/exposed people, would contribute to identifying improved prevention and care interventions in preparation for future health threats, and to promoting participation in health research.⇒ This research is the first to explore community experiences, opinions and perceptions related to COVID-19 in Guinea. A key strength of this study is that it was designed and implemented by an international multidisciplinary team both from France (outsiders) and Guinea (insiders), potentiating complementary strengths to enrich data collection and interpretation. ⇒ We included deductive and inductive coding, allowing for the emergence of unanticipated themes. ⇒ Saturation of themes was likely achieved, showing that our sample size was sufficient to comprehensively explore the perceptions of the members of the target communities. ⇒ We were not able to recruit as many participants as planned due to, among other things, logistical difficulties. ⇒ Despite efforts to diversify the profiles of potential participants, interviewees were mainly male and highly educated: we were therefore less able to explore in greater depth the perspectives of women and of people with little or no education.

Introduction and background The integrated human immunodeficiency virus prevention and sexual and reproductive health services deliver a coordinated response to the dual challenges of high HIV prevalence and unmet needs for family planning. It widens the reach of HIV prevention and family planning messages to all communities . Integration involves combining HIV risk assessment, HIV testing and counselling, treatment and care, and SRH services like family planning, maternal and child health, and management of sexually transmitted infections within the health sector to offer comprehensive health services at one service delivery point or through a strong referral system . Integrated services expand access to and coverage of critical services and improve the efficiency of such services by reducing duplication of service delivery functions while delivering more services per client contact . Nurses are in the forefront of service provision for HIV prevention and family planning in many African countries. Some of these countries, including Malawi, where this study was conducted, have adopted and implemented the integration of HIV prevention and SRH services because it is considered as a promising practice to address unmet needs for contraception and to reduce mother-to-child transmission of HIV . Unfortunately, not much has been recorded on the nurses' actual practice of integration of HIV prevention and SRH services in the healthcare human resource-constrained African countries, like Malawi. --- Nurses' practice of integration of HIV prevention and sexual and reproductive health services in Ntcheu District, Malawi Read online: Scan this QR code with your smart phone or mobile device to read online. http://www.curationis.org.za Open Access in 2010 and 2011 in Malawi, to determine whether patients accessing HIV services were also able to access family planning services either on site or through referral mechanism identified gaps and areas of improvement needed in that country. These included improving coordination between Ministry of Health's Reproductive Health Directory and HIV and AIDS department, training providers to provide integrated services and using task shifting to expand access to services . Later, the 2015 policy review on family planning and HIV integration in Malawi identified various types of HIV services, such as counselling on HIV prevention, voluntary counselling and testing and dispensing antiretrovirals , to be integrated into existing family planning services . The Reproductive Health Service Delivery Guidelines of Malawi supports the need for provider-initiated counselling and testing among family planning patients, who do not know their HIV status. The guidelines highlight that family planning providers, including nurses, should be trained in counselling and testing for HIV, and that there should be proper referral of patients to pre-antiretroviral therapy or antiretroviral therapy clinics, while HIV-negative women should be counselled on risk reduction and dual protection. The guidelines also state that a clear referral mechanism should be established when certain HIV services cannot be provided at the same site as family planning in that country . In response to the Malawi National Policy on Family Planning and the Malawi HIV integration and Sexual and Reproductive Health Service Delivery Guidelines, it became necessary to determine and describe the nurses' practice of the integration of HIV prevention and SRH services as a strategy to effectively combat the spread of HIV and promote family planning in that country. --- Research methods A descriptive qualitative case study was conducted in one family planning clinic within the secondary level hospital in the district health office of Ntcheu, Malawi. The research question was: How do nurses practise the integration of family planning and HIV prevention services in Ntcheu District, Malawi? A modified conceptual framework for assessing facilitators and barriers to knowledge use by Cabana et al. ) was used to guide the study. A descriptive qualitative case study design was found to be the most suitable design as it has already been used in organisational studies . The design allowed the researchers to gain an in-depth understanding of the nurses' practice of integration of HIV prevention and SRH services as a 'contemporary phenomenon' within a 'real-life context' of the selected secondary hospital in Ntcheu District, Malawi . The single unit of study was the nurses' provision of integrated family planning and HIV prevention services, and the context was the practice of family planning and HIV prevention care. The family planning clinic within the secondary hospital and timeframe for data collection formed the boundaries of the case study. The features of the case that were examined were knowledge of and attitudes and perceived barriers to practising integration of family planning and HIV prevention services as three domains of the modified conceptual framework of Cabana et al. ). --- Population and sampling The targeted population were all nurses providing SRH services and HIV services at the Ntcheu District Hospital during the time of data collection from February to May 2016. Purposive sampling, based on the study objectives and the research question, was used to select and recruit the participants who met the criteria to be included in the study and could provide rich information required to answer the research question. As defined by Guest, Bounce and Johnson , a sample of 10 participants, including the two participants from the pilot, was reached through attaining data saturation. This small sample size is accepted in qualitative research and facilitates close relationship between the researcher and the participants, thus enhancing the validity of the findings . --- Data collection tool A semi-structured interview guide was developed by the researchers in line with the research objectives and the concepts of the modified conceptual framework for assessing facilitators and barriers to knowledge use by Cabana et al. that was used to guide the study. The pilot study was conducted in the last week of February and first week of March 2016, using two participants who were purposively selected and met the inclusion criteria of the study. As recommended by Van Teijlingen and Hundley , the pilot study was conducted to allow for the refinement of the semi-structured interview guide if necessary. However, data analysis of pilot data showed that it was not necessary to refine the semi-structured interview guide. This was confirmed by the senior author. The data from the pilot study were added to the data of the main study because the process of data collection was the same in both the pilot and main study. --- Data collection and data management Data collection and preliminary data analysis were conducted between March and May 2016 when the participants were off duty. Individual interviews were conducted with all participants individually to get first-hand information. Observations of non-verbal communication during the interviews were noted as field notes. English was used to conduct interviews because it is the only official language used in the healthcare system in Malawi. Interviews were audio-recorded and transcribed within 24 h, while the information was still fresh in the mind of the researcher. Each transcribed interview was sent to the senior researcher immediately for guidance in determining how collected data were in line with the research objectives. Preliminary data analysis was conducted parallel to data collection in the same manner as described in the 'Data analysis' section. This was to enhance sampling until data saturation was attained. --- Data analysis Formal data analysis commenced soon after completion of data collection and preliminary data analysis which was conducted during the process of sampling and data collection to determine data saturation. Manual data analysis , using the five steps for interpretive content analysis as outlined by Terre Blanche, Durrheim and Painter , was used to analyse the transcribed data. Firstly, the transcribed interviews and field notes were read many times to gain in-depth familiarity with the collected data. Notes on identified meanings were made reflectively where appropriate in the margins of the sheet containing the data. This was followed by coding which involved underlining phrases and sentences that were found to be relevant to the research question and objectives in the transcripts from all participants, starting from the first to the last transcript. Different colours were used to represent each identified statement. Each colour was given a label to ensure identification of which code belonged to which statement within each participant's transcript. Numbering of the individual codes was done for purposes of storage and retrieval for within-case and across-case analysis, resulting in 42 identified codes. As proposed by Miles and Huberman , a second qualified and experienced qualitative researcher was employed as an inter-coder. The codes' applicability was tested on the two transcripts from the pilot study. The codes from both the researcher and the inter-coder were further verified and confirmed against raw data by the senior researcher who is also the second author. The identified codes were grouped to generate categories based on their similarity in meaning. The categories were then grouped to form themes, based on similarity. The whole process was submitted to the senior researcher for verification. Individual themes were further interpreted for meaning to construct an exhaustive, in-depth understanding of the nurses' practice of the integration of family planning and HIV prevention services using each objective of the study and the research questions to construct an inferred meaning of the phenomenon studied within the context of the secondary level hospital in Ntcheu District, Malawi. This extensive data analysis process yielded five themes. Member checking, which refers to verification of findings by the participants , was conducted with the seven participants who were still available at the research setting. They all confirmed that the five themes reflected what they believed to be the nurses' practice of integrating family planning and HIV services in the secondary level hospital in Ntcheu District, Malawi. --- Trustworthiness of the study The four basic frameworks for ensuring rigour, which is quality, authenticity and truthfulness of findings, were adhered to throughout the study. This includes transferability, credibility, confirmability and dependability . Among other things, member checking was done by going back to participants with the findings to check and verify the interpretation of the researchers . It was found that the interpreted data reflected what the participants believe is the nurses' practice of integrating HIV prevention and SRH services in the secondary level hospital in Ntcheu District, Malawi. Intercoding agreement between the researcher and inter-coder was reached and further confirmed by a senior researcher. This ensured that the codes were a true reflection of data from participants. The complete set of analysed data was further checked, and the authenticity of identified themes was confirmed by the senior researcher who is also the senior and second author of this article. --- Ethical considerations Ethical clearance was obtained from the relevant authorities. All ethical principles for conducting research with human subjects as described in the Declaration of Helsinki of 2013 were observed throughout the study . Each participant signed and provided a written informed consent form to participate in the study. --- Results Of the total of 10 participants, including the two participants from the pilot study, two were male and eight were female nurses who have worked in the family planning clinic, a research setting, for a period of between 2 and 8 years as qualified and licensed nurse practitioners. All participants reported having received in-service training on the integration of family planning and HIV services. The age of the participants ranged between 25 and 49 years. Five themes were identified as facilitation of access and acceptability of comprehensive HIV prevention and family planning services, educating and counselling patients, early detection of HIV among women of child-bearing age, professional benefits of integrating family planning and HIV prevention services and resentment of integration of family planning and HIV prevention services. --- Facilitation of access and acceptability of comprehensive HIV prevention and family planning services This theme was derived from data related to nurses' practice of facilitating the access and acceptability of comprehensive http://www.curationis.org.za Open Access HIV prevention and family planning services as captured in the following extracts from the data: 'We ensure that every woman of child-bearing age who comes to our family planning clinic has access to family planning services as well as HIV prevention services at any time and any day … we also ensure that these women have access to the family planning method they want at any time.' The following extract demonstrates nurses' familiarity with facilitating access and accessibility of comprehensive integration of family planning and HIV prevention services by helping women to select the most appropriate family planning method. 'But there are some family planning methods which not every woman was familiar with previously but they are now able to access even when they tested HIV negative. So we do family planning and HIV counselling daily to ensure that the HIVnegative patients are given the right choice of family planning method.' The theme was further supported by the following quote from data: 'With the coming of integration of family planning and HIV prevention services, nowadays we ensure that all women who come for family planning services have access of both family planning and HIV counselling and testing services on a daily basis.' --- Educating and counselling patients This theme emerged from data that related activities carried out by nurses to educate and counsel patients as part of integration of HIV prevention and reproductive health services, such as giving motivational talks on the importance of getting tested for HIV, teaching patients about the advantages and disadvantages of all family planning methods, helping patients to choose the best family planning methods according to their HIV status and continuous counselling on positive living. The following extracts from the data highlighted the theme 'Educating and counselling patients' during integration of family planning services with HIV services: 'We give a general health talk and later a motivation talk on the importance of getting tested for HIV and spacing your children well.' Educating and counselling patients about family planning and HIV prevention services was further captured through the following extract from data: 'Each and everyone who comes for family planning is counselled and tested for HIV and given a motivation talk on the importance of getting tested for HIV. That's what we do that all the time.' Another extract expressed similar sentiments as follows: 'When these women come for the family planning clinic we give them a health talk on all the available family planning methods, their advantages and disadvantages and then later a motivational talk on the importance of getting tested for HIV, and this is our routine.' Another quote supporting the development of the theme 'educating and counselling patients' on integration of family planning and HIV prevention services was as follows: 'At times we come across women who insist on using implants even when they have tested HIV positive and we have explained about the disadvantages of implants to HIV positive persons. We try to explain again and again until they can choose a method appropriate for them, we don't just force them.' --- Early detection of HIV among women of child-bearing age This theme emerged from data that related to nurses' activities related to early detection of HIV among women of child-bearing age, including retesting for HIV after every 3 months, screening for chronic conditions including HIV and knowing HIV status of all family planning patients, as shown in the following extracts from the data: 'When someone tests negative for HIV, we congratulate them, but we tell them that even if they have been found negative, that does not mean they cannot get infected, but to continue preventing the virus but also getting retested after every three months. We insist on this because it may help with early detection of HIV among women of child-bearing age.' 'When a family planning client comes in, we weigh her, check blood pressure and test for HIV. She also gets counselled and tested for HIV before she receives her method of choice for family planning. In many instances, from doing this, we are able to detect other illnesses early.' 'Nowadays all patients who come for family planning undergo HIV counselling and testing before they receive a method of their choice. This helps with early detection of HIV for early treatment.' Early detection of HIV among women of child-bearing age women was further captured from the following extract: 'With the coming of integration of family planning and HIV prevention services we are able to detect HIV among our patients at an early stage through screening when they are not yet sick, and this makes work easier.' --- Professional benefits of integrating family planning and HIV prevention services This theme was derived from data that related to professional benefits of nurses' practice of the integration of family planning and HIV prevention services, as captured in the following extracts from the data: 'If you identify positive woman who comes for family planning, and as a provider you discover that she is also breastfeeding, but was missed for HIV counselling and testing during antenatal care, with this approach, you are able to assist her accordingly, without having to refer her to someone else.' http://www.curationis.org.za Open Access Another extract from data highlighted the benefit of picking up on missed patients as a benefit of family planning and HIV service integration, as shown in the following extracts from data: 'For example let's say a mother was missed on HIV counselling and testing during antenatal care and comes to family planning clinic with her breastfeeding baby, gets tested and is found HIV positive … now with integration it means this mother will be put on ART and baby on Bactrim.' The following quotes were extracts that were also identified as benefits of integrating family planning and HIV services by nurses in Malawi: 'Women who are missed during antenatal care or were on window period, we are able to capture them during family planning clinics … this protects their babies as well as the whole family because they will have a healthy life now.' 'Those mothers missed during antenatal are being captured now with this integration thereby preventing mother to child transmission.' The following extracts highlighted the reduction of HIVrelated deaths as a benefit of integration of family planning and HIV services by nurses in Malawi as follows: 'With integration of family planning and HIV prevention we are preventing or reducing HIV-related deaths because patients are being identified at an early stage. That makes me happy knowing that through my practice, lives have been saved.' 'Integration of family planning and HIV services is a good thing because we are able to identify HIV-positive patients at an early stage when they are not sick, and this will reduce HIV-related deaths in our community.' 'If these women are accessing both family planning and HIV prevention services, this motivates me to do more for my people.' Another professional benefit of integration, that of reduction of workload, was captured in the following extracts from the data: 'But there is also reduced workload. Previously without integration of family planning and HIV prevention of services we were having family planning only three times a week, and during these family planning days you could provide services to 50 patients. But now with integration the services are being provided on daily basis; we only see about 15 patients or less than that a day, so to me this is motivating. That is good for all of us.' 'Integration of family planning and HIV prevention of services seems to reduce workload at times because we are providing all the services on a daily basis … patients come in small groups than before integration, where you could have a big group of patients on a day since family planning clinics were being conducted twice a week.' --- Resentment of integration of family planning and HIV prevention services This theme was derived from data that related to the participants' resentment or hatred of integration of family planning and HIV prevention services, which are captured in the following extracts from the data: 'As for me, I see integration as a very good thing even though it seems like it's too involving and needs more time to be spent with one patient. In that regard I think it is not as good as it was made to be when introduced to us. --- Discussion The findings revealed that nurses' practice of integration of family planning and HIV prevention services enabled them to facilitate access to and acceptability of HIV prevention and SRH services among patients. The practice was enhanced by the training in both family planning and HIV management, which the participants had received. Similar findings have been reported in many studies globally, where the integration of family planning and HIV prevention services has been acknowledged as a strategy to increase access to and acceptability of comprehensive integration of family planning and HIV prevention services for patients . The negative effects of limited or no training of nurses on integration of HIV prevention and SRH services were reported in some African countries, including Tanzania, Ethiopia, Kenya, Rwanda and South Africa . Clearly, training of nurses on the integration of HIV management and family planning are key to improving the nurses' practice of integration of HIV prevention and SRH services and facilitation of access to comprehensive integrated HIV prevention and SRH. The WHO recommends that health providers should facilitate access to HIV counselling and testing to all women of child-bearing age. This not only facilitates access to HIV counselling and testing for all women of child-bearing age, but also maximises the health and well-being of individuals through the timely detection of HIV, prevention of HIV transmission and subsequent access to appropriate HIV prevention and treatment . The UNPFA's recommendations support the nurses' practice of educating and counselling patients of family planning and HIV prevention services as an important aspect in the integration of HIV prevention and SRH services. It further recommendations that health providers must share complete information regarding contraceptive method effectiveness as well as the advantages and disadvantages of each method with their patients . The importance of the findings on nurses' practice of early detection of HIV is further highlighted by other authors. For instance, Ramfolo et al. suggest that knowing the patients' HIV status has benefits for both the client concerned and the provider. For HIV-negative people knowing their status empowers them to protect themselves from becoming infected and to remain negative. For HIV-positive people knowing their status ensures that they can be provided with appropriate treatment, care and support services and be assisted in living positively . The reported professional benefits of implementing the integration of HIV prevention and SRH services, including patients coming in small manageable numbers, were reported to help nurses to feel good about helping their patients and to increase the trust between nurses and their patients, and are well supported in the literature. For instance, Nielsen-Bobbit et al. reported that a reduced number of multiple return visits or referrals were also reported in Tanzania, where health providers noted that integration of family planning and HIV services resulted in an increase in client uptake of services and reduced need for multiple return visits or referrals. Similar findings on benefits of integrating family planning and HIV prevention services were reported in a longitudinal study in Haiti, where Peck, Fitzgerald and Liataud found that integrating additional services including family planning within HIV prevention programmes increased the uptake of HIV testing by 62 times over a 15-year period. In Zimbabwe, it was reported that offering integrated services through community health workers contributed to increases in contraceptive use and referrals to VCT centres and led to improvements in client attitudes and knowledge about both family planning and HIV prevention . In Kenya and the Kingdom of eSwatini, it was found that client satisfaction improved because of the increased efficiency of services, which, in turn, had a positive effect on health workers' own satisfaction . This study also revealed resentment of integration of family planning and HIV prevention services among some nurses. For instance, it was reported that the practice of integration of family planning and HIV prevention is too involving and takes more time, thus increasing the workload and resulting in exhaustion. Resentment of integration of family planning into VCT for HIV as burdensome, tiresome and timeconsuming has been reported by others . --- Limitations of the study Only eight participants were reached for member checking. The two other participants were on vacation. Waiting for their return from vacation would have delayed the conclusion and reporting of the findings. It is common in member checking with participants who work in one institution and have to take vacation at different times. --- Recommendations The nurses' practice of the integration of HIV prevention and SRH should be promoted as a strategy to combat the spread of HIV and promote family planning in other resource-constrained countries. --- Conclusion This study revealed the important aspects of nurses' practice in the integration of HIV prevention and SRH services to combat the spread of HIV and promote family planning in Malawi. --- Data availability statement Data sharing is not applicable to this article as no new data were created or analysed in this study. --- Competing interests The authors declare that they have no financial or personal relationships that may have inappropriately influenced them in writing this article. --- --- Disclaimer The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of any affiliated agency of the authors.

A rapid assessment conducted by the Centre for Reproductive Health, in collaboration with the International Planned Parenthood Federation (IPPF), United Nations Population Fund Activities Background: Nurses play a critical role in their practice of integrating human immunodeficiency virus (HIV) prevention and sexual and reproductive services to combat the spread of HIV and promote family planning in resource-constrained countries like Malawi.The purpose of this study was to determine and describe the nurses' practice of integration of HIV prevention and sexual and reproductive health (SRH) services as a strategy to effectively combat the spread of HIV and promote family planning in Malawi. Methods: A descriptive qualitative case study was used. The research question was: How do nurses practise the integration of family planning and HIV prevention services in Ntcheu District, Malawi? Qualitative data were collected using semi-structured interviews from a sample of 10 participants. Manual data analysis, using the five steps for interpretive content analysis, was used to analyse data. Results: Five themes were identified as (1) facilitation of access and acceptability of comprehensive HIV and family planning services, (2) educating and counselling patients, (3) early detection of HIV among women of child-bearing age, (4) professional benefits of integrating family planning and HIV prevention services and (5) resentment of integration of family planning and HIV prevention services.The nurses' practice of integration of HIV prevention and SRH services has more benefits for both nurses and patients as a strategy to combat the spread of HIV and promote family planning in a resource-constrained country like Malawi.

Introduction We live in a world of scarcity. Resources, including goods and services, are limited. The restricted nature of goods and services, in turn, leads to higher prices and lowered access . This is a reality human societies have lived with for many centuries. Another historical feature of human societies is inequality . Scarce resources have always been distributed unequally to some extent-at various times and in various societies more or less unequally than others. By some accounts, the United States today is experiencing some of the highest levels of inequality seen since at least the beginning of the twentieth century, and perhaps beyond. Indeed, according to one economist, the level of inequality in the United States today is "probably higher than in any other society at any time in the past, anywhere in the world" . These two human realities-scarcity and inequality-in many ways go hand in hand. Scarcity may be one source of inequality-if there is not enough of a resource to go around, some will inevitably end up benefitting more than others. Yet technological advance provides the potential for reducing or even eliminating at least some forms of scarcity. For example, the Internet has ushered in a new era of abundance for informational and creative content by reducing the price of reproduction and distribution of this content to near-zero . Other technologies offer the same promise for physical goods and services. 3D printing, for instance, will almost certainly make abundant a wide variety of physical goods as the technology advances and the price of producing and distributing these goods drops. And robotics technologies may do the same for services as robots become increasingly able to perform, cheaply and effectively, the bulk of services currently performed by humans . If this projection of reduced scarcity in the realm of goods and services is correct, what will this mean for inequality? When scarcity is mitigated, will inequality be tempered as well? It is tempting to think so. After all, if the newfound abundance of goods and services means that these become readily available to most members of society at low cost, at least one potential source of inequality goes away. And with that departure, ideally, the harms that attend unequal societies-harms ranging from slowed economic growth to adverse health and psychological effects on the society's citizens -would be mitigated as well. But this hope is, unfortunately, unlikely to be realized-at least without significant planning and intervention. In the second Part of this chapter, I explain why. First, history belies the assumption that inequality necessarily disappears under conditions of abundance. Different societies at different times have enjoyed relative periods of abundance, and yet inequality has persisted . Scholars have understood this historical truth as confirming the hypothesis that inequality is as much about politics and power as it is about physical limits . Second, with respect to the many harms that attend unequal societies, these harms may not have as much to do with access to goods and services as they do with psychological factors. For example, an increasingly accepted hypothesis put forward to explain the correlation between inequality and the raft of social harms that attends it posits that these harms arise from the adverse psychological consequences of living in an unequal society . These consequences affect not only those at the bottom of the wealth and income ladder, but also those at the top. They arise from inequality itself rather than any objective measure of poverty or standard of living . Third, somewhat paradoxically, certain conditions of abundance may bring with them their own psychological and societal harms-especially if it is not clear to all that the abundance is being fairly distributed. Reminders of abundance, for instance, might increase psychological distress related to concerns about fairness in ways similar to actual inequality . The implications of these insights about abundance and inequality are many, and in the third Part of this chapter I focus on two. First, given the political nature of inequality, we cannot automatically assume that a new abundance of inexpensive goods and services enabled by technology will necessarily translate into widespread access to this abundance. Structural inequalities can lead to bottlenecks that prevent the disadvantaged from accessing even those goods and services that in theory should be within their reach . Second, even if all members of society can access newly abundant goods and services equally, to the extent that other forms of inequality remain, many problems that plague unequal societies like the U.S. will persist. These problems grow from the psychological effects of inequality rather than any ability to access particular goods and services, even if these goods and services have real welfare-enhancing effects. Given these implications, I ultimately conclude that scholars and policymakers must consider how to address the structural and political barriers that might prevent widespread access to the influx of inexpensive goods and services that will improve people's lives. But, more than this, if society wishes to tackle inequality and the social problems that come with it, it must consciously dismantle inequality in all its forms. --- Inequality: Definitions and harms Scholars have predicted a technological revolution that will change our experience of scarcity. Advances in technologies like 3D printing, robotics, and synthetic biology will lead to a new infusion in the market of low-cost goods and services-much like the Internet has done for information and creative content-making these products, in economic terms, abundant rather than scarce . Living in a society rife with inequality-as those of us in the United States do-the question arises as to whether the specific promise of abundance offered by these emerging technologies could positively impact this situation. But before asking if this will happen, we might first ask if we want it to. Relatedly, we should clarify what we mean when we talk about inequality. --- Defining inequality Neither of these are easy questions to answer, and I do not attempt to do so comprehensively in this chapter. As for the latter question-what we mean when we refer to inequality-scholars have taken a range of approaches to address it. For purposes of this chapter I use the term to refer broadly and generally to an unjust distribution of opportunities and resources within a society . Inequality is not necessarily present merely because resources and opportunities are distributed unequally-the justness, or fairness, of the distribution is a relevant consideration under the definition I adopt . Justness is, however, often difficult to gauge, as will be discussed further below . And perceptions of inequality can be equally significant for their ability to give rise to particular social harms as the objective presence of the phenomenon. Further, although absolute levels of distribution are not necessarily determinative of the presence or absence of inequality, the wider and more visible the gap between the haves and have-nots in a particular society, the less likely it will be that distributions are just in fact, and the more likely the gap will be perceived as unjust. Thus, under my definition, inequality manifests when there is an actual or perceived unjust distribution of resources and opportunities, and this is more likely to obtain when the distribution is clearly uneven among citizens or groups. --- Inequality's harms The other question-whether we want to eliminate inequality, and if so, why-is a normative one; as with all normative questions, opinions as to the correct answer can differ. Here, I advance the view that inequality is normatively undesirable, and I offer some reasons to support this position. First, there is the simple fact that the unfairness of inequality feels wrong to many people. Humans have a finely tuned sense of fairness that arises very early in childhood . Subjective perceptions of fairness are rooted in conceptions of human dignity, a value many find compelling and desirable . When the values of dignity and equality are not respected in a society, it causes subjective discomfort among its citizens . More concretely, inequality gives rise to measurable psychological harms. Residents of unequal societies are less happy, exhibit more mistrust and increased anxiety, and have higher rates of depression than those in more egalitarian societies . Perhaps even more concerning than the psychological harms associated with inequality are the social harms that manifest themselves in unequal societies. These harms are wide-ranging, and include, among other things, increased violence, decreased health, reduced life expectancy, higher infant mortality, lower social cohesion, weaker governance, poorer educational attainment, slower economic growth, and lower social mobility as compared to more equal societies . In light of these psychological and social harms, the normative case for eliminating inequality is a strong one. In the next Part, I explore how the expectation of newly abundant goods and services, made possible by technological advances, might contribute to this venture. --- Inequality and abundance In a society of abundance of particular goods and services, one might hope and expect-perhaps for some of the reasons articulated above-that inequality will be mitigated. But will it? In this Part, I address this question. I explain why we should not expect this salutary result without significant policy intervention-and in fact, why without such intervention, the newfound abundance might exacerbate current problems of inequality. I focus first on the political aspects of inequality before addressing the psychological impacts of living in unequal societies-impacts which give rise to a wide range of additional social harms. Each of these frames-the political and the psychological-helps elucidate the limits of abundance of goods and services as a catalyst for eliminating inequality. --- The political nature of inequality One might think of inequality as a problem that arises, at least in part, from scarcity. Definitionally, a scarce resource is one that is incapable of meeting demand . The seemingly inevitable consequence of this is that some will be able to obtain the resource while others will not, leading to inequality. Extending this line of thinking, one might conclude that once a particular resource is no longer scarce, that resource will cease to be a source of inequality. When there is more than enough of a resource to go around, principles of economics dictate that the cost of this resource will approach zero, making it theoretically available to any who desire it . Contrary to this assumption, however, scholars have highlighted the phenomenon of "scarcity amid abundance" . It manifests when some segments of a society experience functional scarcity even when a resource is abundant . Economists, sociologists, and historians have documented this phenomenon in various societies at various historical time points. For example, Amartya Sen has described how famine events have taken place in various societies even when there was enough food to adequately provide for everyone . And Elizabeth Chattergee has explained that India now finds itself in an unfortunate situation with respect to energy, where some groups in the country enjoy a surplus of the resource while others go without . The phenomenon of scarcity amid abundance suggests that even abundant resources might be significant sources of inequality. But why? One potential answer lies in the political nature and power dynamics of inequality. Political and power considerations complicate the market forces that would otherwise lead to widespread access to an abundant resource. In some cases there might be a concerted effort among the powerful in a particular society to withhold resources from those with less power. This effort may arise even in times of abundance because the powerful have a distorted sense of their personal need; or they might fear that the abundance will not endure. For example, David Lamoureux describes how the British in colonial Lagos hoarded land and water for themselves . Or it could be a pure exercise of greed or power . In other cases, the unequal distribution of abundant resources could be a result of simple indifference on the part of those in power; for instance, the lack of will on the part of the British to make the infrastructure investments necessary to provide colonial India with electricity . Another possible explanation for scarcity amid abundance lies in the ongoing impacts of past inequalities. Past wrongs can give rise to ongoing structural inequalities that make it difficult to distribute abundant resources, even when there is a political will to do so. For example, Chattergee explains how the current energy inequalities in India can be traced in part to past colonial rule and Britain's lack of interest in providing India with electricity. Because energy is a resource that requires significant infrastructure investments, the impact of past neglect continues to reverberate today . Scarcity amid abundance is not a phenomenon confined to other places and other times. Examples of inequality in the midst of abundance can be seen in the United States today. The problem of food deserts, for example-where some U.S. communities struggle with nutritional inequality-illustrates the phenomenon and underscores the point that even the most affluent societies can suffer from it . Indeed, even Lemley's example of the Internet making information and creative content abundant demonstrates how the simple market equation of abundance leading to near-zero cost and widespread access can fail to accurately describe the reality on the ground. The fact is that many in the U.S. today are unable to partake in the abundance the Internet offers, perhaps in part due to the political power wielded by major private companies and a lack of will on the part of the government . The political realities of inequality suggest that a resource may not cease to be a significant source of inequality merely because it is theoretically abundant. In Part III I explore what this means going forward as policymakers contemplate how to approach the coming wave of low-cost goods and services. But first, I examine how an understanding of the psychological underpinnings of inequality should also dampen any sanguinity about an abundance of goods and services automatically mitigating inequality. Instead, the challenge for the U.S. is to plan and prepare so that this coming abundance of goods and services can translate into increased welfare and serve as a vehicle for reducing inequality. --- The psychological and social impacts of inequality As detailed above, numerous studies have identified the many social and psychological ills unequal societies face. But to understand whether a new abundance of goods and services will cure inequality and the harms that attend it, it is instructive to understand why these troubles arise in the first place. Scholars are beginning to provide some answers to this question. The emerging picture suggests that the psychological and social difficulties associated with inequality share a causal relationship. Specifically, it appears that the social harms characteristic of unequal societies grow at least in part from the negative psychological influence inequality has on a society's citizens . --- Inequality and mistrust How does this work? As explained, citizens of unequal societies suffer from a number of psychological impacts. One of these impacts is an increased mistrust of both other citizens and the system as a whole . People who live in societies where inequality reigns tend to regard their system of governance as unfair. They are also prone to suspecting that those at the top achieved their favored positions through unethical and dishonest means . This lack of trust leads to more tangible and destructive harms. For example, in an empirical study, Nishi et al. demonstrated how mistrust could impact community formation and ultimately, economic growth. In the experiment, which was set up as a modified public goods game, participants initially allocated a higher share of the wealth cooperated less and acted to preserve their wealth. Those with lower allocations were then forced to choose between being exploited by the "rich" or refuse to cooperate themselves. In games where initial allocations were unequal, mistrust flourished, cooperation faltered, and overall wealth generation was stymied, leading to slowed growth of the game's economy . Other research links the mistrust inequality breeds with the reduced civic participation and ties seen in unequal societies. Those who live in more unequal communities are less likely to participate in social clubs and service organizations, even when they have the ability and resources to do so . Inequality has been shown to be a major driver of this association, with the mistrust inequality engenders explaining the causal relationship . This lack of civic engagement and social support might also help explain the causal relationship between inequality and mistrust on the one hand, and poor health outcomes, including increased mortality and reduced life expectancy, on the other. Frank Elgar, who has studied this phenomenon, hypothesizes that "[s]ocieties with low levels of trust may lack the capacity to create the kind of social supports and connections that promote health and successful aging" . Finally, there is ample evidence that mistrust prompts people to act in unethical and anti-social ways. When people feel that others are getting ahead unfairly, it is easier to rationalize their own unethical behaviors that attempt to even the playing field . Cheating is therefore more prevalent in unequal societies . Inequality also gives rise to higher homicide rates, a causal relationship that is mediated by lowered trust in others . --- Inequality and status competition The ripples moving outward from inequality's stone throw of mistrust are serious and far-reaching. But inequality has additional psychological impacts on the people experiencing it. In addition to mistrust, inequality fosters feelings of envy and jealousy, feelings which, in turn, fuel status competition. Residents of unequal societies place more importance on people's relative positions on the social ladder compared to residents of more equal societies . They also act in ways that demonstrate the weight they place on status, working longer hours to get ahead and conspicuously consuming goods that signal their status to others . Status competition may not at first glance appear to be particularly harmful in and of itself. After all, a motivation to work longer hours might very well improve productivity, and conspicuous consumption might be seen as a relatively harmless activity that also fuels economic growth. However, it appears that status competition generates extreme anxiety about one's relative position in society that leads to all sorts of negative outcomes, including worse health , increased risky behaviors , and higher levels of obesity and drug abuse . One might reasonably assume that these impacts would affect only the "losers" in an unequal society. And to some extent, this is true. Those at the bottom of the status ladder in an unequal society experience at least some of these harms more intensely than those on higher rungs . Remarkably, however, those at the top of the ladder are not immune from the stress of status competition and all the harms, including reduced life satisfaction, that flow from it . The hierarchy-spanning effects of status competition arising from inequality have been demonstrated nicely in the societal microcosm of an airplane. Airplanes, with different levels of seating classes and boarding groups, tend to replicate existing social hierarchies in very visible ways . Payne describes research by Katherine DeCelles and Michael Norton showing how these visible reminders of differential status can affect behavior. After analyzing data from millions of flights, DeCelles and Norton found that so-called "air-rage" incidents, in which a passenger behaved badly or caused a disturbance, were almost four times more likely to occur on flights that had a first-class section compared to flights that did not. The disturbances were almost twice as likely on flights where the economy-class passengers were forced to walk past the seated first-class passengers as they boarded as compared to flights where the economy-class passengers boarded in the middle or back of the plane . As Payne points out, given the cost of an airline ticket, it is unlikely that a typical commercial flight has many truly poor passengers . Yet the status competition triggered by a plane's social hierarchy infected even the relatively welloff airline passengers. Further, the recorded disturbances were not limited to those at the bottom of the airplane's hierarchy. Although economy-class passengers were more likely to cause a disturbance, a significant percentage of incidents were caused by first-class passengers . --- How the psychology of inequality informs expectations for abundance and inequality Understanding the psychological and social impacts of inequality and how they are linked can help scholars evaluate the potential effects of a new influx of low-cost goods and services into an unequal society like the U.S. First, to the extent that these goods and services do make it into the hands of the "have-nots" in our society, they undoubtedly have the potential to improve quality of life. A society that has better access across the wealth and income distribution to a variety of welfare-enhancing goods and services-including medical services-is almost certainly better off than a society in which only a subset of the population at the top of the social hierarchy has access to these things. But whether this new abundance of goods and services will mitigate inequality and the harms that attend it is a different question altogether. The increased access to particular goods and services will certainly mitigate or eliminate one basis on which individuals might distinguish themselves in an unequal society-namely, the differential ability to enjoy these goods and services. But to the extent that other, more fundamental inequalities remain-inequalities in income, income mobility, wealth, access to education and a good job, ability to vote, incarceration rates, and others-a new abundance of goods and services may not do much to remedy the mistrust and status competition that, in turn, give rise to the myriad social problems observed in unequal societies. This is true in part because, as research shows, it is not some objectively low standard of living that causes the mistrust and status competition associated with inequality. Indeed, even the relatively well-off airline passengers in the airplane study were not immune from inequality's psychological sway. As Payne notes in an article discussing his research, even the poor in the U.S. have access to a variety of goods that might have been unattainable for them 20 years ago-including TVs, cell phones, and microwaves . And yet, the problems typical of unequal societies are keenly felt in the U.S. By way of explanation, Payne offers that "[i]nequality makes people feel poor and act poor, even when they're not. Inequality so mimics poverty in our minds that the United States, the richest and most unequal of countries, has a lot of features that better resemble a developing nation than a superpower" . As an aside, this feeling of being poor, despite all evidence to the contrary, may perhaps help explain why so many extremely wealthy people in the U.S. classify themselves as "middle class." These individuals might genuinely feel that they are not particularly wealthy, in part because they are comparing themselves to those who have even more than they do . For example, in her research into inequality, sociologist Rachel Sherman interviewed a woman with a household income of over two million dollars a year who described herself as middle class. In explaining her reasoning for this categorization, the woman stated that "no matter what you have, somebody has about a hundred times that" . Near-universal access to televisions and cell phones, while arguably making life easier and better for citizens, has done little to solve problems of inequality in the U.S., including the psychological impacts of status competition and mistrust that give rise to even greater social harms. It would be naïve to assume that access to more goods and services alone, without an attempt to address underlying issues of inequality, would have any different effect. This is particularly true given that in societies where income and wealth inequality reign, citizens invest more of their effort and money into signaling their status through the conspicuous consumption of positional goods . Positional goods are intended to signal one's position or status in a society. They are scarce-usually made intentionally so by those offering them-and therefore presumably available only to high status individuals with great wealth. Examples of positional goods include brand name items, rare and expensive sports cars, and tickets to high profile sporting events like the Super Bowl. Even as new technologies make a variety of new goods and services available to citizens across the wealth and income spectrums, as long as wealth and income inequality remain there will be status competition that plays out in part through the acquisition of positional goods-which despite technological advance will continue to remain scarce either naturally or through the efforts of those offering them . In fact, as certain goods and services, because of their newfound abundance, lose the power to signal status, we might expect the development of new vehicles for signaling status. Non-fungible tokens, or NFTs, might be one example of this . Policymakers should not expect, therefore, that simply increasing access to a variety of goods and services will address the larger problems that arise from inequality-though these goods and services might indeed make people's lives better in measurable ways. As long as income, wealth, and other forms of inequality remain, they should expect that mistrust and status competition will continue to flourish, leading to the raft of additional social harms seen in unequal societies like the U.S. In fact, there is intriguing initial research suggesting that a backdrop of abundance might exacerbate the negative feelings that characterize unequal societies, making the harms that arise from these feelings even more likely to occur. --- How perceptions of abundance might exacerbate the psychological impacts of inequality In the previous Section Inequality: Definitions and harms explained why a newfound abundance of goods and services, made possible by new technologies, is unlikely to remedy the negative psychological effects of living in an unequal society. But it is plausible to think that this abundance might at least mitigate these psychological effects somewhat. After all, if most people are newly able to get more of what they need, they may become less concerned with what they do not have, which in turn might open the door for them to trust more and compete less. Interestingly, however, a series of studies by Francesca Gino and Lamar Pierce suggest that this might not be the case, and that in fact, a setting of abundance might have the opposite effect-increasing mistrust, envy, and some of the other negative feelings common in unequal societies. Building on work finding that the presence of wealth may encourage people to engage in unethical behaviors, Gino and Pierce set out to study how a context of abundance might affect people's behaviors in an experimental setting . In the study they define abundance as "a large pool of visible resources that are either shared by [societal] members or possessed by individuals within the [society]." Subjects in the study were asked to complete a word task and were given a pile of cash from which to pay themselves based on their performance. In the "abundance" condition, participants were given the cash from a table containing much more money than was necessary to pay all participants, whereas in the "scarcity" condition participants were given funds from a table that contained only enough cash to pay the participants. The researchers found that the abundance condition produced twice as many cheaters-participants who overstated their performance in order to pay themselves more than they had earned-than the scarcity condition. The magnitude of the cheating-i.e., the level of overstatement-was also significantly higher in the abundance condition. In subsequent studies, Gino and Pierce set out to determine what might be prompting the unethical behaviors seen in conditions of abundance. They examined a number of hypotheses, including the possibility that the cheating was mediated by feelings of envy based on a perception of inequity triggered by the abundant cash. And in fact, the authors did find that envy was a prime motivator of the cheating, while alternative hypotheses, like simple greed or participants' perceptions that their actions would harm others less in the abundance condition, were not supported . This series of studies by Gino and Pierce has not been the subject of subsequent research, so the results should not be overstated. However, their findings do dovetail nicely with the psychology literature on inequality discussed above and provide some insights into the feelings and behaviors prompted by conditions of abundance. As explained above, feelings of envy triggered by conditions of inequality can cause people to mistrust others and believe that these others are succeeding unfairly. These perceptions in turn, can lead to unethical behaviors as people rationalize their own attempts to get ahead . The so-called "abundance effect" identified by Gino and Pierce suggests that similar feelings and behaviors might be prompted by the mere presence of abundance, which, absent any evidence to the contrary, can give rise to perceptions of inequity. Extrapolating from the lab to the real world, what might this mean for a situation in which a new abundance of goods and services is introduced into a highly unequal society like the U.S.? As explained above, that event alone is unlikely to remedy the psychological harms that flow from living in conditions of inequality. But, more than this, the new abundance of goods and services-especially if it is not clear that everyone is benefitting equally from it-could exacerbate the existing negative feelings engendered by inequality or trigger additional adverse emotions, as the new visible reminders of abundance activate people's sense that they are not getting their fair share. Further, as explained above, these emotions and behaviors associated with feelings of inequity are causally linked to a wide range of social harms, including increased violence, worsened health, and slower economic growth. Rather than expecting the new abundance of goods and services to remedy these problems, there is reason to believe that it might worsen them absent significant policy intervention. --- Implications The above discussion of the political and psychological forces underlying inequality leads to two major conclusions about how a new abundance of goods and services can be expected to impact an unequal society like the U.S. First, the fact that these goods and services will become theoretically abundant does not necessarily mean that they will be abundant-i.e., widely available across income and wealth distributions-in practice. And second, even if the new abundance of goods and services proves in fact to be accessible to all, this will not automatically mitigate inequality and the social problems that grow from it. It might even exacerbate these problems by further triggering the psychological forces that give rise to them. In this Part, I explore what actions should be taken if policymakers want the coming abundance to offer real gains to the wellbeing of citizens on all rungs of the social ladder. --- Ensuring access to abundance Economic theory predicts that when a resource is abundant the cost of this resource will approach zero, making it theoretically available to all who desire it. As demonstrated by the phenomenon of "scarcity amid abundance", however, this prediction often fails to be realized in practice. The reasons for this are myriad, but as discussed above, they can include hoarding by those in power, the absence of distributional infrastructure, the influence and greed of small but powerful interest groups, or a lack of will on the part of decision-makers. What does this mean for policymakers who hope that a new influx of goods and services brought about by technological change can be enjoyed by all citizens? The first lesson is that this might not happen without identifying and eliminating potential barriers to access. For example, Lemley notes that the Internet has made informational and creative content abundant in the economic sense . Yet, it is not abundant in the practical sense because large swaths of the population-about 18% of African American households, among others-do not have home Internet and so are unable to easily access this content . Susan Crawford identifies cost as the major driver of this lack of access, and points to a lack of competition and government oversight of Internet service providers as the underlying culprits . According to Crawford, the way to make this content truly abundant in both the economic and practical senses would be for the government to invest in the necessary infrastructure and then allow private actors to use this infrastructure to compete for consumers . Scholars predict that 3D printing will lead to an abundance of goods in the same way that the Internet has led to an abundance of content . According to Lemley, for example, the day may soon arrive when most citizens will have access to 3D printers in their homes or public facilities and will be able to manufacture a variety of desired goods with widely available online designs . This prediction might in fact be more easily realized than the goal of universal home Internet access, since 3D printers do not require costly infrastructure that can hinder competition. As 3D printing technology improves and more companies enter the market, then, it is quite possible that the cost of owning a 3D printer will drop to the point where most homes will have one, just as most homes in the U.S. now have a personal computer . However, that scenario is not necessarily a given, and it could also be the case that the cost of 3D printers will remain high for a significant amount of time, leading to disparities in who can take advantage of their manufacturing abilities. This disparity could in turn exacerbate existing inequalities as those most in need of what 3D printing has to offer are the least able to access it. Policymakers might therefore consider what could be done in this latter scenario to ensure equal access to 3D printing across wealth and income distributions. For example, though Lemley talks about 3D printing being available in "public facilities, " this is a scenario that will require planning and funding to be realized. That said, it should be relatively straightforward for the government to provide funds to ensure that 3D printers are in fact available and accessible in libraries and other public places. Desai and Magliocca also discuss government interventions that can be undertaken to ensure that people-once they do have access to 3D printers-can take full advantage of what the technology has to offer, including creating intellectual property infringement exemptions for small-scale printing activities and establishing a notice-and-takedownbased safe harbor for websites hosting files with 3D printing instructions so that these files can also be widely accessed . A similar analysis holds for Lemley's prediction that robots will be able to do for services what 3D printers will do for goods, completing tasks like serving meals, cleaning houses, and driving cars. The challenge for policymakers is in ensuring that all households have equal access to these technologies. Given the expectation of the kinds of tasks these robots will eventually perform, it will not be enough, as it might be with 3D printers, to have these robots available at public facilities. Individuals must have access to these technologies in their own homes. Government subsidies-for example in the form of tax rebatescould help ensure that these important technologies become widely available. In contrast, Lemley's predictions about synthetic biology might look more like the Internet scenario due to the presence of mediators and gatekeepers. For example, Lemley hypothesizes that advances in genetic engineering, when combined with 3D printing, will allow for medical offices to "generate custom genes to order" and create organisms and body parts in-house . However, as the current state of medical care in the U.S. teaches us, the fact that a doctor's office or hospital can do something cheaply and easily does not necessarily translate into better access to these services across the population. The U.S. lags behind other countries in access to affordable health care, which is hypothesized to result in part from a lack of universal insurance coverage . Unsurprisingly, this lack of access hits those at the lower end of the income spectrum hardest . The fact that medical providers may be able to offer advanced services at a lower cost to them will not, therefore, guarantee that all members of the population will be able to affordably access these services. In fact, this possible future state of affairs may end up exacerbating existing inequalities, as those who already have access to medical services will be able to take advantage of even more advanced technologies, while those without access will be left in the cold. And though Lemley entertains the possibility of a time where individuals may be "printing [their] own organisms", most of us will likely be depending on medical intermediaries for these kinds of services for the foreseeable future . If everyone is going to reap the benefits of the new abundance brought on by advances in synthetic biology, then, policymakers must work on solving existing problems of access to medical care. To this end, scholars have hypothesized a number of ways in which the U.S. might improve access to care; offering health insurance to all its citizens, capping costs from co-payments and deductibles, and providing exemptions to out-of-pocket costs for high-value or high-need services are just a few examples . This approach should not only ensure that new medical technologies become widely accessible, but it should also do much to address a current significant source of inequality in the U.S. In sum, what this analysis suggests is that ensuring equal access to newly abundant goods and services brought about by technology will require planning on the part of policymakers. The conventional rebuttal to any call for government intervention, of course, is that the invisible hand of the market will handle things most efficiently and so intervention should be stayed absent evidence of market failure. Here, I have tried to make the case that there is in fact market failure, rooted in the political and power dynamics underlying questions of access to and distribution of resources. What, then, should this intervention look like? In some cases, it might involve the relatively straightforward step of ensuring that a particular technology like 3D printing is available in libraries or other facilities. In other cases, it will involve remedying existing structural inequalities and problems of access, including the current lack of access to medical services-a thorny and complex problem that demands a multi-pronged approach. But in any event, policymakers should not expect that the access issue will resolve itself, no matter what economics might predict, and they should be thinking now about how to implement policies that will help all citizens take advantage of newly abundant goods and services. Solving broader problems of inequality so that the new abundance can lead to real welfare gains Planning to guarantee widespread access to a forthcoming abundance of goods and services is the first step in ensuring that this new abundance does not contribute to existing problems of inequality. But, even if successful, this planning will not mitigate or solve these problems. To be sure, ensuring widespread access to welfare-enhancing goods and services undoubtedly has the potential to improve lives. For example, a society in which more people have access to more advanced medical technologies is almost certainly better off than a society in which this access does not exist. But whether or not the new abundance is made available to all, in a society where other extreme forms of income and wealth inequality exist this abundance will not solve the myriad social problems that grow directly from inequality and its negative psychological impacts. In fact, depending on how policymakers respond, the new abundance could end up reinforcing the psychological distress that leads to this array of social problems seen in unequal societies. If policymakers wish to solve these problems, then, they need to tackle these other forms of inequality head-on, rather than expecting a new influx of widely available goods and services to do the work for them. Exactly how they might do so is beyond the scope of this chapter, but many scholars have taken up the topic and offered a variety of innovative and feasible solutions. Further, lest the task seems too daunting, policymakers need not believe that achieving perfect equality-even if it were possible to do sois necessary to reap the psychological and social benefits of more egalitarian societies. As discussed above, the psychological and social harms of inequality are often triggered by the sense of unfairness and mistrust that arise in situations of extreme and visible inequality that cannot be rationally justified. Indeed, there is at least some evidence that some level of justifiable inequality might be psychologically and socially beneficial because it gives people visible hope that they can improve their own situations in life ; but see Cheung . Policymakers can therefore focus their efforts on addressing the extreme inequality that currently prevails in the U.S.; to this end, a number of proposed interventions, including inheritance and estate taxes, government transfers to bottom earners through universal basic income or earned income tax credits, and increased funding of social safety nets could be highly effective and lead to significant gains in the battle against inequality . --- Conclusion Society may soon experience a new abundance of goods and services as emerging technologies lower production and distribution costs. But the effect of this new abundance on current conditions of inequality in the U.S. has yet to be examined. Though it is tempting to hope that the coming abundance of goods and services will help remedy inequality and the social problems that attend it, my analysis here suggests that this prediction is unlikely to come to fruition without significant policy intervention. Instead, problems of inequality are likely to persist under new conditions of abundance, and in fact may worsen. For those interested in addressing the significant social problems that arise in unequal societies, the solution is twofold. First, policymakers must plan for the coming abundance of goods and services in order to ensure that it is truly shared by all. And second, they must address extreme income, wealth, and other forms of inequality directly, rather than hoping, without basis, that increased access to goods and services will mitigate these social problems. --- 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. --- --- --- 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.

With technological advance has come the possibility of a new era of abundance. Technologies like D printing and robotics promise to lower the costs of production and distribution of goods and services, presumably making these goods and services readily available to those across income and wealth spectrums. This undoubtedly is a good thing. But what will be the e ect of these technologies on existing wealth inequalities and the psychological and societal burdens they impose? Can we expect that this newfound abundance will help remedy the current historic levels of inequality in the U.S. and other western countries? Unfortunately, the answer is likely no, for two reasons. First, history suggests that inequality often persists even under conditions of abundance due to dynamics of power and politics and ongoing impacts of structural inequalities. Having more than enough of a particular good or service to go around does not guarantee that all will have access to this good or service. Second, even if the new abundance enabled by technology extends into all levels of the socioeconomic spectrum, enabling individuals to access goods and services (and their attendant benefits) previously beyond their reach, the harms that attend unequal societies will persist. Increasing evidence suggests that these harms, including increased violence and decreased health, arise not from access (or a lack thereof) to particular goods and services, but from the adverse psychological consequences of living in an unequal society. This is a psychological burden shared not just by those at the losing end of the inequality equation, but also those who enjoy a relative advantage in society. Unequal societies are psychologically harmful to all who live in them, regardless of where these individuals fall on the socioeconomic spectrum, and largely independent of the particular goods and services they enjoy. The upshot is that society cannot rely on new abundance technologies to automatically solve problems of inequality and the social and psychological burdens that plague those who live in unequal societies. Indeed, depending on how society responds to questions of access to these technologies, their introduction might exacerbate various forms of inequality. In light of this, it is crucial to address conditions of inequality head-on, so that the new era of abundance promised by technological advance can lead to real gains in individual and societal wellbeing.

Introduction The Soninke are black people who live in West Africa. They are mainly concentrated in the upper valley of the Senegal River divided into three countries namely Senegal, Mali and Mauritania . They have played a prominent role in the history of this sub region with the Ghana Empire called Wagadu in the Soninke language. They founded also many kingdoms after the decay of the Ghana Empire in Senegal, Mali and Mauritania before the French colonization. The latter is considered as the mother cause of the mutations of the Soninke society. Thence, we are going to base on this historical event which has caused a great cultural hemorrhage in Africa in general and in Soninkara particularly to explain the deterioration of the Soninke way of life from the colonial time to the present day. We will focus on the cultural transformations, brought about by the French colonization in Africa, on the socio-traditional and moral aspect and also on the economic aspect in the Soninke society. In addition to that, other causes like migration and modernism will also be studied in this work in order to better demonstrate how the western cultural influence has deteriorated the socio-traditional and moral values of the Soninke people. Our approach is different from those who usually tackle this issue based only the conflict of identity between the hooro and the other social classes . This study is going to deal with the social mutations on the whole system of life of the Soninke people. --- The Impact of the French Authority on Soninkara To implement successfully their machine of exploitation, the French colonial authority thought fit to destroy or neutralize all the powers of the traditional leaders, to impose a new administration and to settle a new education system in its colonies. At that epoch of colonialism, during the XX century, the French colonialists had no consideration of the way of life of their subjects that they looked upon as people without civilization. They argued that black African people needed to be civilized. For that purpose, they thought they had to format the traditional system of life of the Soninke people rooted in some values which can really make a society developed. Therefore, the imposition of a civilized system based on Judeo-Christian capitalism on the Soninke land, to meet the industrial needs of France, was view by the French colonialists as a great necessity. --- The Destruction of the Traditional Leaders' Power The French colonial system had ousted the traditional rulers by putting an end to the royalty of the tunka nlemmo 1 in Soninkara. The main objective of this domination was the exploitation of the resources of the Soninke land in order to develop France. As the western countries were in competition of industrial development, that enterprise was necessary. Thence, all the local powers of the Soninke society of that epoch were neutralized. The hooro 2 who had the monopoly of the chieftaincy were turn into in- 1 The tunka nlemmo represented the royal family. 2 Hooro is the plural form of hoore which means free-person. The hooro are the upper class of the Soninke society. They are the traditional rulers, the religious leaders and the principal warriors. They have a privilege and prominent over the other social classes namely the nyaxamalo and the komo . termediaries between the French colonial authority and their people. They worked for the French colonialists. The one who were reluctant or tried to revolt were eliminated. Besides, the local army was weakened. A great number of the Soninke warriors 3 were killed for that purpose. The French colonial authority was empowering itself by acting so. That facilitated the realization of its project of developing France with the natural resources of Africa. As a result, the Soninke local traditional leaders had lost their power in favor of the French colonialists. Thus, they served the French colonial administration. They fulfilled their tasks under the control of the French colonial authorities who became their masters. Therefore, the heads of the Soninke people of the colonial epoch turned into valets of the French colonialists who controlled their land. As they lost all the prestige of royalty, they were forced to work for the interests of France. Since the French colonialists had strong military power with sophisticated weapons, they succeeded in extinguishing the Soninke ruling and military powers. Then, all the leaders of Soninkara were listed and controlled by the French colonial administration. Those who were the allies of the French colonial authority were rewarded, and those who were reluctant were intimidated . The traditional chief or king whose duty was to protect his people from the enemies and safeguard their welfare, "no longer ruled in his own right, but as an agent of the colonial authority" . That was the starting point of the substitution of the African political system by the Europeans. Pollet & Winter write: "L'implantation française à Yélimané prit d'abord la forme d'un camp militaire. Elle était à l'image de l'effort politique du colonisateur qui voulait substituer sa domination au pouvoir traditionnel. […] Le changement de régime eut pour conséquence politique principale que la puissance des Dukure 4 fut décapitée; leur déchéance s'opéra au moyen de l'interdiction de la guerre, justification de leur pouvoir et source de leur prestige, et par l'abolition de l'esclavage, moyen d'enrichissement de tous les nobles en général. Les chefs durent obéir plus forts qu'eux et ne furent plus que des intermédiaires entre les administrateurs et la population." . Translation: "The French settlement in Yelimane started by a military camp. It meant the political effort of the colonizer who wanted to substitute the traditional power. The change of regime had as main political result the destruction of the Dukure's power; their collapse was achieved by the prohibition of war, justification of their power and source of their prestige, and by the abolition of slavery, mean of enriching of all the nobles in general. The chiefs had to obey those who were more powerful and were mere intermediaries between administrators and the population." 3 Warrior is called yugo sire in Soninke, warriors = yugu siru. According to the Soninke oral tradition, the yugu siru were eliminated by the tubabu ni . Moreover, to better ensure their power in the Soninke land, the French colonialists had created troubles between the traditional leaders who were essentially hooro and the komo 5 through the abolition of slavery and the creation of "villages de liberté". This is the strategy to divide in order to better reign. The French colonial authority instigated the clan of the slaves to emancipate by claiming their liberty. That was a way of creating a social imbalance to weaken the traditional ruling leaders of the Soninke society so as to put an end to the tunkaaxu 6 and the mangaxu 7 . In fact, all the ways and means were good to extend the French colonial authority power and influence over Soninke land in order to meet the industrial needs of France which was in competition with its neighboring countries. And the social troubles could, really, weaken the authority of the traditional chiefs or kings. It was in this context that, the lower class, the komo have started to claim freedom in Soninkara. Thus, some families of this clan moved to remote lands to set up their own villages to free themselves. For example, we have the villages of Turime and Bema in Bakel 8 . That was a strategy to ensure their power and influence in the framework of their project of expansionism and their agricultural and industrial exploitation in Africa in general. Furthermore, the sons of the chiefs were called to be trained in the French school in the region of Saint-Louis so as to assimilate them the colonial way of thinking and doing or the western cultural values, hence "Assimilation policy". That was a forced integration of the traditional rulers' heirs through western school. --- The Assimilation Policy of the French Colonial Authority The French colonial authority thought that it was vital to indoctrinate the sons of the chiefs the western cultural values in order to better succeed in their enterprise of agricultural and industrial exploitation on the lands of their subjects. This is a cultural colonization which is the worst domination. Thus, the traditional rulers of Soninkara would lose their heirs. That phenomenon could also make the sons of the chiefs hostile to the some traditions of their fathers. In other words, it could jeopardize the continuations of many traditions. In actual fact, the French colonial authority used school as a cultural to stop the traditional training which was a channel of transmission of the Soninke cultural values in the royal court. One can understand that through the great importance that Faidherbe gave to the French school in Senegambia. tard, qu'il n'y eut un seul chef, en Sénégambie, qui ne sortit de cette école." . Translation: "…Faidherbe decided then to open a special school for them where they would learn our language for the suitability of our relationships with their country where they would be brought up as some individuals meant for an environment of semi savage people […]. Faidherbe wanted, twenty later, every chief in Senegambia to be a product of that school." Faidherbe used the French school as a weapon of conquest. He wanted the sons of the chiefs to be transformed into allies of the French colonial authority in order to support the exploitation of the natural resources, in Africa, for the economic development of France. That is why, for this French former colonial governor, they must understand the French language and adopt the European mentality so as to be able to collaborate with the colonial administration for the interests of France. In fact, as far as the sons of the chiefs assimilated the European values and ideologies, that would facilitate the colonial penetration in their society which were being exploited by the French colonialists on the behalf of France. They could provide them much information as the French colonialists did not understand the languages of their subjects. The sons of the chiefs would play both the roles of representatives and interpreters for the French colonial administration. Thus, the colonialist could easily know the gist of their subjects' affairs. They could also distinguish their enemies from their allies. Therefore, the opening of the French school, during the colonial epoch, on the African soil was a way to better control and dominate the local authorities and their people. In other words, it was a device of control. And culturally speaking, when we learn a foreign language, we are strengthening the people to which that language belongs. In consequence, we become culturally dependent. That can even generate a kind of complex which symbolizes a psychological weakness and influence negatively our cultural identity. In actual fact, the promotion of a language can hugely contribute to the cultural and economic development of a society or country. Language is vital when regarding culture and economic development. This issue is among the major difficulties of the black continent's development. The leaders, who should be references as far as African culture is concerned, are transformed into white blacks . They behave as western people to the detriment of their culture. They have western mentality and tastes. They educate their cultures in western culture. As a matter of fact, it is the continuation of the phenomenon of the traditional leaders' sons' indoctrination which is reflected through the contemporary African rulers. The heirs of the traditional authorities were trained to act in the economic interest of France. The current African leaders act in the framework of "France-Afrique" cooperation for the benefit of France. Nothing has changed since the colonial epoch except the external aspect of things. Till nowadays, the African leaders are forced to incorporate western values such as republic, democracy and many others to serve the western countries. All their policies are inspired of West. They have nothing of their own. They are still following the orders of their masters. They are under their control. That is why, they act as the representatives of the western leaders in Africa. This is the effect of the assimilationist policy initiated by the colonial authority in Africa for the benefit of Europe. The western school is an industry which produces western products. It makes some Africa people alienated and many others hybrid. It is a silent killer of the African traditions and values. In addition to the policy assimilation through the French school, the colonial authority set up a new court in Soninkara to compel the traditional leaders and their people to judge according the western laws. --- The Imposition of the French Colonial Justice to the Soninke People Before the arrival of the French colonialists, the Soninke people had their own laws. They were written, but they were inspired of good morality and kept by heart. The disputes between people were mainly settled by reconciliation. In addition to that, there were some cadis who judge basing on the Islamic sharia. But when the French colonialists arrived on their soil, things had changed. The French colonial authority did not recognize the traditional laws of its subjects. It brought an apparatus of justice which was erotic to the Soninke society. Some laws that the Soninke people did not know and did not apprehend. Since then the local justice was replaced by the European one. As justice is fundamental in a society, the French colonial authority broke the Soninke local justice in order to better ensure their domination over the Soninke land. Thence, the customary law and the sharia were not considered by the French colonial administration. Any traditional leader or cadi could not do justice without the knowledge of the French colonial authority if not they risked punishment from their master who controlled everything. There was a reorganization of justice in the colonies in order to maintain the French order. Eric Gasparini writes: "…tout jugement rendu en matière indigène portant condamnation des dommages intérêts ou une amende, ne sera exécutoire auprès qu'après avoir été approuvé par le commandant de l'arrondissement. […] l'administration une politique indigène en matière judiciaire, mais elle est également pourvu de moyens de contrôle importants sur le droit local." . Translation: "…any judgment done for the indigenous affairs regarding the condemnation to damages or fine, will only be executed after being approved by the district commandant. […] the administration has not only an indigenous policy in terms of justice, but it was provided with important means of control on the local law." The French colonial authority overviewed the local justice due to the direct administration of France on its colonies. Pollet and Winter write: "Il fut institué un tribunal coutumier, presidé par le chef de subdivision conseillé par deux conseillers. Ce tribunal rendait ses jugements, tant au civil qu'au pénal, en accord avec la 'coutume saracolaise coranique' dans la mesure où elle n'est pas contraire à la morale et à 'l'ordre publique français. […] La Cour de Nioro 9 tranchait en seconde instance et était seule compétente pour les affaires criminelles." . Translation: A customary cout was settled, led by the subdivision chief, assisted by two counsellors. That court gave its sentences, as well civil as penal, in accordance with the saraolle koranic custom as far as it is not contrary to the French moral and public order. […] The Court of Nioro. The whole judicial power was between the hands of the French administration. Everything was done according to its convenience. Then, the local leaders and their people were subjects, and their land was ruled by the French colonial justice. In the beginning, people were reluctant to the French laws, but gradually, they realized that they had no choice but to admit and comply with them in order to avoid troubles with the French colonialists. As time elapses, many Soninke people went to the court and civil registry established by the French colonialists either to settle some of their familial disputes or for having papers. Since the declaration of the children' birth, marriage and death was a great necessity, the Soninke people were obliged to go to these places to have papers for their identification across their balkanized countries and the world. This also was the reason why, many parents were obliged to send their children to French school. They needed to understand the new world in which they were. And nowadays, civil papers are in the core of the Soninke people's life on account of school and migration. This is an effect of the substitution of the customary law by the French law. This phenomenon marked the start of the modern Soninkara. Furthermore, to transform the African societies in general and Soninkara, the French colonialist imposed the capitalist system to their subjects. The latter would upset the traditional system of the Soninke people. --- The Capitalist System in Soninkara The principal activity to survive of the Soninke people, before the coming of the French colonialists and during the colonial time, was agriculture. They also practiced breeding cattle 10 , fishing and hunting. They exchanged their agricultural products by barter, which was basically the mean of transaction, to have sappe and kambare " and cattle of ship, goats or oxen . As a matter of fact, the Soninke 9 Nioro was a kingdom in Soninkara before the colonization of Africa. That is why the Soninke people had some experience in commerce hence the dioula 11 . The latter were realized not to practice commerce permanently by Dr. Tautain . They did it for a specific objective. Once the latter was reached, they resumed growing crops. Pollet and Winter write: "…le métier de dioula n'est généralement pas une profession que l'on exerce toute sa vie; on fait un certain nombre de voyages pour s'acheter un ou plusieurs esclaves, de gagner le douaire d'une femme, etc., puis on reste à ses cultures." . Translation: "…the dioula's trade was not usually a job that people did in their whole life. They travelled to purchase one or many slaves, to gain the dowry for a woman, etc., then they focused on their crops." The Soninke people's experience in commerce and their crops would interest the French colonial authority during the colonial period. As a matter of fact, the Soninke had dealt with the French in the precolonial time precisely during the slavery in Africa. They were considered as the most civilized people in the sub-Saharan region by the Europeans at that epoch. The latter were impressed by their skills in commerce. Pollet & Winter write: "Les Soninke ont été le sujet d'écrits épars qui se rejoignent cependant sur l'importance qu'ils accordant généralement aux activités mercantiles…, il s'agirait essentiellement d'un peuple de commercants." Translation: "The Soninke people have been the topic of various writings which however recognize the interest that they usually grant to the mercantile activities…, it would be essentially about a group of traders." It was at that period that the yille, a basic commodity in Senegambia, became vital in the transactions. Bathily writes: "Les transactions sur le mil, denrée vivrière de base en Sénégambie, furent un facteur d'une grande importance dans l'histoire économique de la région au XVIIIe siècle. Le rythme de la traite des esclaves au Galam 12 était largement conditionné par les disponibilités en mil dans les magasins des forts." Translation: "The transactions with the millet, basic commodity in Senegambia, were a very important factor in the history of that region in the XVIII century. The rhythm of the slave trade in Galam depended mainly on the availabilities of the millet in the stores of the forts". Even after the abolition of slavery in XVIII century, the French colonizers kept on dealing with those whom they considered the biggest traders in the sub-Saharan zone in slaves . When the French arrived again in Soninkara, in the framework of the colonization of Africa, to meet the industrial needs of France that things began changing. The French authority urged their subjects to seek profit by developing business in agriculture. That traditional basic crop interested again the French colonial authority so much that it imposed the Soninke people to pay the contribution by this agricultural product. Millet which was a basic survival product among the Soninke people was vital in the transactions. Then, a service, created by the colonizer, named "institutions de prévoyance" was set up in some Soninke areas like Dyahunu and Tiringa so as to control the yille. The French colonial authority had even prohibited, in those areas, the selling of this crop to the strangers . The colonial administration had obviously interest in controlling the millet seeds . In fact, each family used the yille for paying tax to the colonizer . The contribution became both a sacerdos and a great worry to the subjects especially when it must be paid by money in the XX century 13 . That duty, really, put the Soninke people under the yoke of the French authority. Furthermore, the introduction of some services in the agricultural affairs like the "institutions de prévoyance" supervised the subdivision chief was something new to the Soninke people of that epoch. That phenomenon has penetrated progressively the agricultural field of Soninkara until the present day. Then, many services have been set up in some Soninke areas in the framework of modernizing agriculture. As in business the sources of incomes are the services and the goods, so the increase of the agricultural exploitation and the creation of various services were advantageous to the colonizer to gain lots of benefits. This is the reason why, the Soninke people were told to enlarge their fields for more crops to ensure the consumption and the payment of the contribution to the French administration. An important amount of seeds must be kept for seeds of the following rainy season in some attics called "greniers de reserve" managed by the colonizer. That service named "institutions de prévoyance" initiated by the French administration was not free. The Soninke cultivators paid that service with their yille. That was a way of exploiting their subjects of the French colonizer. The noticeable phenomenon which upset the socio-economic life of Soninkara was the substitution of the yille, which used to be a capital mean of transaction , by money. The penetration of money has changed lots of things in the Soninke society. In fact, the agricultural exploitation undertook by the French authority to meet the industrial needs of France and the development of commerce, during 13 In 1945, the "Franc des colonies francaises" was imposed to the French colonies in order to allow France to benefit from the natural resources of Africa. https://www.nofi.media/ the colonial time, in the area of AOF 14 made people great seekers of profits. In addition to that, the ecological crisis with the droughts and the payment of tax imposed to the subjects in the French colonies made socio-economic life difficult in Soninkara. Thus, people were obliged to sell a portion of their crops or to move to the agricultural exploitation areas or big cities like Dakar, Bamako for gaining money. That was the beginning of the peasants 15 , transformation into money seekers. People became money seekers for paying tax because the colonizer imposed to the subjects the payment of the contribution by money . They also sought money for meeting their needs such as buying manufactured products from Europe, paying fares 16 and other services. The capitalist system began penetrating the life of the Soninke people. Thus, people moved as far as the "basin arachidier" of Sine Saloum in Senegal, in the framework of the "navétanes" to make a living. As capitalism is characterized by the unceasing seeking of profit, Africa became another market where the colonialists could collect lots of benefits. In actual fact, developing the agricultural exploitation and the commerce of the finished products made by France was a strategy to extend the French market in Africa. As there was a strong competition between the European powers, there were tremendous stakes in developing the capitalist system in the black continent for the French and theirs rivals . Thus, a new trend of survival started in Soninkara. Some Soninke people started to lead mercantile activities for making a living. And progressively, commerce began to gain lots of actors among the Soninke. Then, the importation of the manufactured products favored the opening of modern shops in the Soninke homeland. What would change the social consideration of richness among the Soninke of the XIX century. It was no longer based on the amount of crops stocked in the mara 17 but that of money. That marked the beginning of materialism which has grown gradually. The phenomenon of money acquisition would jeopardize the common life 18 that used to symbolize social cohesion in Soninkara, for the seeking of capitals enhances individualism 19 and selfishness. This system of life would threaten the social cohesion and the harmony in the social relationships of the Soninkara. The possession of private properties can make the 14 Afrique-Occidentale Francaise was the western side of Africa colonized by the French. It encompassed several countries such as Senegal, Mauritania, Mali, Niger, Burkina Faso, Benin, Guinea Conakry and Ivory Coast. Its capital was Dakar. http://www.larousse.fr, Afrique-Occidentale francaise . --- 15 In soninkara, people usually consume the crops they grow. That is to say the crops are meant for consumption. --- 16 Modern means of transports such trains and cars have been brought to Africa since the colonial time. In Soninkara people used trains to travel to cities like Dakar or Bamako. 17 Mara is an attic where people used to store the annual provision. --- 18 The Soninke people believe that the common life is better. They say "Bereke ana koffo nya". That is to say we have benediction in collective; it is more advantageous to live together for individuals. --- 19 For the Soninke, only God is one, but human beings are social creatures, they need one another in order to survive on earth. individual independent or even self-sufficient. Some troubles in the social relationships, therefrom, could put in danger some moral rules such as the respect towards the elders 20 and the sense of sharing concerning one's properties . These fundamental moral values would be threatened by the materialism that capitalism embodies. And unfortunately, that money system enhanced by the industrialization of Europe has killed many social values which symbolized the common life. It has transformed the Soninke people into "big money seekers", hence migration which is nothing but the imposition of the capitalist system's result in Soninkara. --- The Migration of the Soninke People and Its Effects in Soninkara The migration of the Soninke people towards the big cities of, central Africa and France is another phenomenon which is a direct consequence of the French colonization. It has brought about many social mutations in Soninkara. The Soninke people love migration so much that their neighbors recognize them by that feature. For example in Senegal, the Wolof people say that the Soninke 21 love migrating to Europe so much. To study the effects of migration in Soninkara, we are going to divide the Soninke community into three groups: the kaara nko , the Soninke people in the capital cities and those who migrate to France. --- The kaara 22 nko The migration to some capital cities of West African countries, to France and to Central Africa has brought about many changes in Soninkara. Certainly, the houses for dwelling have been well modernized and the living conditions as well, but many customs and traditions have disappeared. And some other have been neglected. People are more engrossed in materials than in social values. That phenomenon has destabilized the social cohesion and the harmony in the relationships in the ka 23 and the debe 24 . Formerly, the hirsaaxu 25 was taken into account in Soninkara. Migration, however, has jeopardized this custom. As the fakke 26 and the possession of materials in the kaani 27 have rendered many people disrespectful, the orders and the recommendations of the hirse 28 are less or no longer considered. The hirse who 20 Soninkara is an oligarchical society. The Soninke people consider much the respect and obedience of the hirsu even if this fundamental is declining so much in the modern Soninkara. --- 21 The Soninke represent 70% of the African migrants in France. They consider France as El Dorado. The migration to France is mainly the dream of the Soninke youth. used to symbolize unity in the koffo 29 is not any more respected as gani 30 . He was the vector of cohesion and harmony in the group. He called the family members to the oneness of the koffo for maintaining strong and harmonious relationships. He looked after the moral education for the welfare of everybody. And he included every member in that task. Moral was in the core of people's life. It was a mean of urging the members to respect religiously the collective life, for the strong attachment to the social life is beneficial for the unity of people. In fact, the main preoccupation of the hirse was the oneness of the group for the social cohesion and the harmony in the relationships in the ka. The abundance of money and materials, however, has undermined the hirsaaxu. If the colonizer had destroyed the power of the traditional for reducing him to a mere servant, the migration, as far as it is concerned, has deteriorated the symbol of unity of the ka: hirse. This phenomenon of disrespectfulness towards the ka nhirse 31 is similar to the death of the general whose troop will undergo defeat because of their broken spirit. Thus, in the kaani, nowadays, we cannot talk any more about unity or oneness because the abundance of money and materials has favored individualism, selfishness, self-sufficiency and indiscipline. Most of the family are not characterized by discipline and modesty, but they are rather arrogant and disrespectful. In consequence, the whole body of moral norms which held people to live together in peace and harmony are, nowadays, affected by these vices. Some values like reverence and obeying the elders have become obsolete. As a result, all the members have become hirse. No one, nearly, admits the authority of the leader. The juniors, the women, the children have a poor consideration towards the hirse. They follow more their passion and desires than the recommendation and the orders of the hirse. This situation is similar to a boat without captain. The one who used to be religiously respected and around him people gathered to have a good social organization sees, nowadays, his authority denied by his members. Each member, almost, is the chief of himself or herself. Therefore, in a social group where everybody is an authority, the social cohesion will inevitably decay. The relationships are strained and the misunderstanding and the internal conflicts will spoil the social balance and the peace between the members. Thus, the stringent responsibilities of the chief such as education and unifying the group that serve to promote oneness in the ka, have been deserted. People are preoccupied with their own businesses. As a matter of fact, migration has enriched the kaara nko, but it has destroyed the social cohesion and the harmony in the relationship. Though they have a lot of comfort, beautiful houses and plenty of commodities stocked in their maro 32 , 29 Any association is called koffo. most of the Soninke people live "alone together". That is to say they are in the same house, but they are not unified. The hirse is no longer the symbol of unity in the ka, for the disrespectfulness and the arrogance of the family members. The abundance of money, in Soninkara, has also changed the consideration in the social relationships. Even the parents show more respect to their sons who are faranci nko 33 . They follow their desires, but those who are in kaara are less considered. This phenomenon is general among the kaara nko. As far as the debe is concerned, the traditional structure dwells intact, but the social cohesion and the harmony in the relationships between the xaabilo 34 are deteriorated. What was more essential in the social relationships between the families no longer exists. Money has given independence to lots of xaabilo especially the lower class. And that has created a kind of emancipation of the komo slaves). In many Soninke villages, the people of the lower class do not admit any more to be considered as slaves hence the social tensions. Since they possess wealth and being also influenced by the western democracy 35 , they claim equality and better consideration. We have noticed that many people of the lower class are richer than their former masters due to the migration. That phenomenon has been the main cause of the social tensions between the hooro and the komo. The latter are fighting in order to build up their identity. Then, they claim respect and an equality of rights between the xaabilo. Though their claims are reasonable, they bump into the refuse of the hooro who do not want to lose their preeminence and their social privileges . That situation has upset the political solidarity and has entailed social conflicts in some Soninke villages. In actual fact, the relationship of domination that used to exist between these two social classes is, nowadays, the principal reason of their conflict. The komo no longer admit obedience and servitude towards their former masters. They do not have any more that will of submission to them, for their fortune. They require, instead, the hooro to recognize their dignity as their fellow creatures. In other words an equal treatment in the social relationships is required. The komo show a will of change and the hooro want a perpetuation of the relationship of domination. The latter try to redefine the social relationships in a way that does not convince the former. That has generated some dynamics which are destabilizing the social cohesion in Soninkara. In some Soninke villages, the social polarization is obvious. For instance, we can quote Tuabou, Baalou, and Laani 36 . This social crisis is almost everywhere in the Soninke villages nowadays. This issue has decayed the basic relations that make up the social order is approximatively family. --- 35 As democracy advocate liberty and equality for all the individuals, the lower class people defy the upper class people so as to be well considered and well integrated in the affairs of the society. They no longer want the noble men to look upon as slaves. --- 36 Tuabou and Baalou are in the county of Bakel and Laani is in Mali. Advances in Anthropology Evans-Pritchard, 1960) in the whole Soninke society. As the axiomatic premises of the social order are affected by a social crisis, the social unity, in the villages, is fragmented. So, the common life is no longer meaningful because people are losing sight of the common interests. They are rather preoccupied with sectorial and private interests. --- The Soninke People Living Capital Cities The rural exodus which is result of the capitalist system, has made lots of Soninke people move to some of the West African countries' capital cities like Dakar, Bamako and Nouakchott. There is a great number of the Soninke living in these big cities. This phenomenon has injured Soninkara socially speaking. Those who have decided in the capital cities cope with a problem of educating their children in Soninkaaxu 37 . The latter is fundamental in Soninkara. For the Soninke people, the one who does not know his or her culture, language, history and genealogy is not a real Soninka nlemme38 . Fortunately, most of the Soninke children born in the capital cities do not really the cultural patrimony of Soninkara. The worst thing is, they do not even know how their homeland is and they cannot speak Soninke. They are influenced by the city life trend so much that they are culturally lost. Even though, the Soninke families living in cities try to keep their social relationship through ceremonies and some organizations, they have failed to educate their children in Soninkaaxu. The kaara nko see it as a shame. Furthermore, the matrimonial alliance with other ethnic groups has impacted negatively on the Soninke living in cities. For instance in Dakar, we notice that those who have married Wolof women have lost their offspring culturally speaking. Many Soninke children cannot speak Soninke because of their Wolof mothers. The latter refuse to integrate in the Soninke way of life. This matter has, fundamentally, destroyed many Soninke family units in the cities. The kaara nko are shocked and disappointed by this issue. They cannot digest the fact that a Soninka nlemme who are not able to express themselves in the language of their homeland. They feel it as a cultural loss. Many Soninka nlemme, in Dakar, are not recognized. In fact, the culture features of a Soninke do not appear on them. They ignore Soninkaaxu. This is the reason, some kaara nko do not consider them as Soninke. --- The faranci nko The migration to France is seen by the Soninke people as a key of success in their life. Formerly, they only focused on agriculture to survive, but since the end of the XX century, things have started changing. Now it is the migration, the greatest source of incomes, which ensure the survival in Soninkara39 . People depend 37 Soninkaaxu means both the Soninke cultural identity and the whole cultural patrimony of the Soninke. Soninkara has become strongly dependent on France. Certainly, migration has been advantageous economically and financially speaking, but it has deeply changed the behavioral aspect of the Soninke people and has negatively affected some social norms. The peasants who have turned into wagers is a phenomenon which has brought about many social mutations in Soninkara. Thus, the value of yille is replaced by that of money. The latter is possessed by the faranci nko who are always considered as the rich in Soninkara. This change of consideration has generated another mindset that is undermining the Soninke moral and social norms. That is why, we notice that the faranci nko are more considered than the kaara nko. They hold the stringent responsibility, in the modern Soninke society, which is ensuring the biraadi 40 in kaara . Since the crops are not regular because of the ecological crisis , all the expenditures of the ka are ensured by the faranci nko. They are the actors who keep afford the homes. It is the main objective of migration among the Sonike. As they like eating much and comfort, they religiously respect the faranci nko. Even the chief of the family who is traditionally obliged to stay in kaara for managing the ka for the common interest, is behind them. The latter usually call them for the commodities, the bills , and many other expenditures. This has ranked the faranci nko over the kaara nko. However, the Soninke migrants in France are undergoing a great influence of the French society. This is the cause of their mutation. To integrate into the French society, they are obliged to abandon or even alienate a part of their cultural identity. The Soninke in France have lost many values especially those who have settled there with their families. They cope with a real problem of education concerning the children. The latter being brought up in a cosmos where there is a lot of liberty, lose sight of their culture. They adopt some values that are utterly contrary to those of their fathers and grandfathers. We observe a break with the Soninkaaxu amid the Soninke of France. They are crying over this crucial social issue. Furthermore, the basic social relations between the xaabilo 41 in France are strained hence gambanaxu 42 . There is, nowadays, a great social conflict between the Soninke of France. The komo do not accept any more the traditional social configuration which gives all the privileges 40 Biraadi can approximately mean commodities . --- 41 The Soninke people are strongly conservative. They have brought in France the same traditional organization they have in Kaara. However, they are coping with a great social conflict between the xaabilo especially the hooro and the komo . The latter claims equal rights and treatment. The former refuses so as to keep their social privileges and preeminence. --- 42 Gambanaxu is a social issue in Soninkara which has appeared in the XXI century. However, if we study meticulously, we will find that it is originated in the French colonization of Africa. Many revolts of the komo have been repressed by the hooro since the colonial time. But the komo are still fighting for equal rights and treatment in the villages. and domination over them. Then, the gambanaxu symbolizes the revolt of the lower class. They argue that all the human beings are equal and must be treated equally. Being in a society which supposedly defends the human rights, the komo have engaged themselves in a combat to fight against the social inequalities of the Soninke societal organization. As a matter of fact, it is them who motivate the komo of kaara to struggle for their liberty and equal treatment of the xaabilo in the Soninke villages. This has been the source of social conflicts in many Soninke villages. The social stability and order are strongly disturbed by the fight led by the lower class for equality for all in Soninkara. This crucial issue has even caused some deaths. We can quote Laani 43 in 2020 as an illustration. --- Modernism Modernism is a movement from the West which is against whatever is traditional and religious. It is like a tradition-killer because it modernizes people and makes them dislike traditional life. And people tend to be more and more modern by loving materials to the detriment of the traditions. The worst thing of modernism is its toxic and pervasive influence on the youth, the hope of the future. Thus, the Soninke people are not spared by this phenomenon. We can say that nowadays, the situation is almost chaotic as far as morals and traditions are concerned. The windows through which modernism has had access to fragment the traditional and moral values are: school and technology. --- Western School The western school is not only a temple of knowledge, but also it is a channel through which some ideologies and doctrines are conveyed. We have noticed that since the youth has started learning in the western school, they have adopted another way of thinking which is totally different from their parents' and grandparents'. That has involved a conflict of generations between the youth and the old people in Soninkara. The western has distorted the mentality of many young Soninke people. The majority of the young people do not like the traditions. They see them something that pertaining to the past . To their point of view, they have a growth mindset, they need to behave in different way as intellectuals. For them, the world has changed, therefore the society has to get rid of the traditions as some white people have done to follow the modern trend. As they have been taught the freedom of thinking and speaking at the western school, they have become hybrid and hostile to the traditions. They, frequently, challenge the traditional way of life of Soninkara. They always utter "these traditions are nonsense". 44 Sibiri says And the old people see their children as white blacks with a narrow mind and some behavioral issues. They state that school is "bad" because it destroys, culturally and spiritually, the younger generation 45 . Usually, they say that the children have learnt, but they know nothing hence the blaming game. For them the French school's content is in conflict with their background . They also say western school only instructs, but it does not educate. Actually, the French school's program is not liked but many African people, because it produces French black people . The content of the syllabus is rather assimilationist. It does not teach any African culture or history. Everything is centered on the West. Furthermore, the concepts such as liberty, right, and democracy are misleading the younger generation. In fact, they have failed to adapt them wisely to their social realities. For them, these concepts involve libertinism which is noticed through their immoral, irresponsible and lawless behavior. Actually, their perception of things involved by these concepts has given them another mindset. That is why, they are in permanent conflict with their background. They lack discipline and wisdom. And these values must go with knowledge in order to make it beneficial to the whole society. Even if they are academically strong, they are morally and spiritually weak. Now, the old people love morals and religion. They are disappointed to see their heirs behaving without "yaagu" 46 . --- Technology Devices Since the information communication technology devices have penetrated the Soninke society, great mutations have been noticeable on the behavioral aspect above of all among the youths. The technology devices such as televisions and mobile phones have driven the nail. They have poisoned the behavior and the faith of the Soninke young people. Nowadays, the youth is morally and spiritually weak due to the misuse of information communication technology devices. Technology has put the young Soninke into contact with the virtual world which renders them stereotype. They take many things as reality from that world which are ruinous for them morally and spiritually speaking. The latter has given them some vague and uncontrollable tastes which make them unaccountable and lazy. Thus, they fail to make efforts to save their cultural heritage. They do not strive to develop their culture. They are followers of their passions: films, football briefly leisure. And, the appearance of social media in XXI century has 45 Till now, there are many Soninke people who prefer Arab Islamic school than French school. This phenomenon has made the old people so much worried. They see their heirs some values which are not at all adequate with their culture and their faith. Their hope is broken when they see, in the villages, the youths behave in an irresponsible way hanging so much to leisure. Since the Soninke culture hates laziness and unaccountability, they feel deeply disappointed. They say "nowadays, the children are nothing for all" 47 . However, the parents are also to be blamed. They buy ICT devices for their children without any control. Most of them have the same tastes as their offspring even if there is a bit nuance. They spoil them by giving much liberty in following the virtual world which affects badly the youth. As a matter of fact, the advance of technology, above of all in the XXI century, has deteriorated the whole body of social and moral norms of Soninkara. Things have moved from bad to worse because of too much freedom. It has enhanced ill-speaking, calumny and debauchery. As a result, people have serious moral issues. The mutual respect and the sense of responsibility have nearly disappeared. Everyone is the chief of themselves following their passions. It is licentiousness and anarchy which have replaced order and stability. The contemporary Soninkara is characterized by the deterioration of the good values. The values which give people a good personality, guarantee of social peace and stability, are declining nowadays. Most of people above of all amid the youths are disrespectful, arrogant, rude and selfish. These discipline issues are jeopardizing the harmonious and cohesive life in modern Soninkra. --- Conclusion Since the French colonization in Africa, Soninkara has been experiencing many social mutations which are deteriorating the values that underspin a cohesive and harmonious societal group. The arrival of the French colonialists caused the disappearance of the Soninke kingdoms and entailed a great social disorder. It is the mother cause of the social mutations. In addition to that, we have migration and modernism which have badly affected the Soninke way of life. They have jeopardized the common life which used to symbolize unity and oneness in Soninkara. Nowadays, most of the Soninke people have discipline issues that have endangered the order and stability of the society. They have also brought about social conflicts between the classes and the generations. There is no understanding between people of the modern Soninke society because of the fragmentation of the social and moral norms that used to guarantee order and stability. --- Conflicts of Interest The author declares no conflicts of interest regarding the publication of this paper. --- 47 The old people, nowadays, are shocked with the irresponsible behavior of the youths and their lack of good morality.

In this paper, the author aims to study the causes of the social mutations of the Soninke society form the colonial epoch to the present day. In fact, we have noticed that the western colonization which symbolizes a huge cultural aggression against the African societies in general, has brought about big transformations in Africa in general and in Soninkara particularly. Since this historical event, the society of the Soninke people has been undergoing a huge cultural influence of the whites through the capitalist system, the French school and the modernism. The traditions and the way of life of the Soninke people keep changing under a bad contamination of the modern Western civilization characterized especially by liberty (freedom of speech and opinion) and a so-called equality between the individuals. And things have started falling apart in Soninkara hence the social conflicts and the loss of good values.

INTRODUCTION HCI has begun to address the design of digital technologies for justice [17,20] in a number of different settings such as street or workplace harassment [5,15], and the potentials of anti-oppressive design [56]. There has also been a movement in the literature towards topics of sexuality [60], pornography [55,69], and sex work [59,61]. This paper sits within these converging literatures, as well as alongside sex work research from other disciplines, to build a nuanced understanding of the ways in which digital technologies can be used alongside other forms of service delivery to advance and promote social justice. We premise our understanding of sex work from the communities that engage in it and build on existing literatures that recognize sex work as a type of labour that should not be criminalized, but rather protected by labour and other relevant laws that promote human rights. Carol Leigh, feminist and sex worker rights activist who coined the term 'sex work' in 1987, explains that the term "acknowledges the work we do rather than defines us by our status [as a sex worker]". Motivated by her "desire to reconcile [her] feminist goals with the reality of [her] life and the lives of the women [she] knew", her activism worked to create an "atmosphere of tolerance within and outside the women's movement for women working in the sex industry" [40]. In its current context however, the term sex work is used to refer to an activity practiced by people of all genders. In this paper, we reflect on the use of digital technologies for service delivery within a peer-led sex worker rights organisation called Stella, l'amie de Maimie. After an overview of the organisation, we focus our discussions on the Bad Client and Aggressor List, which is central to their services. This tool was, and continues to be developed, through peer reporting and aims to provide information for sex workers in Montréal about potentially dangerous individuals. The contributions of this paper are twofold: we contribute to the growing debate around using HCI for social justice. While there have been various interpretations of this, there has yet to be an analysis of the ways in which digital technologies could facilitate engagement with alternative narratives of justice, particularly in settings where workers may be criminalized. To address this gap, we provide implications for design framed in Fraser's idea of multidimensional and 'abnormal' justice that will support the development of digital technologies for settings where restorative justice may be prioritized. This is a particularly timely contribution based on current political, social, and criminal justice debates at national and international scales related to wider issues of nationalism, racism, or the prisonindustrial complex. First, we contextualize our work in HCI literatures, Canadian legal structures, and Stella's organizational practice. Second, we describe our methods and outline how service delivery relates to restorative justice. Third, we develop three implications for design aimed at researchers seeking to develop technologies that supports service delivery with and groups that are stigmatized or criminalized. --- BACKGROUND AND RELATED WORK HCI has been conceptualising justice through the development of a framework for Justice-Oriented Interaction Design [17], social justice in UX [48], the connections of storytelling and social justice [68], or implications for HCI specifically related to sex work [59]. While the debates surrounding justice and HCI have been useful in laying the groundwork for the relationship between 'justice' and HCI, we believe more nuanced discussions of what 'justice' actually means in relation to digital technologies, and the ways in which humans interact with it are needed. In this paper we focus on this, and how it relates to the consideration of restorative justice necessitated by engendering identities that are stigmatized or are made marginal by other socio-cultural means. We hope to address part of this gap in the literature by using Fraser's ideas of three-dimensional justice [22]. In this framework, justice is seen as a constantly evolving process that works towards a more just world on three levels. These relate to three questions that can be simplified to the following: What does justice look like? How can we move towards this idea of justice? And who decides what the answers to these two questions are? After describing the nuances of these questions and their meanings, there are instances where institutional ideas of justice are incongruent with what those affected by these frameworks consider 'just' -Fraser calls this 'abnormal justice' [23]. We must also acknowledge that "social justice is not an outcome of design in itself" [59], but the processes, as well as the wider work of research collaborations involved in these designs in and of themselves are also seen as part of this 'social justice' outcome. We want to also raise the importance of technologies that are useful for research purposes and wider civic and rights contexts, to move towards an understanding of civic design [16], and more thoughtful engagement with Third Sector Organisations [60]. We do this by bringing to the fore the importance of 'just sustainabilities' which "demand new ways of accounting for difference and inequity at the societal scale as cornerstones of truly sustainable design." [17]. Furthermore, it is important to not only engage in respectful and ethical, as well as trusting [12] conduct, but also to ensure the sustainability of these projects in different ways [17]; to engage in holistic explorations of the research collaborations as justice-oriented within which support organisations activists, researchers, and others work. Bringing Fraser's framework into conversation with HCI literatures, we learn to foreground collaboration and collective, situated work to design technologies with communities in mind and with differential understandings of justice -to collectively answer Fraser's three questions not only of wider political structures, but also of our research in and of itself. Furthermore, using multidimensional justice [21,22,71] provides us with a way of unpicking what we mean with 'justice' in HCI and how it relates to wider socio-legal structures and political frameworks. Using this lens to look at sex work specifically, we learn that Canadian sex work laws can be interpreted as an example of abnormal justice: where the government uses criminal law to address sex work, claiming that criminalization of sex work will protect sex workers , while organisations run by sex workers, like Stella, argue that protection requires removing criminalization . In fact, it is well recognized by social justice movements fighting for the decriminalisation of sex work, that the criminal justice approach is not a way of achieving justice for sex workers [3,37]. Instead, justice for sex workers is seen as being able to work free of the threat of police repression, criminal and other convictions, violence, discrimination, and stigma. --- Sex Work, Support, and Technologies Like other industries, the sex industry, and practices of buying and selling sex have evolved alongside societal developments, perhaps most importantly technology [32,52]. Although sex workers are often seen as being CHI 2019 Paper Paper 652 marginalised in society and "hard-to-reach", in regards to technology sex workers have been found to "represent a unique demographic for high technology penetration, [having] multiple devices per person, and intensive usage in their everyday practices" [50]. Sex workers have moved online to advertise or provide their services [14,51,52] and despite legal frameworks which criminalize their work, they are making use of digital technologies such as social media in innovative ways [32,52]. Cunningham and Kendall [14] raise important legal and regulatory questions surrounding the advertising and exchanging of digitally mediated sexual encounters for this growing online market that incentivises reputation-building as well as screening practices. Furthermore, digital technologies have been designed to support peer-sharing [49], to track health information in developing contexts [67], or to support sharing of safety information [59]. What may be lacking in the current research however, is the development of digital technologies with sex workers directly that take into account their agency and skillsets. The interplay between change and control in developing potential futures with digital technologies in organisational contexts is vital in engaging this more nuanced approach [63]. This is because these digital technologies and infrastructures can themselves generate new infrastructure to challenge wider existing structures such as legal contexts. Technologies are also scalable, and possess upward flexibility; providing us with new opportunities for sex work support services in rethinking organizational control [63] or potentials for justice [59]. One example of this is the creation of a Sex Work Database in Canada. This database brings together "academic research, print and visual media, grassroots activism, and commemorative responses related to missing and murdered women and sex work" and functions as an activist archive that brings together documents produced by sex workers that deliberately assembles "an anti-colonial feminist argument that highlights marginalized voices, and embraces principles of social justice and reciprocity" [19]. Learning from this collaborative project, we see that technologies are not only built with embedded values [24], but also that these can support wider political struggles -in this case the 'tagging' of archived documents was seen as activism for sex worker rights [19]. The technological context for the sex industry and the capacity for sex workers to use technology in their activism and service delivery will vary by region and is impacted by the legal context for sex work. In Canada, sex work is criminalized. In 2014, the Conservative government implemented a 'Swedish inspired' legal regime that made the purchase of sexual services illegal, and also criminalized advertising, material benefits , or procuring. They also made changes to the communicating law, which effectively criminalized the exchange of sexual services for the first time in addition to communication and third party involvement. There are a myriad of academic and nonacademic debates surrounding 'what works' when it comes to regulating the sex industry; but many sex workers and allies would support, and campaign for a decriminalized approach [2,10,37]. Further to this, the Canadian Alliance for Sex Work Law Reform state that in addition to removal of all criminal laws against sex work, "[e]xploitation in the sex industry can be addressed using a labour framework that engages provincial legislation related to public health, occupational health and safety, and employment law" [10]. Ultimately, laws impact on the ways that sex workers can share information, communicate about potentially dangerous individuals, engage with clients and employ vital safety strategies in their work. --- Stella, l'amie de Maimie Stella, l'amie de Maimie is a sex worker led organisation and registered charity that provides space and support for sex workers in Montréal, Canada. The organisation was founded in 1995 as part of a HIV-related public health participatory action research project that placed sex workers at the centre of HIV prevention [13,62]. Members of the organisation also played a major role in mobilizing additional narratives and communities in the Bedford vs. Canada legal case [72], which delivered a landmark decision declaring three of Canada's most commonly used prostitution laws as unconstitutional, and through that recognized the human right of safety and security for sex workers. As the organization is made up of a majority of sex workers, Stella's team brings unique knowledge and strategy to fighting violence against sex workers. On their website, Stella state the following as their primary mission: "to improve quality of work and life for sex workers, to educate the greater public on the different ways that sex work happens as well as about our lived experiences as sex workers, so that sex workers might also enjoy and benefit from the same rights to safety and security that are commonplace for other people." [57] Stella works towards this mission through service delivery and activism, underpinned by "solidarity amongst sex workers and by creating spaces where sex workers can access power." [57]. --- CHI 2019 Paper Paper 652 2.2.1 Service Delivery. Stella provides a number of different services to reach their goals, and integrates a rights based approach into everything they do. To further not only their own goals, but also wider-reaching goals of the sex worker rights movement, they build local, national, and international networks and collaborations. Stella produces an eclectic yet unified image of the organisation through their use of artefacts and publications. Seeing these as an artefact ecology, allows us to move beyond understanding the objects as physical artefacts with some level of digital interaction, and instead supports us in considering the ways in which people interact with them in different contexts of everyday life [7,33]. Stella use artefacts such as condoms, crack pipes, or publications that relate directly to their organisational goals. For example, small cards that Stella created to increase sex workers' knowledge about their rights and legal context were designed and formatted intentionally for sustainable use: they are sized like business cards to fit discreetly into a small bag and can easily be passed on to others. The language that is used on the cards to disseminate legal information is easy to understand, colour-coded, and translated into four languages. Working alongside Stella staff to analyse their use of this artefact ecology [7] helped us identify different uses for the artefacts, especially when discussing their political nature. We use this analysis to better understand the use of the Bad Client and Aggressor List described below. --- 2.2.2 The Bad Client and Aggressor List. When sex workers experience violence on the job, they are able to fill out a short form where they are asked to describe the incident and alleged perpetrator. Sex workers are able to report such incidents with Stella through a number of channels . Following this, staff remove any identifying information about the sex worker, and write a brief but detailed description of the alleged perpetrator, which is then added to The List. As such, The List is made up of edited versions of informal reports from sex workers about incidents with presumed clients which either move beyond their agreed boundaries, involve violence, or disrespect. Often these experiences are shared amongst the community, so The List functions both as a warning system and to promote solidarity. To share this information among sex workers, it is incorporated into the monthly Bulletin created by and for sex workers in Montréal. The bulletin also contains many other pieces of information on services available and activities for sex workers, as well as a regular advice-column written by a well-known sex worker columnist. The bulletin is printed and shared in the drop-in centre and on outreach, and is also e-mailed to sex workers and other organizations in the area. Staff were interested in finding out how they could improve this service by considering the use of digital technologies to collect and share this vitally important information to increase its usage and reach. Sex workers have been sharing this kind of information informally for as long as they have been doing their work. Penfold et al. found that inter-agency working supported through a similar system resulted in increased reports of violence in the UK [46]. Bringing together this learning with digital technologies, Strohmayer et al., have explored the use of digital technologies by a UK-based charity to carry out a similar kind of reporting and alerting process [59]. Learning from the work carried out in the UK, we reflect on the current use of non-digital technologies, and have also taken into account the implications for design as outlined by [59] to imagine digital futures with Stella. --- METHODS --- To foreground collective knowledge-building surrounding Stella service delivery and the potential of integrating digital technologies to facilitate a movement towards multidimensional justice [21,22] with and for sex workers, we used a Participatory Action Research framework [9,36,47]. Stella staff were involved in the development of the overall research questions, the methods, interview and workshop schedules, the process of analysis, and writing a charity report as well as this paper. Our fieldwork took place over 3-months at Stella from April to June 2018. We carried out 3 interviews, 3 workshops, observations, a collaborative analysis of the artefact ecology produced by the organisation, and various informal chats with members of Stella staff. The majority of data collection took place in English, though some discussions in the workshops took place in French and were later translated into English by the authors. Audio recordings from the interviews and workshops were transcribed and analysed using thematic analysis [8]. We have loosely categorized the work into three distinct but interconnected stages of research: sensitization of the first author within the organization to contextualize the political and social implications of our three questions of justice within the context of Montréal ; collaborative artefact ecology analysis to better understand the ways in which Stella creates and utilizes key artefacts for service delivery and activism ; a CHI 2019 Paper Paper 652 series of three workshops with staff to discuss in detail the processes and experiences around the Bad Client and Aggressor List, focusing in particular on staff experiences with digital technologies. The workshops each included a diverse group of roughly 12 members of staff including outreach workers, communications staff, and management. Each workshop was based on a flexible schedule of activities and concentrated on different aspects: understanding the information flow involved in producing and sharing The List through a card-based mapping exercise and discussions; understanding the form used to collect this information by reflecting on the existing form in small groups prior to a group discussion based on staff experiences of using the form; and potentials for novel interactions using design fictions that were developed based on the analysis of the two prior workshops and other data collection as a way of facilitating discussion around digital technologies, justice, and the future of service delivery. This paper focuses on the data collected through the workshops, but the researchers' prior experiences and staff's historical understanding of the organization contextualizes this data; affecting the ways in which we understand, interpret, and analyze the data. --- SERVICE DELIVERY AND RESTORATIVE JUSTICE Restorative justice, though sometimes understood simply as an alternative to the criminal justice system, is defined as "an approach to justice that focuses on addressing the harm caused by crime while holding the offender responsible for their actions, by providing an opportunity for the parties directly affected by the crime -victims, offenders, and communities -to identify and address their needs in the aftermath of the crime" [34] by the Canadian department of justice. The term 'restorative justice' was often used by staff when talking about the ways in which their work fits in to wider justice debates, as well as sex worker rights activism. They also discussed The List specifically as an element that contributes to the restorative justice of sex workers who have experienced violence. While it does not provide a space for perpetrators to be held accountable, it does provide a space for sex workers to seek and implement protections in a context where they themselves are often sought out as criminal, where they are not provided with a context to restore the injustices they experienced. It also provides an opportunity for the 'victim' and the 'community' to identify and address their needs in the aftermath of violence [34]. On top of this, restorative justice is built on principles of "respect, compassion and inclusivity" to encourage "meaningful engagement and accountability and provides an opportunity for healing, reparation and reintegration" [34]. Looking at this then, we see that The List provides an opportunity for sex workers to create alternative forms of reporting violence, in a context where the justice system too often either rejects sex workers' experiences of violence or does not account for their realities. As it does not provide a space for perpetrators to be held accountable though, we argue that instead of seeing The List as a representation of restorative justice in its full form, we see it an example of an alternative approach to justice seeking, based in part in the ideals of restorative justice. As shown above, justice correlates to the ways in which Stella work, and how, as one member of staff said: "everything we do is activism, our existence is the revolution." This was said in a joking way, while also maintaining an air of seriousness. What it does however, is clearly show the link between service delivery and activism in the organisation. Service delivery feeds into activism, and vice versa, while also maintaining them as distinct. For example, Stella's weekly health clinic is a direct way of delivering non-discriminatory and anonymous medical services, and is appreciated as such by sex workers from all parts of the industry. At the same time however, and while understanding that creating separate and isolated services for sex workers is not necessarily the end goal, its existence is a form of activism, as this is the only place where sex workers can receive anonymous and non-judgemental health services in Montréal. This kind of service delivery then becomes, in itself, part of a process of restoring justice to the lives of sex workers. At the same time however, those using the services do not necessarily see themselves as engaging in activism or a revolutionary act -they are attending a sexual health clinic. In this way, the organisation engages in a kind of prefigurative politic, or what we term 'tacit activism', that is embedded and necessary, implicit in the actions taken to, in this case, deliver services. Below, we address in more detail how operating in a context of criminality affects the ways in which service delivery, and The List specifically, contributes to creating alternative forms of justice for sex workers. --- Operating in Criminality In Canada and many other parts of the world, sex workers operate within a legal system that delegitimizes and criminalizes their work. in and of itself is "profoundly, profoundly political", and "a very good example of a restorative justice approach, because it's an alternative way of dealing with crime against a person" and seeking justice for sex workers. The List operates in a system of "abnormal justice" [22] while preventing violence, redistributing power to sex workers, and engaging community. All of these are deeply political acts, and demonstrate the need for alternatives to institutional justice seeking. Most academic discussions and literature around tools such as The List or the Ugly Mugs scheme [46] assume they are developed solely for the purposes of harm reduction or violence prevention [43,59]. Our conversations with the team at Stella however, made it clear that there are multifaceted and more complex reasons why sex workers use such peer-communication tools. Here, we describe only a few of these reasons: violence prevention, recognizing agency, affirmation, and community communication. --- Violence Prevention. One of the core reasons for The List is to prevent violence perpetrated against sex workers -a way to share information, to help sex workers avoid particularly dangerous individuals. One participant reminds us that the collected information must be useful to sex workers when she asks: "Would it help prevent someone seeing a client?" or would the information and details collected help identify a client, an aggressor, or a specific situation? Even though The List is a tool for violence prevention, Stella staff understand that there may be barriers to this, and that the reports do not always result in sex workers avoiding a particular client or violent situation. [38,59]). Other times, they may also report the violence in online forums or through social media channels. Using The List allows peer-to-peer reports where sex workers attempt to prevent violence with each other. They must do this 'for themselves' as it is not always something they can rely on from others outside of the sex industry. Particularly, where police may reproduce stigmatizing treatment to sex workers who want to report, this channel for community communication really allows sex workers to communicate with each other and prevent violence, or as one participant stated: "it's really to get that power and give back to somebody else so [aggressors] cannot be harmful." This individual power, when collectivized in community via a widespread communication like The List, then becomes important not only for the individuals making the report and those reading the alert, but also for the community as a whole. This also highlights that The List is a sex worker led, community initiative, and that this is seen as central to its success. --- Humanity in Service Delivery The Bad Client and Aggressor form and list are used by sex workers who come to Stella and by staff to fulfill all of the above aims. It is part of an ecology of service delivery, where human interaction is essential. The ways The List is formatted and distributed is essential to consider, and the Stella team made clear the importance of people within this process. Here, we relate this humanity to human interaction, care, trust, or other related 'human' elements of service delivery. Using The List as a communication tool is equally important to its distribution. When we look at The List beyond its existence as a tool, and instead see it as part of a wider ecology [42], we see that human contact is the start and end-point of the production of the monthly list. In- , you know." Furthermore, there may be situations where the ways in which questions are asked towards the people who have experienced violence may awaken previous traumas, so the interaction must not only be on a caring and human level, but must also be trauma informed [28]. One member of staff said that an outreach worker cannot just ask a person whether they have experienced violence because "it can awaken all kinds of things for the person." The List, or any other digital innovation that may carry out similar work, cannot only be a tool for violence prevention. Instead, they are part of an ecology that supports the facilitation of connections, relationships, and human interaction. While human contact is important to the use of The List, Stella staff does not presume that sex workers want to discuss the incident or engage in follow up interventions such as counselling or completing a police report. When an incident is reported to a member of staff, they are trained to engage in conversations with the individual, to ask them questions such as: "Do you want to press charges? Do you want us to accompany you all the way [through the criminal justice system]? If they want, that's part of our job." Many however, will not want a follow up or an accompaniment from staff, even if they fill in a report form. The choice to request or decline further support from Stella or other organizations must be respected "because we can't assume that everyone wants interventions [or support], if it's just a report." --- Posting Information Online Much of our discussion was focused around putting the information shared via the Bad Client and Aggressor List online, as well as how we could design novel technologies to make it more useful. Privacy risks were often flagged around this, and one participant stated being online "risks identifying workers." This shows how anonymization becomes important in a context where service users may be criminalized, particularly when designing digital technologies to support peer-communications akin to The List. But ultimately the following statement from a member of staff brings the importance of technologies to the fore: "we cannot say [online technologies are] not an option. [They are] an option, definitely, because that's where it's all going." This understanding lead to wider discussions of technologies in society, and the ways in which sex workers use and appropriate them; smartphones are becoming more affordable and available to sex workers in all parts of the industry, internet access more ubiquitous, and peer-alerting networks through forums and social media are being used. Stella currently use some digital platforms and technologies to share information and communicate with sex workers. For example, Stella staff uses a mobile phone application to communicate with a group of sex workers, but at the same time, it is understood that this particular app is "not [a] community for everyone" and that not all online applications work for every sex worker or sex working community. To be able to access mobile applications, sex workers "need to have internet access, a cell phone or laptop, and not everyone has that." The Bulletin is also e-mailed to sex workers and distributed by other organisations. Our discussions on innovative technologies concluded that regardless of whether information is posted online, remained in the current paper and PDF format, or whether we created a hybrid form of these two options, we needed to ensure sex workers were able to obtain, read, store, and use this important information. In discussing this, it reiterated the Stella mandate, that the "inclusion and diversity of sex workers" is essential to service delivery. While online service delivery may be useful in some ways, it cannot replace existing practices "because of the sex worker on the street […] Having a more complex understanding of the access to and use of digital technologies beyond the artificial dichotomy that street sex workers do not have access to these technologies and that escorts do, would allow us to build more useful tools for service delivery. This more complex understanding would provide us with a reason to diversify not only the kind of information that is received and shared, but also the ways in which it is received and shared for a potentially digitally enhanced Bad Client and Aggressor List . One way of building this necessary but complex understanding is to look beyond traditional boundaries and explanations of the different areas of the sex industry. Instead, we can build ecologies of understanding that take into account multiple realities and mixed accessibility to digital technologies. These should not be based on place of work as is traditionally done in sex work research , but instead could be considered in separate but connected areas such as: place of work, digital infrastructures, and access to hardware and software. Instead of seeing these three things as entirely separate from one another, or that one implies the other, we argue that we must look at these three areas together and with an intersectional lens that accounts for the different positionalities and experiences of sex workers. --- IMPLICATIONS FOR DESIGN Based on the above findings, there are a number of different opportunities for future developments in the use of digital technologies to support the ongoing use of the Bad Client and Aggressor List and similar justice-oriented tools for service delivery. Designing for alternative forms of justice is incredibly complex, but also necessary for the future of the design of digital technologies in social, third sector, or civic contexts. As we have demonstrated throughout this paper, digital technologies can play a part in these justice-oriented collaborative efforts in supporting the ongoing labour of volunteers, staff, and sex workers accessing services. In the case of this paper, this has related primarily to violence prevention, solidarity, or rights advocacy. Looking at these three areas in particular, digital technologies can support the work in collecting, sharing, and using information especially when this is collected and contextualized by human interaction. This human interaction may be digitally-mediated, but as we argue below, should not be replaced with novel digital technologies. Instead, we provide implications for the design of these within a framework of restorative and social justice seeking. Using the Bad Client and Aggressor List as a starting point to reflect on the kinds of technologies that could be useful, we have developed three implications for design: technologies need to be adequately contextualized; the need for multiple formats and types of service delivery to reach as diverse an audience as possible; and a recognition that technologies in and of themselves will not be able to solve complex issues of calls abnormal justice. We believe that these implications will support meaningful engagement to design digital technologies with support services, or others engaged in justice-oriented work. --- Contextualizing Technologies for Justice When designing technologies for social justice, we must ensure we adequately understand the contexts in which we design, including but not limited to the social, historical, political, and legal circumstances. To do this, it is helpful to keep in mind the three questions Fraser [22] poses to build a more just world: What is justice? How is it decided what this justice looks like? And who decides the answers to these two questions? Looking at these questions, one must understand the contexts not from privileged positions as researchers and designers, but rather from the position of those one is designing and innovating with and for. In some ways, participatory design and related research approaches may be useful for doing this. For example, Kensing and Blomberg have analyzed the ways in which participatory design relates to issues of politics of design, participation, and methods across personal, organizational, and national levels [35]. Others expand the discussions of PD to public rather than work-life [6] . PD is often design-focused, the explicit motivations of such work however also exists to "strengthen workers' control over their work lives and to create more democratic work environments" [6]. This paper does not discuss PD directly, but we believe that as a growing community of justice-oriented HCI researchers, CHI 2019 Paper Paper 652 we can learn from Beck [4], Irani et al. [29,30,31], Björginsson and Ehn [6] or others to explore the politics in our research and to develop digital practices and 'things' that support workers' engagements with political processes on personal, organizational, and national [35] levels. Following these scholars, we also encourage researchers working in wider justice-oriented research approaches to HCI to also consider the political and activist potentials in their work. The following questions may be useful for reflection: how can the participatory processes affect not only the lives of those directly involved, but also those associated with those individuals? Who is not participating, and how does their absence affect the project? Seeing technologies for justice within the sex industry specifically as an example for this contextualization, we look into the ways in which legal frameworks in particular can affect research. Not only have Cunningham and Kendall raised legal questions for online markets associated with sex work [14], but with the introduction of laws specific to advertising in a Canadian context, and similar laws implemented elsewhere in the world, we must consider what kinds of technologies are designed. We have to consider how they sit within existing and evolving legal frameworks, and the ways in which they either support or counter these developments. While working within institutions such as universities or NGOs requires us to do work that is legal, we do question to what degree we are able to subvert the legal status quo, to move away from the existing abnormal justice [22,23,41], and instead move towards systems that are just . We urge researchers and designers to question political and legal structures that maintain systems of abnormal justice and ask them to not be afraid of disobedience to these systems when necessary. While seeing digital technologies or platforms as tools for translational service delivery [45] or citizen-led developments [25], the collection and dissemination of information related to alleged crimes or in stigmatized, marginalized, or criminalized communities, brings about particular necessities. Here, we must ensure that whatever digital technology we design is more than 'a tool' and instead see it as part of an ecology [42] that is based in social, historical, legal, or ethical contexts as well as personal experiences of those that are part of the ecology itself [29] and wider community [54]. To do this, we must understand the ways in which humans interact with it, how it fits in with other existing digital and non-digital service provisions, and how it sits within particular social, political, historical, and legal contexts. --- Multiple Formats and Shifting Paradigms We also argue that when designing digital service delivery, we must understand that one approach will not work for the complex and interconnected ecology of existing services, service delivery, and heterogeneous experiences of individuals accessing these services. When designing technologies for restorative or social justice, digital platforms, tools, or ecologies may be useful in some ways, but we also must ensure that we do not exacerbate or amplify the digital divide [64]. Rather than unifying services, we argue for the need to diversify service delivery to ensure diverse groups of individuals are reached, but also to allow for people with different degrees of access, interest, time, or money to be able to make use of the pieces of information that they feel is important and useful to them. To provide services that empower their users to make decisions about their own needs [18], and to ensure accessibility for different parts of the community. Our point of view seems to be in direct contrast with some current trends in HCI to build generalizable technologies and platforms that can be easily translated to different contexts and countries, but correlates with other spaces of HCI that relate to designs with ideals of justice at their core . While building a digital platform to collect and share information about potentially dangerous individuals on a national level may work in some countries such as the UK [59], it is important to acknowledge that this may not be translated to other contexts easily. For example, looking at the potential to design a digitally mediated national Bad Client and Aggressor List for touring sex workers in Canada, there are a number of immediate legal concerns , as well as pragmatic issues , or risks associated with such digital tools. These risks and issues become exacerbated in spaces where community members participating in any design work or research are structurally disadvantaged through stigmatization, marginalization, or criminalization. When exploring both the importance of designing for different parts of a community, and the trend of designing globalized technologies, we argue that the humanity of service delivery must not get lost. Ultimately, we urge designers to design for particular communities, in an informed and respectful, ethical, and just way, rather than attempting to design all-encompassing generalizable digital tools that aim to solve complex issues. In relation to designing with and for sex workers, this might mean CHI 2019 Paper Paper 652 moving away from designing technologies only to protect or reduce harm to sex workers and instead work towards the normalisation of sex work as a design space by designing technologies for sex workers' unique business models. In turn, changing the design paradigm in this way could help tackle the stigma and abnormal justice endured by sex workers, which are propped up in some ways by the focus on globalised and protective technologies. --- Technologies are not Solutions Building on the importance of adequately contextualizing technologies, and advocating for the use of multiple formats of service delivery, we now also want to address HCI's tendency to assume that technologies are able to solve complex issues. In this paper, we have described multiple uses and purposes of a particular technology as a way of imagining processes that are more just for sex workers experiencing violence. We argue that The List has been successful in achieving its many purposes exactly because it does not strive to solve the problem of violence, but rather because it is recognized as an intervention that can support the ongoing battle for sex workers rights. It is pragmatically, aesthetically, and emotionally situated within Stella's aims; embracing the humanity and peer elements necessary for The List to do its work [57]. Similar to The List, finding new ways of communicating among groups that are forced to use underground channels is invaluable for these same communities to thrive; this takes place alongside technological and legal developments [14]. What is imperative when designing digital innovations that aim to support these existing channels of communication however, is that we must ensure that the original purpose of these often-analogue systems cannot get lost [65]. For example, in the case of the Bad Client and Aggressor List, the focus on informing, communicating, and empowering must remain. This is similar but distinct to the implication that technologies in sex work support services should aim to facilitate the fighting of stigma related to the industry [27,59], regardless of what kind of digital innovation we develop. This is translatable to many technologies we wish to design within the context of restorative or social justice. Again, raising Fraser's three questions [22,41], we believe that thoughtfully answering the 'what, how, and who' of justice in our research spaces will lead us to genuinely take into consideration rich accounts of the context in which these may be designed. To do this, we should reflect on our privileged perceptions as researchers, and instead foreground those adversely affected by abnormal justice [23]. Building robust and interdependent relationships with the communities we aim to support can help inform this broader awareness of the politics involved in the designs and engagements. Using a participatory framework could also allow us to advocate for change in political and legal structures that build the context within which these designs are created. Through this, we can then use design processes with the affected communities as a way of pinpointing routes towards and enacting genuine political change to tackle the injustices at their roots, rather than designing technologies in an attempt to rectify some of the symptoms of abnormal justice. While we have a rich history of participatory action research and design in HCI and related fields , we believe more nuanced justice-oriented research and methodologies must be developed alongside organisations, groups, volunteers, or workers who are embedded in the design space to be able to meaningfully innovate [39]. To do this, it is important to be in constant communication and collaboration to ensure the context, histories, empowerment, and community that are so necessary to make such technologies useful remain at the center of the innovation. We must not replace existing communication strategies, but rather we need to ensure the developments make sense in the immediate ecologies within which they are placed. --- CONCLUSIONS This paper highlights findings from a participatory action research project between two universities and a sex worker rights organization in Canada. Together, we reflected on the organization's existing use of digital technologies for service delivery, and also imagined possible digital futures. Framing our work in Fraser's ideas of multidimensional justice, and particularly her idea of abnormal justice, we use the learning from this particular example, to develop three implications for the development of digital technologies with, in, and for communities who are often misrepresented, stigmatized, or criminalized. Bringing together our three implications, we argue that nuanced and justice-oriented design of digital technologies can be made possible if we start to see technologies not as solutions to complex social problems, but rather as aides that can support the humanity of service delivery and the people who engage in this kind of work. By developing tools that are multifaceted in themselves, and developing multiple of these technologies for different audience we are able to develop services that cater to the needs of individuals while simultaneously being useful in working towards justice for the often stigmatized service users. Ultimately, we stress the importance of people not only in the development of digital service delivery, but also in the delivery of these services, as well as their continued adoption and adaptation of use.

This paper provides analysis and insight from a collaborative process with a Canadian sex worker rights organization called Stella, l'amie de Maimie, where we reflect on the use of and potential for digital technologies in service delivery. We analyze the Bad Client and Aggressor List -a reporting tool co-produced by sex workers in the community and Stella staff to reduce violence against sex workers. We analyze its current and potential future formats as an artefact for communication, in a context of sex work criminalization and the exclusion of sex workers from traditional routes for reporting violence and accessing governmental systems for justice. This paper addresses a novel aspect of HCI research that relates to digital technologies and social justice. Reflecting on the Bad Client and Aggressor List, we discuss how technologies can interact with justice-oriented service delivery and develop three implications for design.• Human-centered computing~Human computer interaction (HCI) • Human-centered computing~HCI theory, concepts and models

Background Research at the national level in the United States has shown a significant increase in obesity prevalence, particularly over the past two decades [1]. In California, there were more than 7 million obese adults and adolescents in 2011-2012. Overall, the prevalence of obesity in adults has increased by approximately 32% from 2001 to 2011-2012 . The increase in the prevalence of obesity has been observed in Latino, White, African-American, and Asian adults during this period. Moreover, compared to Whites in California, the prevalence of obesity was higher among African-Americans, American Indians, and Latinos [2]. Obesity has been identified as a major risk factor for multiple chronic conditions such as diabetes and cardiovascular disease [3,4]. However, important variation among racial/ethnic subgroups would be ignored if we only focus on the overall population. It has been suggested that the prevalence of and influence of obesity among subpopulations are very different according to races/ethnicities [5]. Some racial/ethnic groups have been more influenced by obesity than others. In the U.S., from 2011 to 2012 to 2017-2018, the obesity prevalence was highest in Non-Hispanic Blacks , followed by Hispanics , non-Hispanic Whites , and non-Hispanic Asians [6,7]. National Surveys [8] show that Asian subgroups have obvious ethnic differences and dramatic increases in the prevalence of obesity, suggesting that these subgroups warrant increased monitoring of obesity and its related risk factors. In California, racial/ethnic differences in body mass index were also observed among adults for both sexes [9]. The reasons of racial/ ethnic disparities in obesity have been explored and have focused on the important roles of a variety of socioeconomic, demographic, and behavioral characteristics [10]. For example, compared to Californian Blacks and Hispanics, Californian Whites had lower poverty rates and higher education levels. However, the reason as to why Asians have lower BMI rates than Whites with adjustment for covariates remains unclear [9]. A number of studies have examined the sex and racial/ethnic disparities in obesity [11]. However, evidence is very limited with regard to differences in obesity rates among Asian Americans; further most of the few existing studies used samples that were multiracial, and therefore may not be generalizable to Asian Americans [12]. Due to lack of data or small sample sizes for Asians in most studies of obesity or BMI, Asian Americans were often excluded, or when they were included, they played a small role in study results [9,13]. However, it is best not to treat Asian Americans as simply one group because they are highly diverse in national origins [14] and socioeconomic status [15] . Furthermore, the prevalence of health conditions in Asian adults such as cardiovascular diseases [16,17] appears different among subgroups. Because of such diversity in multiple dimensions related to obesity, it is important to identify subgroups of Asian Americans at high risk of obesity to help develop targeted interventions for those subgroups. To address the above-mentioned gaps in the literature, we examined the disparities in the patterning of obesity among Asians . We also brought insight into the role of a number of characteristics in predicting obesity among Asian Americans. Because there is a large size of Asian population and Asians are highly diverse in California, it is warranted to understand these disparities and the determinants of obesity as well as associated policy challenges. Consequently, public health resources could be aimed at those most at risk. --- Methods --- --- Study variables Obesity BMI was calculated as weight divided by height squared for self-reported data. Weight status was determined using the World Health Organization Asian BMI cut points in Asian groups as 4 categories to account for racial differences in body fat percentage at the same BMI level: < 18.5 kg/m 2 , 18.5-22.9 kg/m 2 , 23-27.5 kg/ m 2 , and ≥ 27.5 kg/m 2 [19]. --- Race/ethnicity We used the self-reported Asian ethnicity variable constructed by CHIS, which includes five categories: Chinese , Japanese, Korean, Filipino, and Vietnamese. --- Covariates Covariates included age , sex, family income , which adjusts for total household income and number of members in the household: below 100, 100 to 299%, or 300% and above of the FPL), smoking status , marital status , education attainment , physical activity , and fast food consumption . Fast food consumption was determined by response to the following question: "In the past 7 days, how many times did you eat fast food? Include fast food meals eaten at work, at home, or at fast-food restaurants, carryout or drive through" [20]. --- Statistical analysis The SAS PROC SURVEYFREQ procedure was used to weight and estimate population proportions. The Chisquare test was used to compare the prevalence of the study outcome across age, gender, and races and other factors. Specifically, the prevalence of weight status was examined according to race/ethnicity and sex. Demographic and lifestyle characteristics was examined among Californian Asians . The SAS PROC SURVEYLO-GISTIC was used to estimate odds ratios and 95% confidence intervals for the relationship between potential risk factors and obesity. First, weighted multiple logistic regressions were used to examine the association between race/ethnicity and obesity among Californian adults. Second, we used weighted multiple logistic regression analyses to examine the association between race/ethnicity and obesity among different subgroups of Asian Americans by sex. Variables with P values significant at or below 0.20 in univariate analyses were included in the final multiple logistic regression models. SAS version 9.4 was used for analysis and computation of weighted estimates for projection to the California population. --- Results --- Prevalence of obesity in adults of California Table 1 shows weighted prevalence of weight status in Californian adults by sex and race/ethnicity from 2013 to 2014. Overall, the prevalence of obesity was 11.1, 24.8, 33.5, 38.0, and 36.0% for Asians , Whites, Hispanics, Blacks, and other races, respectively; the prevalence of overweight was 32.6, 36.1, 39.0, 33.1, and 37.1% for Asians , Whites, Hispanics, Blacks, and other races, respectively. Using WHO Asian cut points for obesity, 23.3% of Asians were identified to be obese and 40.0% overweight; these prevalence estimates are much higher than those using standard cut points. Regardless of using either cut points for obesity, male Asians had a higher prevalence than female Asians . Among males, the prevalence of obesity was highest among other races , followed by Hispanics , Blacks , and Asians , and Whites . Among females, the prevalence of obesity was highest among Blacks , followed by other races , Hispanics , Whites , and Asians . --- Prevalence of obesity in Asian adults of California The demographic and lifestyle characteristics among Californian Asians with obesity are described in Table 2. Overall, the obesity prevalence was higher in Asians who were males , and were 45-64 years old . Among Koreans, and Filipinos, males had higher obesity prevalence than females . The highest prevalence of obesity was observed in age group 45-64 years old among Filipinos and Vietnamese as compared to that in age groups 18-44 years and 65 years or above . For smoking status, Japanese had lower obesity prevalence in those who were current smokers than those who were not . For marital status, compared to "other marital patterns" and "never married", the prevalence of obesity for those who were married was highest among Vietnamese . Obesity prevalence was lower in those with physical activity than those without physical activity in Vietnamese . The prevalence of obesity was particularly high in Chinese who consumed more frequent fast food . --- Weighted logistic regression analyses in California adults Table 3 shows results from multiple logistic regression analyses for the association between race/ethnicity and obesity among Californian adults. After adjusting for other factors, compared to Whites, being Hispanics and Blacks were associated with higher odds of obesity ; being Chinese, Korean, and Vietnamese were associated with lower odds of obesity . Furthermore, being aged 45-64 years of age , being male, having lower family income, married or other marital status, lower education level, lack of physical activity, and higher frequency of fast food consumption were associated with higher prevalence of obesity. When stratified by sex, among males, being Hispanic and Blacks were positively associated with obesity while being Chinese and Vietnamese were negatively associated with obesity; among females, being Hispanics and Blacks were positively associated with obesity, while being Chinese, Korean, Japanese, and Filipino were negatively associated with obesity. --- Weighted logistic regression analyses in Asian adults in California Table 4 shows results from multiple logistic regression analyses for the association between race/ethnicity and --- Discussion National surveys such as the National Health and Nutrition Examination Survey show that compared to other racial/ethnic groups Asian-Americans have lower obesity rates in the U.S. [21]. A recent study using CHIS has observed considerable ethnic disparity in adult obesity prevalence in California: African Americans , Latinos , Whites , and Asians . The prevalence in African Americans was three times higher than that in Asians [12]. However, it has been criticized that such national or state surveys have not considered the differences between ethnic groups in Asians [21]. This study assessed differences between the prevalence Abbreviations: CHIS California Health Information Survey, CI confidence interval, NA not applicable, FPL federal poverty level, OR odd ratio of obesity in Asian racial/ethnic groups, using definitions by standard cut points and WHO Asian cut points. It was found that overall the prevalence of obesity was 11.1 and 32.6% for overweight . In contrast, using the WHO Asian cut points for obesity, the prevalence of obesity was 23.3% and the prevalence of overweight was 40.0% among Asian Americans; these estimates are much higher compared to those using standard cut points. The reason as to why Asians have lower BMI rates than Whites with adjustment for covariates remains unclear [9]. For the same age, sex, and percentage of body fat, BMI is consistently lower in Asians than in Whites by about 2-3 kg/m 2 . This is because body build and muscularity may be different in these populations [22]. Furthermore, absolute or relative metabolic risk cannot be corresponded similarly in all ethnic groups using standard cut points for obesity [19]. Because of these limitations of BMI measure in Asian populations, lower BMI cut points for Asians have been proposed by WHO using all available data from Asian countries; for Asians, overweight was defined as a BMI of 23-27.5 kg/m 2 and obesity as a BMI of ≥27.5 kg/m 2 [19]. Due to the possibility that lifestyles could be very different between Asians living in their original countries and Asians who have immigrated to live in Western countries , whether these international guidelines for Asians are appropriate for Asian Americans has been subject to discussion [16]. Although there was a debate on whether Asian specific BMI cut points could be used for all Asians worldwide particularly in Western countries [16,[23][24][25][26], a number of studies have been conducted for addressing this issue; these studies found that, compared to other racial/ethnic groups, Asian Americans have lower rates of overweight/obesity [21,27,28], but they have a higher risk of type 2 diabetes, hypertension, and associated metabolic abnormalities [14,[29][30][31][32]. Thus, it appears not appropriate to use standard BMI cut points to examine the prevalence of obesity among Asians Americans, because its impact in these populations may be underestimated. Instead, applying the WHO Asian BMI cut points may have some benefits, e.g., 1) having better estimates of health conditions attributable to obesity [16,29,30] and 2) having significant clinical implication in identifying at-risk Asian Americans [33]. This study found that in California, the prevalence of obesity was 23.3, 24.8, 33.5, 38.0, and 36.0% for Asians, Whites, Hispanics, Blacks, and other races, respectively. After adjusting for other factors, compared to Whites, being Hispanics and Blacks were associated with a higher prevalence of obesity; being Chinese, Korean, and Vietnamese were associated with lower prevalence of obesity. This finding is consistent with a previous study that examined racial/ethnic disparities in obesity among adults in California using the 2003 CHIS, where the prevalence of obesity was highest in Blacks, followed by Hispanics, Whites, and Asians. Reasons that might explain racial/ethnic differences in obesity may be related to a number of factors that are correlated to both race/ ethnicity and BMI, e.g., different SES, demographic characteristics, and behavioral factors [9,10]. For example, Hispanics in California have lower educational attainment and higher poverty concentrations than other ethnic groups; Blacks have higher poverty rates than whites and more than a third have not attended college, and they are more likely than other groups to report no walking [9]. Among Asians in California, the obesity prevalence was higher in those who were males, and were 45-64 years old. It is interesting to observe a higher prevalence of obesity in males among Asians, which is different from that among other races. For example, previous studies found that overall there are more obese women than men. In developing countries , such disparities are more obvious among women [34]. In the U.S., there was no difference in overall obesity prevalence by sex in 2011-2012. However, sex difference in obesity is observed among non-Hispanic black adults [28]. Among Asian Americans, women likely pay more attention to healthier lifestyle, e.g., engaging in more physical activity, and thus are less likely to be obese. Asian Americans were the fastest growing race/ethnic group in the U.S. in the past decade, and these numbers will continue to rise in the coming decades. Asian Americans are expected to double in population size with a projected increase to more than 43 million by 2050 in the U.S. [12]. California has the largest Asian population in the U.S. [35]. Compared to 2001, obesity prevalence in 2011-2012 has increased among Chinese , Japanese , Filipino , and Vietnamese . However, it has not changed significantly among Korean [2]. The present study found that compared to Chinese, being Japanese or Filipino was associated with a higher prevalence of obesity, which is consistent with previous findings [2]. In addition, another report also showed that among Asian adults, Filipino adults were more than twice as likely to be obese as Asian Indian , Vietnamese , or Chinese adults , and were 70% more likely to be obese as compared to the overall Asian population [36]. It is important to investigate the obesity disparity in Asians, because they have pronounced socioeconomic disparities across ethnic groups, with some ethnic groups' SES far exceeding national averages, while some other groups' SES being lowest in the U.S. [15]. This study had some limitations. First, the data were collected almost 8 years ago; as such, the data may not present the current demographic composition in California as the latter has changed in the last decade. Second, data were collected by self-report, making responses prone to social desirability bias and recall bias. Third, although not necessarily a limitation, our findings may not be generalizable to Asian populations in other states. Fourth, because this study was cross-sectional, the directionality of cause and effect of the association between race/ethnicity and obesity cannot be established. Thus, further studies need to use longitudinal data to explore their causal relationships. However, CHIS has large sample sizes, even for subgroups, and results are applicable broadly to the adult population in the United States, especially for Asian population. To date, few obesity reports in California have provided important information regarding obesity among adults in the state; the present study offers new insights into obesity research among Asian Americans residing in California. --- Conclusion The prevalence of obesity was found to be higher in adult Asian Americans in California by using the Asianspecific standards than previously acknowledged. Further, ethnic/racial disparities in Asian Americans in California were observed in 2013-2014. Compared to Whites, being Chinese, Korean, and Vietnamese were associated with a lower prevalence of obesity. Among Asians, compared to Chinese, being Japanese and being Filipino were associated with higher prevalence of obesity. These findings can help design more effective interventions to reduce racial and ethnic disparities in obesity, especially for Asian Americans. --- --- Abbreviations BMI: Body mass index; CHIS: California Health Interview Survey; CI: Confidence interval; FPL: Federal poverty level; OR: Odds ratio; RDD: Random-digit-dial Authors' contributions KW and SG had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: SG, KW, YL, AA. Acquisition, analysis, or interpretation of data: SG, KW, ZZ, AA. Drafting of the manuscript: SQ. Critical revision of the manuscript for important intellectual content: ZZ, SG, KW, AA, YL. Data collection and statistical analysis: SG, KW. All authors have read and approved the final manuscript. --- --- --- Competing interests The authors declare that they have no competing interests. ---

Background: Obesity has been generally understudied in Asian Americans. It is important to identify subgroups of Asian Americans at high risk of obesity to help develop targeted interventions for those subgroups. This study aimed to examine the disparities in obesity among Asians (i.e., Chinese, Filipino, Japanese, Korean, and Vietnamese) living in California. Methods: A sample of Adult Americans in California (n = 47,970) including Asian American adults (n = 3810) aged 18 years or older were obtained from the 2013-2014 California Health Interview Survey (the U.S. nation's largest state cross-sectional health survey). Body mass index was calculated using self-reported height and weight. Weight status was determined using the WHO Asian BMI cut points in 4 categories: < 18.5 kg/m 2 (underweight), 18.5-22.9 kg/m 2 (normal weight), 23-27.5 kg/m 2 (overweight), and ≥ 27.5 kg/m 2 (obese). Multiple logistic regression analyses were used to estimate odds ratio (OR) and 95% confidence interval (CI) after adjustment for covariates. Results: Overall, the prevalence of Asians was 23.3% for obesity and 40.0% for overweight. The obesity prevalence was higher in Asians who were males, aged 45-64 years old, had higher family income, were current smokers, never got married, had lower education level, had an insufficient level of physical activity, and had more frequent consumption of fast foods. After adjusting for other factors, compared to Whites, being Hispanics and Blacks were associated with higher odds of obesity (OR = 1.47, 95%CI = 1.31-1.65; OR = 2.04, 95%CI = 1.65-2.53, respectively); being Chinese, Korean, and Vietnamese were associated with lower odds of obesity (OR = 0.28, 95%CI = 0.18-0.45; OR = 0.14, 95%CI = 0.04-0.46; OR = 0.28, 95%CI = 0.14-0.58, respectively). Compared to Chinese, being Japanese and Filipino were associated with higher odds of obesity (OR = 2.75, 95%CI = 1.52-4.95; OR = 2.90, 95%CI = 1.87-4.49, respectively). Conclusions: The prevalence of adult obesity was high among Asian Americans in California. Ethnic/racial disparities in obesity among Asian Americans in California were observed in 2013-2014. Compared to Whites, being Chinese, Korean, Vietnamese were associated with lower odds of obesity. Among Asians, compared to Chinese, being Japanese and being Filipino were associated with higher odds of obesity. These findings can help design better interventions to reduce racial and ethnic disparities in obesity, especially for Asian Americans.

Background In modern society, rising standards of living and advances in public health and medical care have prolonged the average lifespan, while lower fertility rates have decreased the proportion of young people in the population [1]. Both of these trends have increased the proportion of middle-aged and older people in the population [2]. This has been particularly true in China. In 2010, 24.3% of China's total population, or 324,327,480 people, were between 45 and 65 years old [3]. Health-related quality of life is an individual's satisfaction or happiness with the dimensions of life insofar as they affect or are affected by "health" [4,5]. To date, a number of questionnaires have been developed to measure HRQOL and the 36-item Short Form Health Survey is the most commonly used [6,7]. The SF-36 has been adapted and applied in more than 40 countries as part of the International Quality of Life Assessment Project [8]. Since Li first introduced and tested the version of the SF-36 for use in China in 2002 [9] several subsequent studies have confirmed its high reliability and validity for use in the Chinese general population and among patients with chronic diseases [10][11][12][13]. Because health during middle age is important both in itself and as a predictor of health at older ages, knowing more about HRQOL among the middle-aged population is especially important in an aging society like China [14]. But very little is known about HRQOL among the middle-aged population of China. The limited available research indicates that: HRQOL for this group is not very good [15][16][17][18]; physical scores are higher than mental health scores in most studies [12,16,17], with the exception of one study in Shanghai [6]; and HRQOL scores differed by sex, age, occupation and education [15][16][17][18][19][20]. However, none of these earlier studies was conducted in rural areas, where hundreds of millions of middle-aged Chinese people live and where socioeconomic conditions influencing HRQOL may differ from those in urban areas. Prior studies in China have emphasized the individual factors influencing HRQOL, while the international literature on HRQOL demonstrates the importance of household factors. A large, longitudinal, nationally representative survey of British adults conducted in the 1990's emphasized the importance of household membership and characteristics among the social factors affecting HRQOL [21]. One U.S. study found that household-level variables accounted for 4.5% and 26.1% of the total variance in self-reported individual physical and mental health status, respectively [22]. Another study of older adults in rural Vietnam found that people who were currently in married partnerships and from wealthier households reported better health [23]. This article analyzes new data collected in 2013 to describe HRQOL among people between 45 and 65 years old in an area of rural Mid-east China, examining both individual and household-level predictors. --- Methods --- Study sample In March 2013, we conducted a cross-sectional survey of eligible adults in 600 rural households in four counties in China: Zengdu in Hubei province, Nanle in Henan province, Huoshan in Anhui province, and Ji'an in Jiangxi province . The sample sites were drawn based on convenience sampling from 12 participating counties in the Sino-Japanese technical cooperation project, "Infectious Diseases Prevention and Household Health Care in Mid-east China." Selection criteria for project sites included distance from the local airport and relative per capita income level compared with the provincial average. As such, each site was typical for Mid-east China, which is less developed than China's eastern seaboard. We used a two-stage sampling method to select the surveyed families in the four sample counties. In the first stage, we selected three villages in each county by judgment sampling. Our goal was to select a representative sample of village environments, especially related to economic level and health service availability, based on the investigators' previous knowledge of the study area. Chinese counties usually have a central city where the county government is located that has the biggest and best hospital in the county and that also serves as a commercial and cultural center. Typically, villages closer to the county center are richer and those farther away are poorer. To obtain a sample including a broad spectrum of village environments, we selected study villages based on their distance from the county center and per capita income in 2011. In each sample county, one village was selected that was near the county center and had a relatively high per capita income, another that was at an intermediate distance from the county center and had a per capita income similar to the county average, and a third that was farther away and relatively poor. In the second stage, we randomly selected 50 households in each of the twelve sample villages based on household lists from the local public security bureau. We collected data from eligible adults in the 600 sampled households using two questionnaires. The first was a household questionnaire completed by an adult who was randomly selected among those at home at the time, and the second was an individual questionnaire administered to each adult household member. Household and individual questionnaires were linked by codes. Trained interviewers visited each household to distribute the questionnaires. If the participating adults were illiterate, they were interviewed verbally; if they could read and understand the written questionnaires, they completed the questionnaires by themselves. Due to no adult being at home during any of the four days of data collection, we missed 23 sample households. This study was approved by the Nanjing College for Population Program Management Ethics Committee. All participants gave informed consent prior to inclusion in the study. No one refused to participate. While this may seem unusual by international standards, it is not surprising for research in rural China. Factors contributing to participation in this study included: The interviewers were all local people. Villagers would not usually refuse an "insider" according to Chinese culture; If the participants were busy at that time, the interviewer would make an appointment to return at another time; Participants were given a small gift for their time. This article presents findings for 428 surveyed adults who were between 45 and 65 years old from 320 households; 87.2% of these study subjects completed the written questionnaires by themselves. An additional 31 subjects in households that we contacted in the age group that would have been eligible for the study were excluded because they were not in their village during the data collection period; 28 of these were working outside of their counties. --- Variables --- Questionnaire development and testing This study included two questionnaires. One was a questionnaire of household factors designed by the investigators. The other was an individual questionnaire including personal factors and health status. We used SF-36 to measure health status. Other questions covered factors potentially related to HRQOL based on published reports and the investigators' knowledge of the life of middle-aged people in rural China. Both questionnaires were pretested among 30 households in rural Nanle County in Henan Province of China. --- Health This study used the SF-36 to measure participants' HRQOL. The SF-36 is a 36-item questionnaire measuring HRQOL and can be administered in 5 to 10 minutes. The response format of the items varies from two to six response categories. One item measures current health status compared with one year earlier, while the remaining 35 items yield an eight-dimensional profile of physical and mental health during the past four weeks. The eight SF-36 dimensions-physical functioning , role limitations due to physical health problems , bodily pain , general health , vitality , social functioning , role limitations due to emotional health problems , and mental health -were individually scored and also combined into two categories representing physical functioning and wellbeing and emotional wellbeing: the Physical Component Summary , including physical functioning, role limitations due to physical health problems, bodily pain, and general health; and the Mental Component Summary , including vitality, social functioning, role limitations due to emotional health problems, and mental health. Scores of the 8 individual health dimensions and of the PCS and MCS were calculated according to the SF36 manual. Because a full Chinese SF36 normalized model is not yet available, we used the results of an earlier Chinese study as the normal model to calculate PCS and MCS [24]. --- Individual characteristics Based on previous studies [15][16][17][18][19][20], this study collected data on respondents' date of birth ; sex; highest level of education ; current marital status ; and current occupation . --- Household factors We constructed questions to assess household factors based on variables that seemed most relevant to the local context based on the investigators' knowledge of these areas. Participants were asked about the number of people in their household, whether they lived with people other than their spouse; their town or village of residence; the size of their home in square meters; the kind of water their household members drink daily ; earned household income during the prior year; and if their household members always, often, seldom, or never cared about each other. We included the question regarding household members caring about each other because this has been a popular topic in Chinese mass media recently. --- Analysis We used Epidata 3.2a to record responses and analyzed them using SPSS 17.0. Interviewers and investigators double-checked all questionnaires in the field, so no questionnaire items were missing for more than 1% of respondents. In the few cases where this did occur, we substituted the mode value or mean value . After describing the distributions of the variables of interest, we estimated the bivariate associations of the PCS and MCS scores with each individual and householdlevel characteristic using one-way ANOVA. Due to the relatively small samples and the left-skewed distribution of PCS and MCS scores , binary logistic regression was then applied to analyze the associations of PCS and MCS scores with their potential individual and household-level predictors in a multivariate model. For this analysis, we dichotomized PCS and MCS as lower than 50 versus greater than or equal to 50 . For the binary logistic regression analysis, the potential related factors as independent variables were entered in 2 blocks: block 1 included only individual-level factors, while block 2 included household-level factors. The results of effects are shown as β, OR and significance . P < 0.05 was considered significant. Many of the households sampled had no members in the age group of this study. Others had more than one eligible member included in the study. Insofar as people in the same household are more likely to have similar HRQOL than people in different households, this may violate the assumption of independence in our multivariate model. On the other hand, excluding multiple eligible subjects in the same household from the study would have resulted in underrepresentation of married persons in our sample. To examine this, we conducted a separate multivariate analysis excluding more than one participant from the same household . The results of this analysis were almost identical to the analysis of all 428 subjects, except for wider confidence intervals due to the smaller sample size. For simplicity, we present only our analysis of all 428 subjects. --- Results --- Individual and household characteristics As seen in Table 1, the mean age of respondents was 54 years, and nearly half were female. Most were literate and 43% had middle school or higher diplomas. More than half were farmers. All but 6% of respondents were currently married, and half lived with people other than their spouses. Their mean household per capita living space was 46.06 m 2 , and 49% used tap water/pure water as daily drinking water. Average household per capita income was 8098 ¥ . Seventy percent of respondents reported that their household members always cared about each other. --- Health status Scores for the 8 individual dimensions of health ranged from 63.46 for general health to 89.18 for physical functioning. Average PCS and MCS scores of participants were 46.97 and 53.11 respectively, with scores concentrated at the higher side of the distributions. In Table 2, we compare our results to those of two other Chinese studies. When calculating PCS and MCS scores, the Hong Kong study was used as the normal model for all three [6,24]. Table 3 presents PCS and MCS scores by respondents' individual and household characteristics. One-way ANOVA analysis showed that: PCS differed by age, education, occupation, household per capita income, drinking water supply, and frequency of household members caring about each other; and MCS scores differed by education, household per capita income, daily drinking water supply, and frequency of caring about each other. --- Binary logistic regression analysis of health-related household factors Tables 4 and5 show the results of binary logistic regression analysis with PCS and MCS as dependent variables. For PCS, in the block 1 model , R squared equaled 0.125 and the model was significant . After adding household factors, the R squared of the block 1 and 2 model increased to 0.207. The household factors thus explained 8.2% of PCS variance. Household per capita income, drinking water and frequency of household members caring about each other were significantly associated with PCS. Those who were from families with higher income, drinking safe water, and always caring about each other had better PCS scores. For MCS, R squared was 0.031 in the block 1 model and increased to 0.138 in the block 1 and 2 model. Household factors thus explained 10.7% of MCS variance. MCS scores were higher among respondents with higher household per capita incomes, safe drinking water, and who reported a higher level of family members caring about each other. --- Discussion and conclusion Compared to the Shanghai study, the average RP, BP, GH, SF, RE, and MH scores of this sample were lower by more than 5 points. Compared to the Hong Kong sample, the average RP, BP, RE, and MH scores of this sample were lower by more than 5 points but their VT and RE scores were higher by more than 5 points. The findings for our sample of middle-aged people in rural Mid-east China indicate that they had better mental health but worse physical health than the Hong Kong sample, for which PCS and MCS were both 50 because this was the reference standard. Both mental health and physical health were worse in our sample than in the Shanghai sample and HK sample [6,25]. Their scores in 8 dimensions were similar to scores of a Swedish middle-aged population [26] but different from scores of Australian middle-aged women whose PF, VT and SF scores were lower and GH score was higher [27]. This study also found that the respondents' MCS scores were higher than their PCS scores, in contrast to findings from most previous studies in China [12,16,17], which focused on urban areas, but similar to those from several earlier studies in Shanghai [6], Japan [28], India [29] and Australia [30] that included some participants from rural areas. The Shanghai study also showed that whether participants were from rural areas was significantly related to RP, BP and MH [6] with people in urban areas reporting higher stress levels than those in rural areas. Three comparison studies in China showed that urban people had lower subjective well-being and higher stress than rural people [31][32][33]. A study among Chinese middle-aged intellectuals showed that stress was one of most important HRQOL-related factors [18]. Our results indicate various factors associated with HRQOL among the rural middle-aged population in Mid-east China. Age, educational level and economic status were strongly related to HRQOL. Increasing age was associated with deteriorating physical and mental health; higher education was associated with better physical and mental health, perhaps reflecting increased knowledge about health matters; and lacking a paid job and low household per capita income were associated with poorer mental health. These findings are consistent with previous studies conducted in Singapore, Pakistan, India, Mexico and elsewhere [6,29,[34][35][36][37][38]. This research did not find significant sex differences in HRQOL. This was different from studies of Chinese middle-aged intellectuals [15][16][17][18] but similar to other studies in Shanghai and Hong Kong [6,25]. In the studies of intellectuals, women had lower PCS and MCS scores than men. This may due to their higher level of stress in their work and daily lives. Under traditional Chinese culture, these women may need to work harder than men to obtain the same rewards at work while also being mainly responsible for housework. Rural middle-aged women, on the other hand, usually only do housework and a limited amount of farm work and may thus be under less stress. Of particular interest in this study was how individual and household-level characteristics related to HRQOL. Binary logistic regression analysis showed similar results to earlier studies from Vietnam and the United States [22,23]. Household-level characteristics explained a substantial amount of the variance in HRQOL among middle-aged people in Mid-east China. Individual-level characteristics explained more PCS variance than MCS . Household-level characteristics explained more MCS variance than PCS variance . The percent of variance explained by householdlevel characteristics was different in these three studies, probably due to differing samples, methodology and cultural factors. But all found household income to be an important household-level predictor. This study also found two other novel household-level predictors of HRQOL: drinking water supply and frequency of family members caring about each other. Middle-aged people whose household used safe water and cared about each other more frequently had higher PCS and MCS scores. Safe drinking water is currently a very popular topic in China. People are worried about water safety. Our research found that only half of rural families used tap water or pure water although the Chinese government had promised that rural tap water supply coverage would reach 75% by 2010 [39]. In our sample, lower household per capita income families had a higher proportion of not using safe water . Being poor and using unsafe water are thus double negative factors for some people. These results support the importance of promoting safe tap water. Pipes not only need to reach every household. The pipes must always contain safe water, and the service must be affordable. Caring about each other more frequently was also a positive factor for health of this middle-aged population. The explanation for this result may indicate: If household members care about each other more, they will be happier and adjust to problems more quickly; Good mental health may be an important positive factor for physical health; If they care about each other more, other household members will have more chance to know about each other's health, to provide support, and to encourage appropriate medical treatment. This study has limitations. The cross-sectional nature of data limited our ability to understand causal mechanisms. Participants with low educational level may not understand SF36 as well, leading to inaccurate health assessment. We do not know the accuracy of responses to items such as income. Also, we used both verbal and written responses in this survey; whether there is a difference between these two data collection methods is unknown. Although SF36 has been validated elsewhere, more work is needed to determine how well it correlates with objective health status in this population. Although we believe our study villages to be fairly typical of rural Mid-east China, they were a judgment sample selected to give a broad range of villages. It was not possible to weight our sample to give results that would necessarily generalize to the three provinces studied, let alone to the rest of rural China. This study also excluded 31 targeted subjects because they were not at home during the data collection. Most of these were working outside the province, suggesting that the HRQOL scores of these "healthy workers" may have been higher had they been included, thus producing a downward bias on average scores. Despite these limitations, these results are important because this is the first study of HRQOL in middle-aged rural Chinese, a group larger than the entire population of all but a few countries. In conclusion, this study provides cross-sectional evidence of the pattern of self-reported health status among a middle-aged population in rural Mid-east China. The findings revealed problems in terms of equity in health with some groups reporting better physical and mental health than others. We also found that household factors, including household per capita income, drinking water supply and frequency of family members caring about each other, were strong predictors of HRQOL. According to these results, health inequality among middle-aged rural Chinese might be reduced by increasing education, income, and accessibility of safe water. The importance of a household environment in which people always care about each other appears to be fundamental. --- Competing interests All authors declare that they have no competing interests. Authors' contributions JZ was principal investigator of this study and contributed to conception and design, acquisition of data, analysis and interpretation of data, and writing the manuscript. XR contributed to conception, design, and acquisition of data. NH contributed to analysis and interpretation of data and writing the manuscript. All authors read and approved the final manuscript. --- Authors' information JZ is an associate professor of Social Medicine and Population Management at the Humanities and Social Sciences Institute in Nanjing University of Posts and Telecommunications. XR is a health project management expert in the Chinese Family Planning Science and Technology Research Institute. NH is a physician/epidemiologist and a professor of Family and Community Medicine and of Epidemiology and Biostatistics at the School of Medicine of the University of California, San Francisco.

Background: China has an enormous and growing middle-aged population. Little is known about health-related quality of life (HRQOL) for this group, especially in rural areas. We examined HRQOL and its individual and household predictors among middle-aged people in rural Mid-east China. Methods: HRQOL questionnaires and information about individual and household characteristics were collected from 428 subjects aged 45 to 65 in 12 villages in Mid-east China. We examined the eight dimensions of the SF-36 instrument, along with the Physical Component Summary (PCS) and Mental Component Summary (MCS) using a reference sample in Hong Kong for standardization. Individual and household predictors of PCS and MCS were examined by one-way ANOVA and binary logistic regression analysis. Results: Self-reported HRQOL was similar to that seen in middle-aged populations elsewhere. Based on univariate analyses, PCS differed by age, education, occupation, household per capita income, drinking water supply, and frequency of household members caring about each other; MCS differed by education, household per capita income, drinking water supply, and frequency of caring about each other. Individual and household-level factors accounted for 12.5% and 8.2% of the variance in PCS, respectively, and for 3.1% and 10.7% of the variance in MCS. Conclusions: HRQOL among middle-aged people in rural China appears similar to that observed elsewhere, and varies by income, education, and other factors. Household factors, particularly the extent to which household members care about each other, are significant predictors of physical and mental health. In addition to improving general socioeconomic conditions, efforts to improve HRQOL for middle-aged people in rural China need to focus on the family environment.

A major shift happened in 2009 when, for the first time, people globally were using their mobile devices more for data transfer than they were for voice communication [1]. Subsequently, the devices that were once primarily used for making phone calls are now being used in ways that far exceed their original design. With the ability to access the internet on the go with our mobile devices, we have officially moved beyond a personal computing paradigm and into a pervasive computing paradigm. This move into pervasive computing culture with the rise of the mobile internet is indeed a global phenomenon. At the end of 2011, there were 5.9 billion mobile phone subscribers, equaling roughly 87% of the world's population who had and used a mobile device [2]. The mobile internet has thus interceded into the global problem of the "digital divide" by providing access to areas that do not have the infrastructure and wealth to support hardwired connections to the internet as used in the desktop model of computing. The costs involved in providing mobile internet access is dramatically less in --- OPEN ACCESS developing countries that previously did not have the infrastructure established upon which the backbone of the internet would be founded [3]. In considering the range of topics that are of immediate concern when studying the rise of mobile internet culture, the issue of the digital divide is one that resides at the forefront. Any study that begins addressing this issue immediately faces an obvious fact: the ideal of mobile devices bridging the digital divide because of increased access is by no means a clear-cut issue. As people have been arguing for some time, from the likes of Andy Carvin [4] to Mark Warschauer [5], simply providing access to these technologies does not solve the issue of the digital divide. The problem instead bleeds across issues such as economics, class, the creation of relevant content, technological literacy, pedagogical training and innovations in education, and the development of meaningful communities that connect with each other in embodied ways through these technologies. The first article for this special issue on "Social Transformations from the Mobile Internet" takes up this dilemma, looking at another important question about the role of mobile technology in the digital divide: does the low-cost nature of the mobile internet actually end up providing lower-quality access to low-income teens in the United States while simultaneously costing them more in the long run? In their article, "Mobile Phones Bridging the Digital Divide for Teens in the US?", Katie Brown, Scott Campbell, and Rich Ling provide an insightful reading of a recent Pew Internet study that they were involved in producing. A pair of statistics served as the foundation for this article: 27% of teens with mobile phones use the devices to access the internet; this statistic dramatically jumps up to 41% when looking at mobile internet use from teens coming from low-income families [6]. In seeking to address this gap, Brown, Campbell, and Ling argue that the upfront costs of purchasing either a desktop or a laptop computer to access the internet is prohibitively expensive for low-income families. Thus, in deciding to opt for the "initially cheaper mobile access, the poor ultimately pay more for the internet, as mobile internet subscriptions are more expensive in the long run. Thus, the equipment divide is not narrowed. In sum, it is expensive to be poor" . Their conclusions-which are grounded in a 2009 study and point to profound repercussions for the future of the mobile internet-address the paradox that "those with the least money are paying the most for the internet." They continue: "Once we understand that it is the teens themselves paying for this service, it becomes clear that, despite being more expensive and less elegant than computer accessed internet, mobile internet is better than no internet…. This speaks to the drive to participate created by the internet; it is striking that teens with no other form of access would prioritize internet access, thereby illustrating the centrality of technical access to our society" [6]. Broadening the implications of this study to the global uses of the mobile internet, it becomes apparent that there tends to be an overemphasis on the technologies rather than on finding a bridge between mobile media and the wider cultural contexts that situate these technologies. For example, the spread of the mobile internet has been lauded as one of the key factors in facilitating the Arab Spring uprisings, the UK riots, and Occupy Wall Street movement. While the mobile internet was indeed one of the key technologies that was used to organize people in countries around the world, Nathan Jurgenson notes in his article, "When Atoms Meet Bits: Social Media, the Mobile Web and Augmented Revolution," that focusing predominantly on the technologies of these movement misses the bigger picture. He writes, "However, calling the Arab Spring a 'Twitter Revolution' or, as Jillian York says, 'Not a Twitter, Not a WikiLeaks: Human Revolution' both fail to account for how technology and society, the digital and the physical, media and humans, have imploded and augmented each other. We cannot focus on one side, be it human or technology, without deeply acknowledging the other" [7]. Thus, imagining the mobile internet as a separate sphere of human activity is founded on the illusion that the sphere of the internet is distinct from physical space; similarly, a focus on the technologies as the catalyst of social transformation creates an false distinction between the online and the offline worlds. In the pervasive computing era of the mobile internet, the online and offline are so mutually constructive that theorizations need not regard these as cleanly delineated spheres. Jurgenson writes, "Our offline lives drive whom we are Facebook-friends with…and what we post about. Our offline histories, social-locatedness in various structures, demographics, epistemological standpoints, etc., all influence how we behave online. And what happens on Facebook influences how we experience life when we are not logged in and staring at some glowing screen. For example, social media users are being trained to experience the world always as a potential photo, tweet, check-in or status update. The logic of social media sites and smart phone technologies fundamentally influence how we experience reality even when offline" . As social media move onto the streets, accessed through the mobile internet, this pervasiveness requires us to theorize the activity carried out with these technologies-form the everyday to the revolutionary-through the lens of what Jurgenson calls "augmented reality": a reference to the technologies of AR which understands the term much broadly beyond just technological interfaces like Layar and the recent demonstrations of SixthSense and Google Glasses. Instead, for Jurgenson, "augmented reality" points toward "a larger conceptual perspective that views our reality as the byproduct of the enmeshing of the on and offline. This is opposed to the view shared by both conceptual positions outlined above that views the digital and physical as separate spheres-what I have called 'digital dualism'" [7]. This augmented reality feature of the mobile internet deeply affects the immediacy with which we access information about the people and places around us. The contextaware, site-specific information that is accessed through mobile media is accomplished in a way that unprecedented in other media. This medium-specific feature, one that is of primary concern to scholars of the mobile internet, is addressed in Kjetil Vaage Øie's article, "Sensing the News: User Experiences when Reading Locative News." The study, which focuses on user interactions with a site-specific mobile news application, LocaNews, during a festival in Voss, Norway, is driven by questions about how user experience changes when news stories become available and shift based on user location. Øie argues, "We construct and develop our reflective self in a different way because of contemporary media and infrastructure. We are now more likely to see our physical locality as part of something bigger than we know of: it does not have to be our place of origin where we were born and raised, but rather the physical surroundings that surround us at any time." He continues: "Even though we are always in place and place is always with us, both place and our perception of it is in a state of constant change. Technology and increased mobility are some of the elements that constitute the always changing context, since they can provide perceptual gains and losses, and bring about different ways of relating and experiencing localness." [8]. The consequences for accessing news in this way gesture toward the emerging trends in journalism as well as demonstrating the key factors of the mobile internet: context-awareness, immediacy, and site-specificity. While these concerns have been important throughout the history of journalism, as Øie notes, the utilization of the mobile internet for accessing contextually-specific news addresses the "augmented reality" that Jurgenson argues for in his article . Here, the medium and the physical reality that contextualizes that medium become so intimately intertwined that they must be considered as co-constitutive of one another. This level of pervasiveness of the mobile internet impacts already existing media like news media and, as Gerard Goggin notes in his article, extends into altering already existing technologies like the automobile. In "Driving the Internet: Mobile Internets, Cars, and the Social," Goggin argues for a plurality of "mobile Internets" that, in their pervasiveness, become "central to social belonging and cultural participation," resulting in "widening expectations of its availability at all times. This is the case with cars, where mobile Internet use is becoming more commonplace, involving email, browsing, Facebook updates, tweets, apps, Google Maps, and so on" [9]. He continues: "We can see mobile Internet's use in cars as forming part of wider systems of technology use. Late modern life is underpinned by intensifying forms of automation, sensing technologies, real-time data gathering and analysis, and surveillance, especially in cities-infrastructures deeply involved in the reconstruction of place and mobilities. These are technologies that remain much more in the background than the heavily advertised options of plugging one's smartphone into cars, yet their implications are highly significant" [9]. The significance, as Goggin notes, is highly understudied and thus requires attention that considers the specific and varied contexts through which the mobile internet is accessed. By combining the Actor-Network theory of Bruno Latour with the mobilities theories of Mimi Sheller and John Urry, Goggin traverses a range of car-internet sites of inquiry. This includes the study of mobile internet use while driving and the safety concerns therein, as well as the implications of the car itself being a networked object, which he notes, leads to the notion that "at the least, the electronic circuitry, the rise in software in cars, the emergence of the automobile as a key site, or node, in the 'Internet of things', and so, can be seen as an intensification of the mediation of automobility; if not, amounting to a strong claim, of a new kind of media-'cars as media'" . Utilizing a range of other examples of the car/mobile-internet intersection , Goggin argues that the various domains of life that are located at this intersection are connected in a meaningful way, "not only through the car's ability to convey the driver to these different places, but through the pervasiveness of the network technologies that bridge them, brought together by the mobile phone" . The final article in this special issue of Future Internet, "The Player as Author: Exploring the Effects of Mobile Gaming and the Location-Aware Interface on Storytelling," brings together many of the topics discussed in the other articles, such as the pervasiveness of the mobile internet, its impacts on everyday life, the ways that mobile media are changing the ways that we understand site-specificity, and the transformations that mobile media bring to previously established technologies. Ben Bunting, Jacob Hughes, and Tim Hetland trace the implications of the design of a "hybrid reality game" that provides players with a narrative interface that utilizes location and site-specificity as a central component of the immersive interactions. Drawing from a discussion about the design of their game, University of Death, the authors explore the ways that mobile technologies are impacting the ways people imaging gaming space and narrative in an era of pervasive computing. University of Death is a game that takes place across the city of Pullman, Washington, USA, and is "half scavenger hunt and half puzzle game: players travel across the city, following clues and GPS coordinates to a series of story-caches" [10]. The story-caches are essentially geocache containers that hold different elements of the narrative of the game, which involves "a film-noir-style murder mystery set in two different time periods: the story's 'present' and three years into its 'past'. The player experiences this story directly by moving through the city of Pullman and seeing the physical world through the eyes of three different virtual characters. Through the use of these different perspectives, we attempted to achieve the melding of storytelling and ludic play that enables worldmaking" . The mobile internet is able to enact this process of play-as-worldmaking scenario because it positions the playeras-author through an interface that significantly connects with the player's environment. Bunting, Hughes, and Hetland explore this deep connection between interface and place by exploring mobile games' implications on narrative's role in gaming, the transformation of the "magic circle" of play in a pervasive game, immersive interface design, and the role of the senses and imagination in these games' process of worldmaking. My goal in bringing these essays together for this special issue on the "Social Transformations from the Mobile Internet" was to gesture toward some of the key concerns that will undoubtedly surface in the coming years. From the role the mobile internet will play in issues of the digital divide to the blending of place and interface in emerging social revolutions; from its impacts on existing media like news to its impacts on existing technologies like the automobile, and, finally as seen in the ways it is transforming our everyday spaces into extraordinary spaces like the city-as-gameboard, the mobile internet will continue to become an integral element to the cultural spaces, objects, and relationships that define us.

The social transformations brought about by the mobile internet are extensive. In discussing the broad range of these transformations-positioned as a shift from personal computing to pervasive computing-this editorial elaborates on the key contributions addressed by the articles in this special issue of Future Internet. These articles touch on topics such as the digital divide, the role of the mobile internet in revolutions like the Arab Spring and the Occupy Wall Street movement, the development of site-specific and context-aware news, the incorporation of the internet into existing technologies like the automobile, and the utilization of the mobile internet to transform everyday spaces into game spaces.

INTRODUCTION The closed system of the Democratic People's Republic of Korea prevents assessment of mental illnesses, such as, prevalences, risk factors, treatments, and prejudice. Furthermore, there may be a huge gap between what North Korea officially discloses and the real situation. Therefore, we considered that studies of North Korean defectors could better reveal the statuses of mental illnesses. Since 1994, the number of North Korean defectors has in-ness and stigma) play a substantial role in shaping individuals' attitudes toward mental health service. 8 As the stigma of mental illness substantially influences the attitudes of South Koreans and reduce the utilization of mental health services, we considered that the same to that of general population and North Korean defectors. 7 Despite of much research on the mental health of North Korean defectors, few studies have addressed the stigmas of mental illnesses in North Korean defectors. This study addresses the stigmas of mental illness such as those concerning psychosis, alcoholism, attempted suicide, and depression among North Korean defectors. In addition, it examines associations between stigmas and sociodemographic characteristics of North Korean defectors. --- METHODS The stigmas of mental illnesses were examined in 639 North Korean defectors aged 19 to 65 residing in a Settlement Support Center for North Korean Refugees from June 2012 to February 2013. All 639 had been in South Korea for less than 3 months. We excluded the illiterate subjects and the self-surrendered subjects who had been lived in more than 1 year in South Korea. Of these 639 subjects, 545 completed the Perceived Devaluation-Discrimination scale , [9][10][11][12] the Korean version of Psychiatric Diagnostic Screening Questionnaire , 13,14 and sociodemographic questionnaires an overall response rate of 85.3%. Two physicians with more than 6 months of experience of treating North Korean defectors performed the interviews. Before interviews, physicians explained only the symptoms of depression, because of subject unfamiliarity with the term. The study was approved by the institutional review board of Inha University College of Medicine and by the Settlement Support Center for North Korean Refugees. All study subjects were informed of the purpose and the methods of the study and provided written informed consent. --- Perceived Devaluation-Discrimination Scale The stigmas of mental illnesses were assessed using a 12item scale adapted from the Perceived Devaluation-Discrimination Scale, [10][11][12] which was translated and validated by Lee. 9 The scale measures the severity to which participants believe that a person with a mental illness will be stigmatized if their disorder were known. 12,15,16 Individual responses to questions were scored using a six-point Likert scale, as follows: 1) Strongly agree, 2) Agree, 3) Somewhat agree, 4) Somewhat disagree, 5) Disagree, and 6) Strongly disagree. Summed individual scores approximated to a normal distribution, where a higher total score indicates greater perceived stigma. The PDD was subdivided into four scales, which assessed perceived stigma regarding four mental illnesses, namely, psychosis, alcoholism, attempted suicide, and depression. Some words in the psychosis questionnaire, such as, "mental patients", "mental disorder", and "mental hospital" were changed to "patients with psychosis", "psychosis", and "a hospital to treat psychosis. " Questions were changed in the same way in the other three PDDs. 17 The North Korean defectors answered all four types of PDD. However, 3,055 South Koreans were randomly allocated into four groups, each group answering only one type among four types of PDD. 17 Cronbach' s alpha coefficient for psychosis, alcoholism, attempted suicide, and depression PDDs were 0.828, 0.782, 0.832, and 0.756, respectively. --- The Korean Version of Psychiatric Diagnostic Screening Questionnaire Psychiatric Diagnostic Screening Questionnaire , 13 a useful self-reporting test with 125 questions. Kwak et al. reported an PDSQ optimal cutoff scores for ten types of psychiatric status, that is, major depressive disorder, posttraumatic stress disorder, obsessive-compulsive disorder, panic disorder, alcohol-related problems, generalized anxiety disorder, hypochondriasis, and social phobia among the general population of South Korea. 14 We defined mental illness based on the presence of one of ten types. --- Sociodemographic and clinical variables We assessed sociodemographic variables, which included age, gender, level of education , and marital status . Date of escape from North Korea, number of days spent in transit countries, experience of compulsory repatriation to North Korea, and number of family members in South Korea were documented. --- The 2011 Epidemiological Survey of Mental Disorders in Korea To estimate the prevalence and correlates of mental disorders in South Korea, a nationwide epidemiological study was conducted from July 2011 to October 2011. 19 The Korean version of the WHO-Composite International Diagnostic Interview was used as an assessment tool, and the Perceived Devaluation-Discrimination scale was also used to measure the stigma associated with mental illnesses. 17 The North Korean defectors answered all four PDDs, but 3,055 South Koreans were randomly allocated to four groups, and each group answered only one of the four PDDs. 17 We directly compared the stigma level between North Korean defectors and the general population of South Korea as the same PDD scale was applied to the both groups. --- Statistical analysis The independent t-test was used for bivariate analysis using stigma levels for psychosis, alcoholism, attempted suicide, and depression among North Korean defectors and the general population of South Korea. In addition, linear regression was used to examine associations between demographic predictors/general characteristics/mental illness and stigma in North Korean defectors. The demographic predictors analyzed were age , gender , level of education , and marital status . Point estimates from linear regression models are presented after adjusting simultaneously for demographic covariates, namely, age, gender, the level of education, and marital status. Missing data was excluded from the analyses. The analysis was performed using SPSS 12.0 , and statistical significance was accepted for p values<0.05. --- RESULTS --- Demographic characteristics of the North Korean defectors The mean overall age of the 545 defectors was 35.5 years , and for male and female defectors mean ages were 32.9 years and 36.3 years , respectively. Female defectors accounted for 78.0% of the defectors, and 45.0% of all defectors were less than 35 years old. Nearly 80% had attended secondary school, 49.5% were married, 23.5% were previously married, and 27.0% have never been married . --- Perceived Devaluation-Discrimination Scale Mean PDD scale scores for psychosis, alcoholism, attempted suicide, and depression were 49. 19 1). --- Relationships between demographic characteristics and perceived stigmas of mental illnesses Table 2 presents associations between perceived stigmas of psychosis, alcoholism, attempted suicide, and depression and sociodemographic characteristics in defectors. Perceived stigmas of psychosis, alcoholism, and depression were significantly higher in defectors who were married or previously been married than in those that had never been married . Demographic characteristics were not found to be associated with perceived stigma of attempted suicide. --- Relationships between general characteristics, mental illnesses, and the stigmas of mental illnesses More than three-quarters of North Korean defectors were diagnosed to have a mental illnesses by K-PDSQ. The presence of a mental illness was not found to be associated with perceived stigmas of mental illnesses. Table 3 presents the effects by number of days spent in transit countries, compulsory repatriation, family in South Korea, and mental illness on the perceived stigmas of mental illnesses in defectors Those that spent more than one year in transit countries, had lower stigmas for psychosis and alcoholism. North Korean defectors that had experienced compulsory repatriation to North Korea or had one or more North Korean family members in South Korea had higher perceived stigma for depression. The presences of psychosis, alcoholism, and depression were not found to be associated with perceived stigmas of mental illnesses. No significant differences in the PDD scale for psychosis, alcoholism, and depression were observed between defectors with or without each mental illness . --- DISCUSSION This report documents for the first time stigmas of mental illnesses and associated factors held by North Korean defectors. In this study, the North Korean defectors had higher perceived stigmas for psychosis and alcoholism than South Korean subjects. In North Korea, a psychiatric clinic is called 49th . Generally, North Koreans think that 49th patients suffer from severe hallucinations, delusions, or inappropriate affect. Furthermore, 49th is diagnosed as schizophrenia or a manic state by general physicians and after diagnosis individuals are isolated from society and sent to a 49th ward in secluded area. Most North Koreans have very negative images of 49th. 20 Alcoholism was rare in North Korea until the late 1990s because alcohol production and distribution were under government control. However, since the late 1990s, individuals have been allowed to produce alcoholic drinks at home, and alcoholism is encountered more often. 20 Some studies have reported that the prevalence of alcoholism is high among North Korean defectors. 21,22 Alcoholism can cause many problems, such as, divorce and family dysfunctionality, 23 and this is likely to negatively affect North Korean stigma regarding alcoholism. It was interesting to find that perceived stigma associated with attempted suicide was similar for defectors and South Koreans. Because in North Korea suicide is regarded as a treasonous act and the social classes of family members are degraded, it might be expected that the country's suicide rate North Korean South Korean *psychosis, alcoholism, and depression of North Korean defectors were diagnosed according to Korean version of Psychiatric Diagnostic Screening Questionnaire, † the data was quoted from Choi's article, 17 Psychosis, alcoholism, and depression of South Korean were diagnosed according to the Korean version of the WHO-Composite International Diagnostic Interview, ‡ existence of attempted suicide were assessed with the question: "Have you attempted suicide?". N: number, SD: standard deviation would be lower and that North Korean have a more negative image of attempted suicide than citizens of other countries. 20 On the other hand, in 2011, the suicide rate in South Korea was very high at 31.7 per 100,000 of the population, 24 which would a suggest lower perceived stigma for suicide. Furthermore, experiences of defection might have diminished stigma of attempted suicide. For example, North Korean defectors consider escape as a suicide-like behavior due to the threat to life involved, and usually prepare themselves to commit suicide in of capture. In this study, North Korean defectors had lower perceived stigma for depression than South Koreans. Depressive or anxiety symptoms are usually diagnosed as neurasthenia, cardiac neurosis, or autonomic dysfunction and treated by internal medicine doctors or neurologists. 20 Because North Koreans lack any concept of depression, it was not surprising they exhibited little stigma for depression. It was assumed that the subjects of North Korean defectors were likely to have the highest perceived stigma for psychosis followed by alcoholism, suicidal attempt, and depression. Their mean PDD score of alcoholism, however, was higher than that of psychosis among the South Koreans. 17 As we mentioned previously, North Koreans have a much more negative image of psychosis than other mental illnesses. 20 Regarding sociodemographic variables, the North Korean defectors who were married or previously married had higher perceived stigmas for psychosis, alcoholism, and depression than those who had never been married. Previous studies have also reported that the never married exhibit lower stigma for mental illness than the married. 25,26 Other studies have reported that respondents with children show a more negative attitude toward mental illness patients because of concerns about child safety. [26][27][28] Thus, the higher perceived stigmas of married or previously married North Korean defectors could be partially explained by these concerns. In the current study, age was not found to be associated with higher perceived stigmas in North Korean defectors. Prior studies have reported that younger subjects better accept mental illness. 17,29 This difference could be a unique characteristic of North Korean defectors, who are used to a closed social system, have a powerful negative image of mental illness, and no concept of depression. North Korean defectors that had spent more than one year in transit to South Korea were found to have lower perceived stigmas for psychosis and alcoholism. Thus their experiences in transit may have reduced these stigmas. On the other hand, because North Korean defectors had relatively lower stigmas for attempted suicide and depression, these stigmas were not changed by time in transit. North Korean defectors with experience of compulsory re-patriation to North Korea or with more than one family member in North Korea had higher stigma for depression. Compulsory repatriation and the existence of a family member in North Korea undoubtedly create stress, although it is not clear how this affects stigma for depression. It is evident that the effects of compulsory repatriation and of having a family member in North Korea need more intensive study. Despite the higher prevalence of mental illness in North Korean defectors, the presence of mental illness was not found to be associated with stigma regarding types of mental illness. In a previous study, it was reported that the presence of mental illness did not affect stigma for mental illness in South Korea. 17 The life adjustments made by North Korean defectors in South Korea are clearly related to mental health. [30][31][32] Kim et al. 6 reported that the majority of North Korean defectors have not used psychiatric services due to lack of information, although such aid was desperately needed. The higher perceived stigmas of North Koreans defectors for mental illnesses and an insufficiency of information seem to explain the rare use of psychiatric services. Among South Koreans, the principal barriers to psychiatric service utilization are associated with misconceptions about mental illnesses. 7 To increase the psychiatric service utilization ratio, psycho-education should be provided, people should be informed about the importance of mental health, and efforts should be made to counter stigmas exhibited by North Korean defectors with respect to mental illness. This study has several limitations that warrant consideration. First, the North Korean defectors answered all four PDD types, but the South Koreans answered only one. Thus, because defectors answered many more questions, decreased concentration and boredom might have interfered with the process. Second, the North Korean defectors were recruited specifically at a Settlement Support Center established to meet their needs, and thus, had not been exposed to South Korean culture. Therefore, the results could not be applied to the North Korean defectors settled in community of South Korea or general population of North Korea. Our results provide a means of estimating stigmas exhibited by the North Korean population, but we caution against making hasty generalizations. Third, the prevalence of psychosis determined in the present study was too high, presumably because North Korean defectors have persecutory ideas and are still traumatized by experiences of escaping and associated with adapting at the Settlement Support Center. Evidently, the PDSQ was not appropriate for diagnosing psychosis during the early stage of settlement. In conclusion, we found that North Korean defectors had higher perceived stigmas for psychosis and alcoholism than South Koreans. In addition, we found that marital status is associated with higher perceived stigmas for psychosis, alcohol-ism, and depression. Further studies are needed to determine the effect of psycho-education on the stigmas of North Korean defectors and to document serial changes in stigmas for mental illnesses associated with the receipt of education at the Settlement Support Center for North Korean defectors.

creased significantly due to political, economic, and social difficulties. It was reported at the end of 2013 that North Korean defectors in South Korea numbered more than 26,000. 1 In response, the government of the Republic of Korea (South Korea) has helped resettle North Korean by providing support for living, housing, education, employment and others. 2 Mental health is one of the most significant factors in the contest of the successful adaption to a new environment. 3,4 In 2005, Medecins Sans Frontieres (South Korea) reported that 37.6% of North Korean defectors experienced psychological problems. 5 And according to another study, more than 40% of North Korean defectors had somatization and/or depression. Yet, 83% of those surveyed answered had not received any forms of psychiatric help. 6 In fact, South Koreans use mental health service less than Nigerians, South Africans, Americans, Latinos, Australians, and Israelis, 7 and it has been reported that in South Korea, culture-influenced personal beliefs (knowledge about mental ill-

Introduction As defined by Healthy People 2030, "social determinants of health are the conditions in the environments where people are born, live, learn, work, play, worship, and age that affect a wide range of health functioning, and quality of life outcomes [1]". We feel these constructs are actually Social Drivers of Health rather than Social Determinants of Health because "determinants" suggests nothing can be done to change our health fate [2]. As such, we refer to the SDOH as social drivers of health interchangeably with more common social determinants of health phrasing. The SDOH include non-medical factors like socioeconomic status, education, neighborhood and physical environment, employment, social support networks, and access to quality health care [3] . According to the World Health Organization , addressing the SDOH appropriately is fundamental for improving health and reducing long-standing inequities in health [4]. As outlined in Figures 1 and2, each framework organizes the key components into categories, establishing broad areas across multiple domains into one framework. While the literature surrounding the influence of the SDOH is robust among chronic conditions such as cancer, diabetes, and cardiovascular disease, the ways in which the SDOH affect health outcomes for people with inherited bleeding disorders remains unclear [5][6][7]. The inherited bleeding disorders community includes more than 3 million Americans [8]. The most frequently occurring bleeding disorders include von Willebrand Disease , hemophilia A, and hemophilia B. Less frequent, yet also serious, inherited bleeding disorders include deficiencies of factors II, V, VII, XI, XIII, fibrinogen, α2-antiplasmin, and platelet disorders such as Gray platelet syndrome and Glanzmann thrombasthenia [8]. These bleeding disorders prevent the blood from clotting normally, thus causing affected individuals to experience prolonged bleeding after an injury, surgery, or physical trauma [8]. While there have been great advances in the development of products to treat some bleeding disorders, this population faces a variety of difficult challenges which create barriers to care and potential disparities for health outcomes. Health inequities for PwIBDs may be due to geographical location, financial barriers, access to quality care and treatment, educational hurdles, and psychosocial obstacles [9][10][11][12]. For many PwIBDs, the distance to their hemophilia treatment centers , specialized, multidisciplinary health-care centers providing team-based care to address physical, psychosocial, and emotional needs, can affect how often they see their health care The inherited bleeding disorders community includes more than 3 million Americans [8]. The most frequently occurring bleeding disorders include von Willebrand Disease , hemophilia A, and hemophilia B. Less frequent, yet also serious, inherited bleeding disorders include deficiencies of factors II, V, VII, XI, XIII, fibrinogen, α2-antiplasmin, and platelet disorders such as Gray platelet syndrome and Glanzmann thrombasthenia [8]. These bleeding disorders prevent the blood from clotting normally, thus causing affected individuals to experience prolonged bleeding after an injury, surgery, or physical trauma [8]. While there have been great advances in the development of products to treat some bleeding disorders, this population faces a variety of difficult challenges which create barriers to care and potential disparities for health outcomes. Health inequities for PwIBDs may be due to geographical location, financial barriers, access to quality care and treatment, educational hurdles, and psychosocial obstacles [9][10][11][12]. For many PwIBDs, the distance to their hemophilia treatment centers , specialized, multidisciplinary health-care centers providing team-based care to address physical, psychosocial, and emotional needs, can affect how often they see their health care The inherited bleeding disorders community includes more than 3 million Americans [8]. The most frequently occurring bleeding disorders include von Willebrand Disease , hemophilia A, and hemophilia B. Less frequent, yet also serious, inherited bleeding disorders include deficiencies of factors II, V, VII, XI, XIII, fibrinogen, α2antiplasmin, and platelet disorders such as Gray platelet syndrome and Glanzmann thrombasthenia [8]. These bleeding disorders prevent the blood from clotting normally, thus causing affected individuals to experience prolonged bleeding after an injury, surgery, or physical trauma [8]. While there have been great advances in the development of products to treat some bleeding disorders, this population faces a variety of difficult challenges which create barriers to care and potential disparities for health outcomes. Health inequities for PwIBDs may be due to geographical location, financial barriers, access to quality care and treatment, educational hurdles, and psychosocial obstacles [9][10][11][12]. For many PwIBDs, the distance to their hemophilia treatment centers , specialized, multidisciplinary health-care centers providing team-based care to address physical, psychosocial, and emotional needs, can affect how often they see their health care provider, if ever [5]. Some PwIBDs experience financial barriers related to cost of clotting factor products, insurance coverage/caps, copay assistance program issues, and out-of-pocket costs [5,10,13,14]. PwIBDs may experience bias in treatment, such as non-white PwIBDs reporting higher levels of chronic pain, thereby affecting their overall quality of care [5,9,12]. There is a lack of awareness among this chronic disease population and their medical providers regarding the signs of a bleeding disorder, leading to delays in diagnosis and excess morbidity, as well as the importance of early therapy [9,14]. If left unaddressed, intervention strategies to reduce these barriers/disparities will continue to be disease-specific, often targeting individual and health systems' factors without addressing the SDOH [11]. The apparent lack of literature investigating the impact of SDOH on health outcomes in PwIBDs led us to conduct this systematic literature review. To examine this, we focused on PwIBDs and disease progression, rather than people at risk of a bleeding disorder, and we examined multiple clinical and non-clinical outcomes including bleeding frequency, chronic pain, cost, mortality, and QOL. The purpose of the review is to examine the impact of a broad range of social drivers, rather than limit the review to a specific subset. Given the current literature and the diverse topics, the KFF adapted framework shown in Figure 2 was used to select variables for analysis and categorize current knowledge on the impact of the SDOH outcomes for PwIBDs. --- Materials and Methods --- Eligibility Criteria The following inclusion criteria were used to determine eligible study characteristics: published in English during the last decade , published in PubMed, available in free full text, study population was diagnosed with an inherited bleeding disorder, the study measured one the following clinical outcomes: bleeding frequency, chronic pain, mortality, QOL, and/or cost of treatment. --- Information Sources and Search Strategy A reproducible strategy was used to identify studies investigating the impact of the SDOH outcomes for PwIBDs following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines for literature reviews [15]. The Medical Subject Headings heading for SDOH was not introduced until 2014 [16], therefore, we searched for articles three years prior to ensure inclusivity. A full description of the search terms and search process is illustrated in the appendix . --- • Community/social context: ways a person lives, works, plays, and learns (i.e., social integration, support systems, and community engagement [3]. --- • Education: access to education and its quality [3]. --- • Food: access to healthy/nutritious food [3]. --- Data Collection Process and Study Risk of Bias Assessment Data collection from the eligible articles is shown in Tables 23456, where each table is specific to a SDOH category. The data was extracted for each article on the study design, study objective, number of participants, study population, and impact of outcome measured . The risk of bias and demonstration of causal relationship due to most of the articles being cross-sectional studies is discussed within the limitations section. High prevalence of joint deterioration and pain, with the ankles and the knees being the most affected joints among all age groups. A significant impact of hemophilia on professional and economic levels was particularly evident. Moreover, significant anxiety and depression symptoms were found on 36.7% and 27.2% of adults, respectively, and a belief of chronicity and symptoms unpredictability was particularly prominent among adults and children/teenagers. QoL was moderately affected among adults, but less affected in teenagers and children. High prevalence of joint deterioration and pain, with the ankles and the knees being the most affected joints among all age groups. A significant impact of hemophilia on professional and economic levels was particularly evident.Moreover, significant anxiety and depression symptoms were found on 36.7% and 27.2% of adults, respectively, and a belief of chronicity and symptoms unpredictability was particularly prominent among adults and children/teenagers. QoL was moderately affected among adults, but less affected in teenagers and children. High prevalence of joint deterioration and pain, with the ankles and the knees being the most affected joints among all age groups. A significant impact of hemophilia on professional and economic levels was particularly evident. Moreover, significant anxiety and depression symptoms were found on 36.7% and 27.2% of adults, respectively, and a belief of chronicity and symptoms unpredictability was particularly prominent among adults and children/teenagers. QoL was moderately affected among adults, but less affected in teenagers and children. The collected utility values reflected real-world data and can potentially serve as health-state weights in future cost-utility analyses, although it is important not to use EQ-5D-3L-, EQ-5D-5L-, and SF-6D-derived utility values interchangeably. The HRQoL data further documented the physical burden linked to hemophilia and its complications. Haemophilia A carriers had significantly lower median scores for the domains of "Pain" and "General health" compared to control subjects. Such findings highlight the need for further investigation of the effect of bleeding on HR-QOL in this population. Heavy menstrual bleeding and health-associated quality of life in women with von Willebrand's disease --- Crosssectional Investigate whether women with VWD experienced heavy menstrual bleeding and an impaired health-associated quality of life. --- women Women with VWD Of the 30 women that were included in the present study, 50% suffered from HMB, although the majority received treatment for HMB. In addition, almost all the included women perceived limitations in the overall life activities due to menstruation. The health-associated quality of life for women with HMB was significantly lower with regards to 'bodily pain' compared with Swedish women of the general population. --- Two cases Women and girls with bleeding disorders Multidisciplinary management is important to preserve quality of life and social participation for women from menarche onwards. Healthcare provider perspectives on inequities in access to care for patients with inherited bleeding disorders --- Crosssectional Understand healthcare provider perspectives regarding access to care and diagnostic delay amongst this patient population. --- respondents Healthcare providers HCPs felt that there were diagnostic delays for patients with mild symptomatology , women presenting with abnormal uterine bleeding as their only or primary symptom , and patients living in rural Canada . Fewer respondents felt that factors such as socioeconomic status or race influenced access to care, particularly as compared to the influence of rural location . High prevalence of joint deterioration and pain, with the ankles and the knees being the most affected joints among all age groups. A significant impact of hemophilia on professional and economic levels was particularly evident. Moreover, significant anxiety and depression symptoms were found on 36.7% and 27.2% of adults, respectively, and a belief of chronicity and symptoms unpredictability was particularly prominent among adults and children/teenagers. QoL was moderately affected among adults, but less affected in teenagers and children. --- Results --- Study Selection The search resulted in 1466 original articles, and the titles reviewed for inheritable bleeding disorder populations resulted in 27 articles for retrieval . Then, reviewing the titles for inherited bleeding disorder populations, SDOH, and language yielded 13 articles that were included in the systematic review . Several of the articles assessed multiple of the SDOH categories as health barriers, therefore there were overlaps of articles within each area. Overall, three were categorized as Economic Stability, one was categorized as Neighborhood and Physical Environment, zero were categorized as Food, two were categorized as Education, twelve were categorized as Community and Social Context, and four were categorized as Health Care Systems. --- Results --- Study Selection The search resulted in 1466 original articles, and the titles reviewed for inheritable bleeding disorder populations resulted in 27 articles for retrieval . Then, reviewing the titles for inherited bleeding disorder populations, SDOH, and language yielded 13 articles that were included in the systematic review . Several of the articles assessed multiple of the SDOH categories as health barriers, therefore there were overlaps of articles within each area. Overall, three were categorized as Economic Stability, one was categorized as Neighborhood and Physical Environment, zero were categorized as Food, two were categorized as Education, twelve were categorized as Community and Social Context, and four were categorized as Health Care Systems. See Table 1, line 14 in the appendix for the search terms that generated the identified records. --- Study Characteristics and Outcomes of Studies Study designs included cross-sectional, case studies, and literature reviews. Sample sizes ranged from 30 to 798. Sample population and setting both varied, as well as the study results . Given that statistical analysis was not conducted in each of the studies, we reported on the main findings of the studies relative to the respective SDOH category and health outcome measured. Relative confidence intervals were included, if applicable. Table 7 incorporates all the SDOH components and respective clinical outcome. Each of the thirteen articles measured one or more of the clinical and/or non-clinical outcomes. See Table 1, line 14 in the appendix for the search terms that generated the identified records. --- Study Characteristics and Outcomes of Studies Study designs included cross-sectional, case studies, and literature reviews. Sample sizes ranged from 30 to 798. Sample population and setting both varied, as well as the study results . Given that statistical analysis was not conducted in each of the studies, we reported on the main findings of the studies relative to the respective SDOH category and health outcome measured. Relative confidence intervals were included, if applicable. Table 7 incorporates all the SDOH components and respective clinical outcome. Each of the thirteen articles measured one or more of the clinical and/or non-clinical outcomes. Hemophilia B is known to cause substantial functional limitations and reduced health-related quality of life . --- Okide et al., 2020 [17] Many community-dwelling adults with hemophilia may choose to work in jobs that are unsuitable for them, so as to obtain or maintain insurance coverage. At the same time, many with insurance coverage face rising costs of co-payments and lifetime restrictions. In addition to uncertainty about their ability to keep working, health insurance concerns can be distressing to people with hemophilia and their families. --- Pinto et al., 2018 [18] The occurrence of bleeding episodes in the previous year was reported by 71 adults, 15 With regards to which factors might affect care received by women with inherited bleeding disorders, the majority of respondents felt that lack of patient awareness around "normal" versus "abnormal bleeding" and lack of HCP awareness were the main barriers to care. For almost all respondents , symptoms of excessive bleeding were the most common reason for referral, followed by a positive family history and abnormalities seen on routine bloodwork When asked how satisfied they thought their patients were with their quality of life, 3% of respondents felt that their patients were very satisfied, 58% felt they were satisfied, 20% felt they were neither satisfied nor dissatisfied, 1.5% felt they were dissatisfied, and none felt that they were very dissatisfied. Seventeen percent were uncertain. With regards to how long they estimated their patients' travel time to clinic appointments to be, approximately half estimated 30 minutes to one hour; 33% indicated one hour to two hours, 11% estimated over two hours, and a minority said it took less than 30 minutes. When then asked whether they thought that access to a multidisciplinary clinic could improve quality of care for women with inherited bleeding disorders, the vast majority indicated yes; one respondent indicated no, and 6 respondents were unsure. Participants reporting a higher frequency of joint pain and history of joint surgery had statistically significantly lower EQ-5D-3L utility values than participants who did not experience joint pain or who had not had joint surgery previously . Participants with a history of long hospital stays due to hemophilia reported significantly lower SF-36 PCS scores than those without such a history. Also, patients with more frequent visits to medical professionals regarding hemophilia reported significantly lower SF-36 PCS scores than those with less frequent visits. In contrast, no significant differences were found for SF-36 MCS scores. --- Gilbert et al., 2015 [21] Haemophilia A carriers reported more severe bleeding symptoms than control subjects. The median Tosetto bleeding score of haemophilia A carriers was significantly higher than for women in the control arm . Also, haemophilia A carriers reported greater menstrual blood loss than controls as indicated by a significantly higher median pictorial blood assessment chart score . Haemophilia A carriers had significantly lower scores in the "Pain" and "General Health" domains of the Rand 36-item Health Survey 1.0 than controls. Haemophilia A carriers had a median "Pain" score of 73.75 compared to a median score of 90 for control subjects . --- Our analysis indicates that haemophilia A carriers tend to have poorer HR-QOL than women who are not haemophilia A carriers, particularly in the areas of pain and general health. Physical activity level has been positively correlated with bleeding risk among patients with severe and moderate hemophilia. In addition, significant differences were found in the prevalence of bleeding events, as those who exercised strenuously were more likely to incur bleeds due to trauma, and 55% of PWH actively engaged in sports reported bleeding episodes associated with PA. On the other hand, there was no significant correlation between PA level and bleeding frequency of target joints or joint function, suggesting that the risk of bleeding is dependent on bleeding history, hemostatic control, and sport participation. Physical inactivity is the fourth leading risk factor for mortality, accounting for 6% fo deaths globally. Continuous PA, rather than the type of exercise, is an important determinant of health-related quality of life, even for people with hemophilia. --- Govorov et al., 2015 [23] In the study population, 66.7% of the women with VWD type 1 reported HMB compared with 36.4% of the women with VWD type 2 and 25.0% of the women with VWD type 3. In the dimension of bodily pain, the group of women with heavy menstrual bleeding had significantly lower scores compared with those of women in the general Swedish population. This implies that the women in the study population with HMB experienced an impaired health-associated quality of life due to pain. The health-associated quality of life according to SF-36 appeared to be lower in the study population compared with Swedish women in the general population. However, the differences in median SF-36 scores were not statistically significant. [25] HMB and post-partum haemorrhages are the most frequent bleeding episodes seen associated with lower quality of life and iron deficiency anaemia. The health-related quality of life in 13 years old girls with a bleeding disorder is lower compared to their healthy peers. In contrast, there is no apparent difference in psychosocial functioning and HRQoL between young adult women with a bleeding disorder compared to peers. --- McLaughlin et al., 2017 [26] Compared with patients with no to mild chronic pain, those with moderate to severe chronic pain had 25.5-point and 10.0-point reductions in median PCS and MCS, respectively. Adolescent and young adult females with a bleeding disorder reported lower physical HRQoL when compared with AYA men in this study, even after adjustment for other sociodemographic and clinical factors. --- Neuner et al., 2016 [27] Multivariate analyses within patients, siblings, and peers revealed no differences in self-reported overall wellbeing and all KINDL-R subdimensions in group 1 . In group 2, differences occurred inmultivariate analyses in self-reported overall wellbeing and all subdimensions . The greyed out cells represent the health outcomes that were not measured in each study. --- Health Care System The articles that assessed Economic Stability analyzed barriers such as employment, financial income, and cost of treatment. The article by Burke et al., reported that economic instability was associated with considerable clinical, humanistic, and economic burden of hemophilia B in the U.S. [14]. The mean annual bleed rate was 1.73 ; approximately 9% of patients experienced a bleed-related hospitalization during the 12-month study period. Nearly all patients reported chronic pain, and the mean EQ-5D-5L utility value was 0.76 [14]. Pinto et al., reported a significant impact of hemophilia on professional and economic levels where among the adult participants who had an occupation, either a full or a part-time job, or a student status, 28 reported missed days from work or school due to hemophilia [19]. Additionally, pain showed a wide duration range, varying from 1 month in three age groups to 612 months in the adult's group [18]. While Okide et al., conducted a literature review, a finding from published literature was that many community-dwelling adults with hemophilia may choose to work in jobs that are unsuitable for them to obtain or maintain insurance coverage [17]. At the same time, many with insurance coverage face rising costs of co-payments and lifetime restrictions [17]. Arya et al., was the only study to assess the SDOH category of Neighborhood and Physical Environment. The authors conducted the first study to assess healthcare providers' perceptions around inequities in care for patients with inherited bleeding disorders living in Canada [5]. A main finding from this study was that rural location was felt to be a significant contributor to both delayed diagnosis and decreased access to care, the effects of which may not necessarily be mitigated by universal healthcare delivery. With regards to how long they estimated their patients' travel time to clinic appointments to be, approximately half estimated 30 min to one hour; 33% indicated one hour to two hours, 11% estimated over two hours, and a minority said it took less than 30 min. [5]. Major themes analyzed for the articles that fell under Education were increased school absences and positive impacts of a prophylactic treatment. In the literature review from Okide et al., the authors recommend that community-based adult education and nursing interventions rooted in psychosocial counseling and support services can be helpful in this regard [17]. As reported by Pinto et al., among adults, work/school absences were reported by 42.4% participants and could last up to 293 days. These findings explicitly illustrate the negative impact of hemophilia on work-or school-related activities, corroborating the data from other surveys [18]. Nonetheless, PwIBDs who are on a prophylactic treatment have reported an effective reduction in the number of school and/or work absences [18]. The SDOH category of Community and Social Context analyzed common themes such as QOL and the benefits of sports participation within this population of PwIBDs. Atiq et al., reported that patients who participated in sports had a higher bleeding score item for muscle hematoma 0.38 ± 0.95 vs 0.24 ± 0.75 , which remained significant after correction for age and sex [19]. Additionally, the authors found a linear association between more hours of physical activity per week and a better general health status [19]. Carroll et al., conducted a series of questionnaires, including the EQ-5D-3L and SF-36 version 2, where participants who reported a higher frequency of joint pain and history of joint surgery had statistically significantly lower EQ-5D-3L utility values than participants who did not experience joint pain or who had not had joint surgery previously [20]. In a cohort study, Gilbert et al., reported that hemophilia A carriers had significantly lower scores in the "Pain" and "General Health" domains of the Rand 36-item Health Survey 1.0 than controls [21]. Hemophilia A carriers had a median "Pain" score of 73.75 compared to a median score of 90 for control subjects . For the "General health" domain, hemophilia A carriers had a median score of 75 compared to a median score of 85 for controls [21]. In one study by Goto et al., physical activity was reported to be positively correlated with bleeding risks among patients with severe and moderate hemophilia, where 55% of people with hemophilia who actively engaged in sports reported bleeding episodes associated with physical activity [22]. This study was the only study to touch on mortality out of the 13 articles, where the authors reported that physical inactivity is the fourth leading risk factor for mortality, accounting for 6% of deaths globally [22]. Articles pertaining to the Health Care System category analyzed themes such as access to care for minorities and for patients with rare bleeding disorders, geography, and gender bias. Arya et al., asked practitioners about access to care for visible minorities, patients of lower socioeconomic status , for patients living in rural Canada. Specifically, when asked whether they believe that patients with low SES experience less access to care, 46% indicated 'yes'; 39% indicated 'no', [5]. Moreover, participants felt that patients with bleeding disorders of an unknown cause receive less access to care as compared to those with a bleeding disorder of known cause a lack of certainty [5]. Fortytwo percent of respondents felt that women with inherited bleeding disorders, including symptomatic carriers, experience less access to care as compared to men [5]. --- Discussion This systematic review is the first to review, analyze, and synthesize literature regarding the impact of the SDOH outcomes in the inherited bleeding disorders population. Using a reproducible strategy, 1466 articles were identified, which were reviewed based on predetermined inclusion criteria. When categorizing by the KFF adapted framework, studies tended to cluster into the Community and Social Context, Economic Stability, and Health Care Systems categories. Few studies investigated the SDOH category of Education, and none of the studies investigated the category of Food. Most of the studies were crosssectional, thus limiting the conclusions that can be made regarding causation. Nonetheless, based on the studies reviewed, an association was identified with bleeding frequency, chronic pain, cost, and QOL. This suggests that the SDOH have an association with inferior health outcomes and may influence the clinical progression of inherited bleeding disorders. The impact on mortality was seldomly measured, but when QOL was explored, the study findings did show an impact. --- Summary of Evidence by SDOH Articles categorized as Economic Stability considered topics related to employment, expenses, and cost of treatment. Burke et al., found a persistent and comprehensive economic burden of hemophilia B on patients receiving FIX prophylaxis, with substantial FIX treatment-driven costs to payers and society [14]. The direct non-medical and indirect costs associated with hemophilia B may comprise a relatively small proportion of the total cost, but nonetheless represent a significant burden to patients, employers, and society in the form of lost income and productivity for both patients and caregivers [14]. Direct nonmedical costs were mainly driven by caregiver expenses, both professional and informal, and indirect costs were comprised largely of hemophilia-related unemployment and early retirement [14]. Furthermore, a survey conducted by Pinto et al., in Portugal showed that people with hemophilia do not experience higher unemployment than the general male population, nonetheless, 50% of the sample population declared an impact of the disease on their professional activity [18]. A literature review conducted by Okide et al., found that many community-dwelling adults with hemophilia may choose to work in jobs that are unsuitable for them to obtain or maintain insurance coverage [17]. Such a finding emphasizes the remaining unmet needs for reducing the economic burden with sustainable population health strategies and treatment options for people living with inherited bleeding disorders. In the Neighborhood and Physical Environment category, the main finding was that rural location was felt to be a significant contributor to both delayed diagnosis and decreased access to care. These results are supported by qualitative interviews with PwIBDs where geographical barriers are a recurrent theme [5]. While there is no existing literature on specific geographic barriers for PwIBDs, the literature supports the concept of travel distance to their specialist being a potential barrier to care for persons with hemophilia [5]. The Hemophilia Experiences, Results, and Opportunities initiative found that a significant number of young adults experienced difficulties visiting their HTCs with travel distance and travel time being the main barriers to accessing care. This has also been speculated to be the case for adults with hemophilia [5]. The study authors recommended telehealth or e-health initiatives to help minimize limitations to care for patients living in rural locations without easy physical access to an HTC. The COVID-19 pandemic gave the opportunity for HTCs to further evaluate and identify new ways to address access and utilize digital innovation through the implementation of telemedicine. Ultimately, the need for social distancing while still delivering integrated, comprehensive care opened the ability to serve individuals who had previously faced challenges accessing an HTC. The relaxation of regulations enacted during the pandemic provided room to utilize telemedicine for HTC services, and broadening the HTC patient cohort whose geographic location was a significant barrier to accessing an HTC in some areas. While this is not a complete solution to pervasive health and social inequities, as not all individuals have access to a device and/or reliable internet, it was a positive step forward. Few studies investigated the impact that a bleeding disorder has on Education. Pinto et al., referenced a study done by Shapiro et al., where it was observed that children with higher bleeding rates missed more days of school and tended to have lower academic achievement as compared to the general population [19]. Among adults, work/school absences could last up to 293 days [18]. Interestingly, the number of participants with a college education in Pinto et al.'s study was similar to that of the Portuguese general population in the 25-64 age group, thus showing that the impact of hemophilia is not reflected on lower educational achievement [18]. While there is a lack of evidence to show the impact of a bleeding disorder on educational status, research has shown that prophylactic treatment can effectively reduce the number of school and work absences [18], which is a great accomplishment for this patient population. The category of Community and Social Context mainly explored QOL outcome measure. Many of the articles focused around the area of physical activity. While sports participation and physical activity are associated with a better QOL and reduced frequency of spontaneous bleeding due to improved strength and fitness partners, many people with bleeding disorders lack physical activity due to their physical condition and fear of hurting themselves [19,[22][23][24]. This indicates the need for strategies focused on the encouragement of regular engagement in physical activity among people living with bleeding disorders, promoting education about its benefits and potential risks, as well as guidelines on how to adequately deal with eventual injuries. Furthermore, the need for a multidisciplinary comprehensive care team to address physical, psychosocial, and emotional needs of people with hemophilia and other inherited bleeding disorders as a priority target in the improvement of health status and QOL in people with bleeding disorders was raised [14,[17][18][19]27]. Given the COVID-19 pandemic, this is a crucial time to lend a voice to the importance of patient access to a multidisciplinary team to deliver whole-person, integrated, accessible, and equitable healthcare by interprofessional teams [28]. We call upon the medical community to lobby payors to continue to support integrated and accessible care for our patient communities. In the Health Care System category, a main finding by Arya et al., highlights that gender and geography are two key SDOH in the care for patients with inherited bleeding disorders. Travel distance and lack of healthcare provider awareness were noted to be potential barriers to care for patients with undiagnosed bleeding disorders and women alike [5,23,25]. Moreover, the study highlighted that factors outside of the SDOH, such as the degree and type of symptomatology, may contribute to diagnostic delay [5]. --- Summary of Evidence by Clinical/Non-Clinical Health Outcome Bleeding frequency and chronic pain were the most measured clinical outcomes, with half of the articles finding statistical significance. Pain experience in hemophilia has been associated with poor psychological functioning, functional disability, and diminished QOL, additionally increasing the burden of the disease itself [17,20,21]. This association is an important finding of the review. In addition, Burke et al. found that despite individuals diagnosed with hemophilia B receiving FIX prophylaxis treatment, they continue to experience breakthrough bleeding and can eventually develop hemophilic arthropathy [14]. Mortality was seldomly measured, only being identified in one of the thirteen articles which was itself a literature review, so thus statistical significance did not apply. The article measuring this outcome was categorized in Social and Community Context. This suggests that more research is needed regarding the impact of the SDOH on mortality. In terms of the two additional measured non-clinical outcomes, QOL was measured in twelve of the thirteen articles, with half of the articles finding statistical significance. Cost was rarely measured in two of the thirteen articles, with one of the two articles finding statistical significance. The studies measuring QOL were mostly located in the Social and Community category, whereas studies measuring cost were located amongst the Economic Stability and Health Care Systems categories. --- Limitations There are limitations to this study that are worth noting. The search was limited to free full text English articles published in PubMed in the last decade . Given that we did not receive any funding for this review, only free full text articles were reviewed. Since studies with significant results are more likely to be published, the studies in this review may reflect publication bias. The SDOH category of food was not explored in the literature, and thus only five of the six categories in the KFF adapted framework were analyzed. Additionally, the health outcomes were not defined and/or measured consistently throughout the different articles, hindering us from cross-comparing the outcomes effectively. Lastly, most articles were observational designs, impeding the ability to comment on causation. Conclusions from this review are therefore qualitative and meant to guide future research, which is why the National Hemophilia Foundation will further explore this topic and make future analysis/recommendations. --- Conclusions This systematic literature review suggests that the SDOH has an association with health outcome and may influence the clinical progression of inherited bleeding disorders. Nonetheless, further research is needed to better characterize the direct impact of the SDOH outcomes in inherited bleeding disorders. Researchers should be careful to clearly define the health determinant of interested and ensure that the measure is deliberately being examined in studies. The impact of the SDOH key areas of neighborhood environments and food must be explored further, as they are likely contributors to bleeding disorderrelated outcomes, excess morbidity, and excess cost of care for people living with bleeding disorders. Additionally, the SDOH literature must be strengthened, allowing for evidencebased interventions which will be useful in developing national policies for inherited bleeding disorders research and clinical care. Naming, defining, and understanding the root causes for the SDOH is the first step towards the mitigation of pervasive health and social inequities within the inherited bleeding disorders community to improve population health, reduce health disparities, and enable PwIBDs to thrive. --- Implications As representatives of the NHF, we would like to call to attention that for the past 40 years, HTCs have developed and perfected the multidisciplinary shared decisionmaking model [29]. We believe that those in our community who receive care at HTCs have greater likelihood of achieving the outcomes we desire for equitable and optimal health-longer lifespan, improved functioning, decrease in emergency room visits, hospitalization, and healthcare costs. The guidelines by which HTCs operate were established by the Medical and Scientific Advisory Council of NHF and the World Federation of Hemophilia's Guidelines for Management of Hemophilia. We are proud to report that HTCs are fulfilling the promise of high-quality primary care, as described in the article "In Implementing High-Quality Primary Care: A Report from the National Academies of Sciences, Engineering, and Medicine", featured in JAMA's May 2021 issue [30]. We continue to put improving health equity, access, and inclusion in our strategic plans for ongoing improvement to reach more individuals and remove barriers to care. Additionally, the COVID-19 pandemic facilitated the ability to serve PwIBDs who had previously faced challenges accessing an HTC through the implementation of telemedicine. While this is not a complete solution to pervasive health and social inequities, as not all individuals have access to a device and/or reliable internet, it was a positive step forward. Further, we propose that the comprehensive care delivered in HTCs can serve as the model for other chronic illnesses and medical conditions, including fostering open communication and collaboration with multidisciplinary teams, including primary care providers. --- Data Availability Statement: Not applicable. --- --- Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/jcm11154443/s1, Figure S1. PRISMA 2020 Expanded Checklist [15]. Author Contributions: K.L., K.N., M.H. and L.A.V. All participated to the planning and writing of the review. All authors have read and agreed to the published version of the manuscript.

The ways in which the social drivers of health, also known as the social determinants of health (SDOH), affect health outcomes for people with inherited bleeding disorders (PwIBDs) is unclear. This systematic review of the published literature examines the impact of SDOH on health outcomes in PwIBDs. Articles that included the following parameters in PubMed informed this study: published in English between 2011-2021; available in free full text; study population diagnosed with an inherited bleeding disorder; and study measured at least one of the clinical/non-clinical outcome measures: bleeding frequency, chronic pain, mortality, quality of life (QOL), and/or cost. The main findings from the 13 included articles emphasized the unmet need for reducing the economic burden with sustainable population health strategies and treatment options for PwIBDs. Rural location was also a significant contributor to both delayed diagnosis and decreased access to care. Furthermore, the need for a multidisciplinary comprehensive care team to address physical, psychosocial, and emotional needs of PwIBDs was raised as a priority target in the desire for equitable and optimal health. This systematic literature review suggests that the SDOH are associated with inferior health outcomes and may influence the clinical progression of inherited bleeding disorders.

Introduction Violence against women is a global scourge, a complex social problem with wide-ranging and expensive harms for communities, economies, and society. Intimate partner violence , often referred to as family violence, domestic violence, or domestic abuse, is predominantly violence against women . It encompasses a range of coercive controlling behaviours, including physical, sexual, psychological and emotional, and economic and financial abuse, to manage and entrap an intimate partner . 'Traditional' forms of IPV, for example, physical abuse, are relatively well-researched if not easily solved, nor wholly understood, with economic and financial abuse long considered an unfortunate consequence of experiencing other forms of abuse. Now, economic and financial abuse is understood to be a standalone form of violence , with prevalence studies suggesting 1 in 5 women face this type of violence . To date, governmental, policy, and scholarship have focused on integrating health, legal and justice, and social systems to address IPV and coercive control. Drawing the financial system and its institutions into these efforts, however, is in its infancy. Social service providers are pushing global momentum with, for example, Women's Information and Referral Exchange in Australia and Surviving Economic Abuse in the U.K. working toward banking solutions. Similarly, in Aotearoa New Zealand , the social sector is aiding victims-survivors facing the realities of this oppressive form of violence and working with motivated banks. Banks hold a unique and powerful position in an intimate relationship, power yet to be fully tapped into by the bank to both avoid harm and do good. Notable exceptions exist, both at an individual banking level and industry level, including the Australian Bankers' Association Industry Guideline for financial abuse, and the U.K. Finance Financial Abuse Code of Practice. The Co-Operative Bank supported the economic abuse-focused prevalence study above and its update . Business ethicists have tended to take a broader view of banker professional responsibility. The focus on banking stability and systemic importance is constrained for the most part to the "too big to fail" setting, especially post-Global Financial Crisis , see for example Moggia , Linsley andSlack , andEY . Corporate social responsibility is also not a new topic in ethics , nor finance . de Jonge places a feminist lens on CSR, exploring the responsibility of workplaces to support their employees experiencing violence. More generally, how employers may best support staff experiencing IPV is surveyed by MacGregor et al. , suggesting there is no 'one size fits all' approach. However, scholarship is sparse on if, when and how a financial institution should respond to societal issues, such as IPV, traditionally outside their 'remit' of financial services. I aim to fill this gap by bringing together two distinct but interrelated strands of the business ethics literature to argue why and illustrate how the financial institution, specifically the retail bank, plays a vital role in society's response to financial abuse as IPV. The first is systemic harm , that is, the financial institution as systemically important to society. Therefore, its actions are felt beyond its direct interest groups such as consumers and shareholders. By focussing on the bank's relationship with a consumer, I expand Armour and Gordon's systemic harms into the realm of retail banking from the post-GFC reform context mentioned above. The second concept used is consumer vulnerability, an idea of increasing prominence for all corporates but arguably even more significant for systemically important institutions. Herzog cites Armour and Gordon for the term "systemic harms" and suggests two paths: 'narrow' and 'broad'. Herzog addresses the so-called narrow path, opting to guide the prevention of additional harm to society. I address 'broad' duties and the narrow, encompassing both positive and negative avoidance of harm. In the former, providing space for a dedicated domestic and family violence response within the bank's core business allows a targeted and hopefully more impactful approach. For the latter, an example may be protecting an existing customer's safety by ensuring their mailing address cannot be accessed by their abuser. I propose the role of the bank is relevant both as remedy and prevention: providing remedy to a victim-survivor of IPV and, specifically, financial abuse within a framework of consumer vulnerability, and can play an active role in shaping a new path forward for all consumers' healthy financial relationships. Like CSR, consumer vulnerability has been explored in a variety of contexts. Industry bodies and regulators have created codes of practice and guidelines for various sectors, banking and financial services included, and scholars have a long tradition of highlighting vulnerabilities and their impacts. In one such study, Graham examines case studies of consumer vulnerability responses from the service provider's perspective . They examine policy frameworks, complaints procedures, and remedial powers given to the consumer-facing staff at the participant organisations; however, they do not explicitly address violence against women or IPV. Like Graham, the perspective presented here is that of the victim-survivor as an existing or prospective bank customer. While it is reasonable to expect a bank may match its outward stance with an internal support framework, the scope is on the consumer/bank relationship rather than the employee/employer context. While this article focuses on the retail banking sector, related questions of consumer inclusion and exclusion are highly relevant to all business sectors. An increasingly complex and fast-paced digital world widens the scope of a corporation's influence beyond its direct network. I invite readers to define "financial institutions" in the broadest sense, with the argument made here transferrable to any consumer-facing organisation, especially those providing economic resources necessary to live a full life in contemporary society. Examples that immediately spring to mind are telecommunication and utility companies, as both industries also face questions of what consumer vulnerability means in an operational setting. To the best of my knowledge, this is the first article to theorise the argument for retail banks taking an active role in combating economic harm in intimate partnerships. Further, framing victim-survivors of financial abuse and IPV as vulnerable provides a framework for putting theory into practice. I present two juxtaposing examples, one positive and one negative, to illustrate the real consequences of bank and financial institution inaction in this space. The stories are drawn from a collection of twenty-three women's stories of violence gathered over two related qualitative studies, both under institutional ethics committee approval . The article is conceptual rather than empirical and is not a complete retelling of their complex stories nor a presentation of the findings from those projects. Rather, the examples underscore the importance of considering victim-survivors as vulnerable and the financial institution Financial Abuse in a Banking Context: Why and How Financial Institutions can Respond 1 3 as systemically important in preventing further harm and actively responding to financial abuse as part of a collective societal response. --- What is Financial Abuse? Financial abuse is defined as behaviour that restricts, controls, exploits, or removes another person's access to money, economic resources, or participation in financial decisions. While the literature tends to use the terms 'economic abuse' and 'financial abuse' interchangeably, recent work by Sharp-Jeffs clearly describes financial abuse as one facet of the more broadly defined economic abuse. The former focuses on monetary and financial resources, whereby the latter takes a wider view of economic resources, including housing, employment, and education. There are instances of overlap, such as credit ratings or mortgages, which may impact housing . Neither economic nor financial abuse require physical proximity to perpetrate , allowing abuse to continue unabated postseparation and severely restricting victims' ability to move on with their lives . Both are mechanisms of partner and systemic entrapment . In their work on "nonviolent coercive control", Crossman et al. find victim-survivors reported higher levels of fear post-separation than those in their "violent" and "nonviolent" comparison groups. Additionally, online platforms and social media also aid financial control and stalking behaviours, giving rise to newer methods of abuse such as so-called transaction abuse . Consequences of financial abuse include trauma-related health issues, poverty, debt, lost income, and unemployment , restricting a victim's ability to end the relationship and seek safety for themselves and their family. Inequity in systems, including legal and justice , social , and financial , exacerbate violence and contribute to wider 'systemic entrapment' as inequity intersects with disadvantage. Examples of disadvantage include gender, ethnicity, immigration, ongoing impacts of colonisation for Indigenous women, socioeconomic status, health, and financial capability-each is compounded by societal norms, traditional gender roles, and the taboo of open money conversations. Thus, any response to financial abuse, and IPV, is increasingly understood to require a collective shift from victim empowerment toward holistic, system-wide approaches involving the whole of society and its communities . At the time of writing, the global community is facing ongoing uncertainty due to the COVID-19 pandemic, with varying degrees of high financial stress, isolation and forced proximity. Financial distress and hardship have increased, leading to reports of increased conflict over money matters within households . Preliminary studies show rates of family and intimate partner violence have increased, as movement restrictions heighten barriers to safety for victim-survivors locked down with their abusers . Krigel and Benjamin provide insights into the transitional path from physical through to economic abuse, suggesting the typical experience of IPV is not only complex but changes form over the course of the relationship. As nations flatten their pandemic curves, ease lockdown restrictions , emergency financial support ends , and the economic impact of the pandemic becomes clear, it is logical to hypothesise other forms of IPV and coercive control may become more prevalent. Combined with general financial uncertainty and instability, a higher number of retail banking consumers are at risk of vulnerability and for victim-survivors of IPV, the risk is acute. --- Banker Responsibility: Consumer Vulnerability & Systemic Harms The Retail Bank as Systemically Important Banks are corporate institutions of systemic importance. That is, it is in the interests of a nation and its citizens that their banking system functions well, as the impacts of failure are adverse consequences for everyone in that economy, including those outside that one bank's direct business . In the context of banking institutions, stakeholders are not simply those with direct relationships, such as shareholders and customers, but rather the bank is, and should be, invested in the society in which they operate . The ethics literature provides much discussion of banks concerning their role and responsibility as systemically important actors in a wider economic system, with post-GFC reforms a case in point . However, scholarship is limited on banking ethics and household or consumer issues outside the vulnerability context examined below, providing little guidance on if, when and how a financial institution should respond directly to social problems, such as IPV. For households, it is banks lending to mortgage holders to provide shelter, finance businesses and thus allow employment, and facilitate the day-to-day money management of individuals and their families. In extraordinary times, such as those of the COVID-19 pandemic, it is banks providing temporary relief for households through adjustments to debt repayment terms in the event an individual-through job loss, income reduction or extended leave/furlough-is struggling to service their debt. It is important to conceptualise the bank or financial institution as systemically powerful not only for a society and its economy, but also as wielding unique power within a household or intimate partnership-power that can inadvertently cause harm or do good. Money can indicate other aspects of an intimate relationship, especially the power dynamics between partners . An individualistic approach to money may give individuals a greater perceived right to control the money they earn in the workforce , conflicting with the concept of an equal partnership . Further, household finance is often an uncomfortable topic for couples and families to discuss, leading to conflict in the most egalitarian of relationships . The so-called 'money taboo' prevents open discussion of personal finances, including household financial matters . Complexity in the various layers of household money matters enables controlling partners to have financial control, often with severe consequences for victim-survivors and their children/dependents . Seemingly benign methods of household money management can exacerbate unequal power dynamics, with institutional practices enabling abusive partners' financial control. Examples of the mechanisms used by couples to manage their money and financial matters include individual and joint bank cheque, savings and loan accounts; joint debt outside banks ; investment ownership ; family trusts; and primary residence ownership. In each instance, external organisations use their own and industry-standard policies and protocols to regulate and govern the use of these tools by an individual and/ or couple. The banking sector especially has a stake in the prevention of and response to financial abuse , given their almost-total reach across global populations and intimate knowledge of a household's financial matters. In this paper, I extend existing understandings of 'systemic harm' to motivate the bank's role in directly addressing economic harm in the context of IPV, moving beyond economic stability and corporate social responsibility work dominating banking, financial and corporate ethics. The Experience of Financial Abuse as Consumer Vulnerability Hill and Sharma define "consumer vulnerability as a state in which consumers are subject to harm because their access to and control over resources are restricted in ways that significantly inhibit their ability to function in the marketplace." This definition is appropriate here for two primary reasons. First, it is general enough to apply to any industry and thus adaptable to banking, financial institutions, and/or services. Second, and arguably more importantly for our purposes, it echoes the definition of financial abuse, provided earlier, almost word for word. This latter reason also highlights an important distinguishing factor for victim-survivors of financial abuse and IPV more widely, from other life experiences and circumstances commonly understood as potential points of vulnerability. Victim-survivors are subject to both the intentional actions of their abuser and those unintentional actions of their financial institution. Both forces exacerbate a victim-survivor's experience of vulnerability and directly impact "their ability to function in the [financial] marketplace". The definition above also provides instructive scope for any organisation seeking to build actionable policies and practices and provide practical guidance for their staff. Namely, despite the introductory section of this article stating that intersectional disadvantages compound financial abuse and IPV, disadvantage alone does not render a consumer vulnerable. "Disadvantaged groups are disadvantaged because they are unequal…in a specified context" ; however, disadvantage alone does not automatically ensure that an individual is vulnerable to harm from an organisation . There is a difference between disadvantage and vulnerability: the circumstances experienced by an individual that adversely impact their ability to control and/or have access to resources and/or a marketplace. That is, circumstances opening someone to harm are what render them vulnerable, not any particular disadvantage they face. Vulnerability is thus predicated on the interaction an individual has with another party or organisation. By definition, all those experiencing circumstances rendering them vulnerable are open to harm from those outside parties that have the power to restrict their autonomy and agency in a market. However, harm arising from an individual's vulnerability is not binary nor guaranteed and exists on a spectrum-a note important to remember when prioritising financial abuse over other areas of potential vulnerability banking customers face. The adverse consequences for institutions behaving poorly with their so-called 'vulnerable' consumers have recently been highlighted, gaining the notice of professional bodies, regulators, and financial institutions alike. For instance, Australia's Royal Commission into Misconduct in the Banking, Superannuation and Financial Services Industry has led to significant legislative and regulatory change across the sector. In the N.Z. context, third-tier lenders have come under scrutiny for predatory lending practices. They are now subject to stricter lending rules under the amended Credit Contracts and Consumer Finance Act . Protecting individuals facing not disadvantage but life circumstances leaving them vulnerable has been highlighted as essential to ensure an equitable and inclusive society. Such ethical discussions are not new, with marketing practices that may be considered exploitative and their ethical implications considered for decades . The stakes of getting it wrong are high, nor is it only a problem for the finance sector, as demonstrated by the recent AUD 50 million fine for telecommunications company Telstra, for predatory sale practices and "unconscionable treatment of Indigenous phone…customers" . Where does this leave victim-survivors of financial abuse? Violence is undoubtedly traumatic to live with and experience. However, research finds that women do not necessarily feel vulnerable until they need to seek help from outside organisations . For victimsurvivors of financial abuse, their vulnerability in relation to their financial institution may not be felt until they are required to interact with their bank or a budgeting or debt service. For our purposes here, it is helpful to draw a line between the abuser's actions creating the circumstances that make the victim-survivor vulnerable to harm and the harm inadvertently caused by their subsequent interaction with the financial institution/bank. Distinguishing between the two forces eroding a victim-survivor's agency and inclusion in the financial marketplace is key to outline the bank's role and response, ensuring further harm is avoided. However, when it comes to operationalising the 'inclusion' of customers experiencing circumstances that render them vulnerable, the implementation of novel codes of practice is more complex. --- The Case for an Informed Banking Response to Financial Abuse --- Aotearoa New Zealand: Background and Context Aotearoa New Zealand provides a unique setting for exploring the role of retail banks in responding to financial abuse in the wider context of IPV. Intimate partner violence rates are among the highest globally and the highest in the developed world. N.Z. data places lifetime prevalence at 1 in 3 N.Z. women experiencing physical and/or sexual violence . When one includes psychological abuse, this statistic increases to 1 in 2 women under N.Z. legislation, economic and financial abuse is categorised as psychological abuse. No economic and financial abuse population-based prevalence data exists for N.Z. at the time of writing. However, recent research found IPVrelated financial abuse doubled in prevalence between 2003 and 2019 from 4.5 to 8.9% . Despite the Anglo-centric focus of the examples provided here, it is worth highlighting that Māori women are over-represented in IPV statistics . In terms of the pandemic, normality has mostly resumed although select industries bear most of the economic burden. As do Indigenous peoples and minorities worldwide, Māori and Pasifika communities face additional structural inequities, including employment and health inequality, impacting their resilience to economic shocks . When it comes to banking, various banks approach domestic and family violence differently. For example, some may have a dedicated in-house team to provide a 'one stop shop' for victim-survivors, while others provide information on their website. However, the N.Z. banking sector has no governing code of practice or guidelines to provide a framework for responding to economic and financial abuse or IPV more generally, unlike Australia or the U.K. . Finally, N.Z. households are more likely to have an interest in a family trust entity relative to those in other Anglo-Western nations , adding a layer of complexity to money management and property settlement when an intimate partnership ends. --- A Note on Methodology To illustrate my conceptual argument, I offer two cases of banking experiences: one negative and one positive . Both stories highlight the tangible impact banks can have on the experience of women facing violence, their financial security both during their 'relationship' and post-separation, and the challenges financial institutions face in providing support for victim-survivors. In writing this article, I have opted to outline where improvements are crucial to avoid furthering economic harm of vulnerable victim-survivors and juxtapose harm with good. The scope of the discussion is retail banking, however, the understandings of vulnerability in the context of financial abuse are relevant beyond the financial services sector. Names have been changed, and some details omitted to protect participant anonymity. Anna and Eloise's stories were each selected from twenty-three interviews collected in two larger qualitative studies, undertaken in 2018 and 2019/2020 respectively, to examine what post-separation financial abuse looks like from women's experience . Both larger projects were designed and developed in alignment with a Constructivist Grounded Theory methodological approach, deemed suitable for the wider research agenda the studies sit within and the critical social justice perspective taken by the researchers. Constructivist grounded theory "provides tools enabling researchers to go deep into studied life and see it from varied vantage points" . The flexibility inherent in the approach allows continuous analysis during data collection, the culmination of which may be new theory and understandings on a topic whilst the research is underway, described by Charmaz as prompting "new ideas, revised directions, and can lead us to retrace our steps". The approach is therefore iterative, as questioning assumptions may lead to further data collection to verify findings and resultant theory. For those interested or unfamiliar with the approach, Väyrynen and Laari-Salmela provide a useful overview of grounded theory in business ethics research. Both the 2018 and 2019/2020 studies focused on women's experiences, and I use gender-specific language here, as IPV is predominantly gender-based male violence against their female romantic partner . However, violence is also experienced in non-heterosexual relationships , and men can be victims of IPV in heterosexual relationships. I acknowledge these experiences as important; however, they were not the subject of the studies described here. The purpose of the current article is not to present comprehensive findings of the larger studies but instead to briefly describe the contrasting banking experiences of two women, both of whom recounted complex stories of IPV. Further details of methods are available in Vogels and Scott . Dr Christina Vogels collaborated with me on the 2018 project, with interviews and fieldwork completed in partnership. Her contribution is gratefully acknowledged. This section will proceed as follows. Qualitative data collection and participant recruitment for the two larger studies are described to provide context for Anna and Eloise's interviews, followed by my self-reflexive positioning . Preliminary analysis follows, providing a foundation for selecting Anna and Eloise's stories. --- Data Collection Women were invited to participate in the 2018 and 2019/2020 studies through two advocacy organisations, which acted as crucial conduits between participant women and the researchers. For the 2019/2020 project, some women contacted the author directly following media interest in the earlier study. For eligibility, we required women to be permanently separated from their former partners. The 'snow-ball' purposive sampling approach ensured recruitment could be carefully managed in terms of the number of potential participants so that no eligible woman was refused a place in the study due to time or funding resource constraints. The inclusive nature of the approach was ethically important to the researchers, given traditional silencing of women's stories of violence. While the recruitment strategy sought to avoid further silencing participant women and their voices, I acknowledge we primarily spoke to those women selected by the advocacy organisations. Twenty-three women were interviewed: fifteen over four months in 2018 and a further eight in late 2019 and early 2020. Qualitative methods are appropriate to capture the complexities of lived experiences of IPV and, specifically, financial abuse. Interviews were semi-structured and lasted between 90 and 120 minutes, with each woman provided the opportunity to recall her experience of violence both during and after the 'relationship' ended. Interviews were audio-recorded and later transcribed by one of three research assistants , then collated with any additional documentation provided by participants. Transcripts were read for accuracy and each interview summarised in preparation for further analysis. Anna and Eloise, the cases presented in this article, were two of the twentythree participant women interviewed. --- Researcher Reflexivity: A Qualitative Requirement Each successive interview challenged my prior understandings of the role banking institutions have in the private lives of individuals and couples. Charmaz asserts constructivist grounded theory's "emphasis on reflexivity… prompts us not only to examine who we are in relation to the research but also to remain reflexive about how we use grounded theory strategies" . I therefore offer my self-reflexive statement here, as prelude to the preliminary analysis section that follows. Trained in finance, both academically and professionally, I am fascinated with how money and financial resources influence behaviour and shape individuals' lives. Societallevel factors, including the money taboo, gender pay and investment gaps, and the ownership we may place as individuals on 'our money', prevent open and constructive money conversations and can contradict the equality we look for in our intimate relationships. When paired with the emotional complexity of intimate partnerships, it seems logical that unequal power dynamics are exacerbated by inequity in financial resources. By extension, in the context of IPV, money and financial resources can be used as weapons of entrapment against an intimate partner . A white New Zealander , raised in Australia, I am a university-educated working mother of young children, who assumed the role of 'stay at home' mother initially and now, am primary breadwinner while my husband is the primary carer of our children. Overlaid by studying and working in a male-dominated profession, each of these roles influenced my collection and interpretation of all stories collected during the study. My analytical and emotional response to the research project has evolved and matured over time, as my professional background has strengthened to include wider understandings of how money is used to exert control over another . For instance, it has taken some years to fit comfortably within the identity of 'feminist' and at time of writing, 'pragmatic feminist' is closer to the mark . Exploring fluid researcher identities is left for a future article and mentioned here solely for the purposes of full and candid self-reflexive positioning. --- Preliminary Analysis: Banking Relevance to Experiences of Financial Abuse The semi-structured interviews resulted in in-depth narratives recounting women's complex lives and experiences of IPV. The focus of the 2018 and 2019/2020 studies described in the 'Data collection' section above was financial abuse. Specifically, the ongoing impact it, the abuser, and institutions have on women's lives as they rebuild them after their 'relationship' with the perpetrator ends. Given our focus on money management and finance, participant women were asked about their banking arrangements during and after their relationship. Of the twenty-three women we spoke with, approximately one quarter redirected the conversation at this point, or stated the bank was not a factor and/or indicated they had not disclosed nor thought to disclose their situation to their bank. Such redirection was not unusual over the course of the twenty-three interviews, as women sought to tell the researchers their story, their way and interwove their narratives of financial harm with other forms of violence. Somewhat jarringly the financial lives participant women lead could not be easily disentangled from the wider environment of violence they face, either during their 'relationship' or afterward . This complexity of context and experience underscores the importance of the conceptual argument made in this article, positioning the bank as important to a consumer rendered vulnerable by their experience of violence. Eighteen women discussed their experience with banks as either directly or indirectly relevant to their experience of violence, often briefly with a relative minority responding in detail/at length. Anna and Eloise's narratives were selected from this smaller group of eighteen. --- Case Selection: Anna and Eloise Participant women we spoke to overwhelmingly related negative banking experiences during our talk. Of these, three alluded to the arguably nefarious intent of individual actors at the bank. While perhaps indicative of poor organisational practices and processes, this article remains focussed on industry and institutional level systems and policies, seeking not to focus on any one bank or financial institution or actor within that organisation. These three stories were thus excluded from case selection in this article. Of the fourteen remaining 'negative' stories, Anna's interview explicitly documented the unintentional harms financial and banking institutions cause via their pre-existing policies, practices, and systems. She highlights many of the elements contained within the interviews of other participant women, bringing together the tentative codes and preliminary understandings built over previous interviews and illustrates the cumulative impact of successive institutional actions. Anna's articulate recount draws direct links between the abuse, bank actions and how the real consequences played out, clearly demonstrating the urgent need for action by organisations to respond to financial abuse and IPV, more generally. Some participant women named a comparatively better experience with one bank over another banking experience. However, of the eighteen who discussed banks as relevant to their experience of abuse, only one woman was emphatically positive when discussing her bank. In a short passage of our interview, Eloise's narrative illustrated the 'tone' any institution can take when seeking to avoid doing further harm to the victim-survivor. More than any other participant, Eloise outlined how the bank helped and the corresponding impact on her experience of violence. Her story lays foundational groundwork toward building supportive understandings of victim-survivors as consumers facing vulnerability. For those women who compared one banking experience to another, usually contrasting positive with negative during their interview, the importance of an empathetic tone was apparent-regardless of whether the victim-survivor had disclosed their circumstances to the bank. Eloise's banking experience, while unique in terms of the other interviews, clearly points toward the direction we might take in building better banking responses-and provides wider guidance for all organisations looking to define their role in combating IPV. To illustrate the complex lived reality of both women, Anna and Eloise's experiences of violence and their subsequent banking interactions, I present each as a vignette: a short summary of their experience. Doing so allows focus on the elements of their narratives most relevant to the topic of this article whilst retaining the wider context of their lives, as relayed to the researchers. While contrasting markedly, when Anna and Eloise's stories are taken together, an understanding of why and how institutions must respond to financial abuse in the context of IPV can be built. --- Anna: An Example of a Poor Banking Response Anna, a mother of four, spoke of two relationships during our talk. The first, her 21-year relationship with her former partner, is the focus of this case. I will use minimal commentary, preferring to use Anna's words where possible, although the potential points of bank intervention are emphasised. Anna had been with her partner for 14 years before they had children. She says the relationship had always had "moments" of violence, but that "she didn't really know because a lot of the times you don't know it's a controlling, abusive relationship". Their finances were completely joint, including bank accounts and two rental properties-everything in both names. They were never married. Anna pointed to a change when she had children, as "when it started to get quite dark". She stopped working and says "there's a sense of entitlement" to him becoming the sole breadwinner. Money became tight, and she did not have enough. As her pre-children career in hospitality was no longer practical, Anna opted to find other sources of income, including study. She said, "Okay, I can study, but that is when he started getting violent…I went to sit my first exam management 101. I think it was five essays, and … he broke my hand." Then, he opened a separate bank account and got wages put into that-Anna does not know when he did this; she guesses when she started having children. Anna had been seeing a counsellor and recognised the abusive nature of her relationship-she says, "You plan, to get out of abuse you have to plan." For Anna, that involved staying in the relationship until her broken ribs had healed, as she knew she would struggle with four young kids at home . For brevity, by 2006, Anna had a protection order, although she distrusted organisations she'd had contact with, including the police, lawyers and other government institutions. "So, I start to feel really scared. Scared of her [the lawyer], scared of the police, scared of him, scared for my kids."-Anna. For example, the lawyer suggested she take the joint credit card to pay the legal fee incurred in obtaining the protection order, NZD3000, and Anna refused: "I can see that being used against me, for clearing out the credit cards like that." On another occasion, the police asked what she had done to make him so angry that he would breach the order. Her abuser went on to breach the protection order "probably every couple of months for seven years." In terms of banking, Anna was on the Domestic Purposes Benefit for those seven years and "never missed a mortgage repayment". Her former partner was not contributing to the mortgage or maintenance on the family home, nor paying child support. At one point, when she estimated they had about 50% equity in the family home, she requested an NZD10,000 loan to replace leaking windows. Her abuser refused to sign the documentation from the bank to extend the loan. Later, she applied for a mortgage holiday as she wanted to work now her twins were at school. The mortgage holiday would enable her to repay her credit card debt of approximately NZD3000, and she could start work without the debt burden. Anna explained her situation to the bank employee; he provided assent. She filled out the paperwork and, 2 weeks later, was declined. Upon querying why Anna was informed she was on welfare. When Anna challenged their reasoning, given she had never missed a repayment, the bank advised she required her former partner's signature. Anna explained she would try but that she did not think he would sign. He did not. In her words, "I had tried on three occasions to separate from him financially". The bank then advised Anna to "stop paying the mortgage, and we'll go for him, that might make him start the proceedings." When I asked Anna how she felt about that advice, she told me she was sceptical at first, but in the end "trusted the bank manager" and stopped paying. When her former partner arrived home to N.Z. from a stint working overseas, his bank accounts were frozen. Anna says the bank didn't tell her they would freeze his accounts, and had she known, she would never have asked for the mortgage holiday nor stopped paying. I asked why. In Anna's words, "He would do something…I mean this guy had the potential to kill, he was that type of person." Out of fear, Anna then went home to pay the mortgage up to date. She had no money in her account to do this . The bank's advice had consequences. Not paying her mortgage allowed Anna's former partner to tell the court she was forcing a mortgagee sale on the family home and was therefore playing games. The judge ordered the protection order lifted and the children to remain with their father. The house was ordered to be sold via an agent. Despite claiming to the court he had no money and therefore needed the house sold, her former partner bid on the house at auction. Anna eventually settled in the Family Court for much less than her lawyer thought she should have, to have it over and done. Anna says, "I needed counselling…none of that was available. I had no money. I needed to understand the dynamics of abuse. I was hugely vulnerable." Unfortunately, Anna then met a man online, and the vignette related here is not the end of her story. --- Eloise: An Example of a Better Banking Response Eloise, a mother of two adopted daughters, was mid legal battle at the time of our talk-her story, as written here, illustrates the severity of harm faced by victim-survivors and its ongoing and long-term nature. She met her former partner, Edward, overseas and married him one year later, returning to N.Z. while he stayed in his home country. When he later joined her in N.Z., Edward was "really upset" she did not own a house. Eloise is unsure why he thought she did, as she had not indicated she owned property. To appease Edward, Eloise bought a house, originally under the impression it would be under both of their names. Edward refused, insisting they have a contracting out agreement , as was custom in his culture. Doing so meant his assets overseas were partitioned from Eloise's in N.Z. . Before the agreement was signed, Edward decided he would return to his home country to visit his mother. When Eloise advised she did not have the money to send him, he "smashed [her] head into a wall". Unable to work or complete her study, Eloise separated from Edward and moved into her property, getting a flatmate to help cover the mortgage as she could no longer work full-time. Following their first separation, the marriage became somewhat onand-off, with Eloise funding Edward's lifestyle, trips to his home country and sending money to his mother and friends. However, Edward became increasingly violent. Eloise explained various situations where Edward would become enraged, including if she walked past him as he prayed or prepared food he did not like, and physically assault her. "I remember when he had his hands around my neck, I stopped struggling. And it was a conscious decision, you know. And immediately fear left me, the panic or the fight for life just left me. That survival instinct just left. … And he stopped squeezing. When I gave up, I think he must have thought that he'd already done the job. … I think in that moment, if I hadn't believed, all that rubbish [I deserve to die. And everything he says is right] and not relaxed, I'd be dead. And still, I kept throwing money at him so he could see his mummy."-Eloise Eloise detailed several specific events in addition to her quote above, including one where she asked him to look at her camera as it was not working, and he smashed it on the floor in front of onlookers. She was so embarrassed she separated from him immediately, although he remained in the house and paid her a nominal amount in rent. Despite her re-applying for divorce, he wanted to reconcile-Edward's visa needed renewal, and he asked Eloise to halt divorce proceedings. She did. Regarding Eloise remembers the time Edward was overseas fondly, despite him using her money to travel, explaining during these periods she felt she was independent, able to use the phone freely and go for walks. However, when he returned, it would be 3 weeks of hell as Edward re-adjusted to N.Z. life. Eventually, following the adoptions of their two daughters, Edward agreed to separate from Eloise. Less than a year later, he approached her for NZD7000 to go overseas. Eloise gave him the money. They had commenced separation proceedings, and Edward placed a caveat on Eloise's house, claiming it was relationship property and that due to his poor English, he had not understood the contracting out agreement he had signed. The caveat meant that she could not sell the property, and despite Edward promising to remove the caveat several times over the 2 years that followed, he has not. Three years after their final separation, Edward almost succeeded in killing Eloise. Police were called; however, Eloise was advised not to go to court as "they're going to crucify you" because "you could say, anyone's done this". Instead, they suggested she seek the help of support services . While Eloise said attending a family violence service and meeting other women facing similar circumstances has helped immensely, she is angered any of them have experienced what they have. Post-separation, Edward refused to pay child support. Despite the Family Court acknowledging the abuse Eloise and their daughters have suffered, at the time of her interview, Eloise was concerned his claim of fifty per cent of the value of the property plus legal fees would be awarded. To fund her own legal expenses, she took a mortgage out on one of her two properties, which ended up going to a mortgagee sale. Over the separation proceedings, Eloise credits her bank as getting her through: "I've never dealt with such a terrible situation with such lovely people." She appreciated the credit controller kept her updated on what she could expect, such as a bailiff arriving, and reassured her throughout the process. However, Edward's lawyer halted the proceeds of the mortgagee sale being released and eventually, her bank settled with her former husband to release the funds. The money from the sale was used to purchase a new property, and her loan was restructured. The bank allowed Eloise to pay interest only on the mortgage. When the interest-only payments come to an end, Eloise will need to pay the mortgage in full , and the house will go to a mortgagee sale. At the time of our conversation, Eloise advised Edward had had her accounts frozen-she believes with the intent to ensure she cannot provide for herself or the girls. She lived off the generosity of friends and was working 80-hour weeks. She had formulated a plan for the care of the girls once she lost her home, knowing her credit rating was too poor to rent a property and was facing homelessness herself. "It doesn't mean that you've got your own name on your bank account, that you have financial freedom. Or you have control over your own life or what you spend. And that has been the hardest lesson."-Eloise Eloise has stayed in sporadic touch with me since her interview. She and Edward are still in the Family Court, and proceedings to financially separate from him remain ongoing. --- In Brief: Unpacking Trauma To unpack what is going on for Anna, Eloise, and other victim-survivors of financial abuse, I will first revisit the concepts of entrapment and vulnerability expounded on above. Second, I relate the experience of financial abuse as consumer vulnerability to our bank and financial institution setting . Finally, referring to Herzog's 'narrow' and 'broad' view of systemic harms, we can operationalise the consumer vulnerability of an IPV victim-survivor into two categories: the narrow, or 'do not make it worse', and the broad, or 'attempt to make it better'. The former is reactive, whereas the latter is a proactive stance. Considering the bank as a systemically powerful institution both at a macro and micro level is helpful as we conceptualise the financial institution's response to individual disclosures of violence by existing customers and any preventative role in society's response to IPV they may play. Economic abuse is the use of financial resources, including employment or education, to control an intimate partner. Coercive and controlling relationships rarely appear obvious to outsiders, and victim-survivors may not immediately realise they are being abused. Victim-survivors, by definition, may not have the level of independence and agency to make decisions benefitting themselves and their families-they may not act in ways outsiders think they can or should. Using a lens of consumer vulnerability, an organisation may recognise a consumer's circumstances render them vulnerable to harm. However, the victim-survivor may not perceive they are vulnerable until the interaction with the organisation leads them to feel this way or know their vulnerability. In Anna's case, she simply wanted to be kept safe from her abuser. The systems in place were inappropriate for doing so. In Eloise's case, her bankers responded to her situation with empathy. They made her feel as though they were doing everything they could, within the parameters of their organisation, to support her. Through these two examples, a clearer picture of both what it means to be vulnerable and how a financial institution can respond emerges. --- A 'Narrow' Banking View of Systemic Harm There have been instances of privacy and system failure in a banking setting, including disclosing a victim-survivor's new address to her abuser . Such examples are often of operational failure and result in internal reviews of procedure to ensure the error is not repeated in the future. Arguably more insidious, although no less severe, is outside advice that causes harm. Two examples are present in Anna's story. The first is a practice that continues today: the bank requiring victims to re-engage with perpetrators to fulfil policy and procedural requirements-for example, requiring both signatures for all changes to a loan or account. For Anna, the bank required her to re-engage with her abuser to make autonomous financial decisions to benefit herself and her family. The second example is when her abuser blocked her attempts to financially separate from him; the advice she received was inappropriate given the legitimate concerns she had for her safety. Unlike in Eloise's case, where each step of the process was outlined to her, for Anna, the implications of accepting a banker's actions were not spelt out. Victim-survivors are the experts when it comes to keeping themselves safe. When weighing the advice of professionals or outsiders, it is crucial for a victim-survivor to know exactly what will happen should they assent to a specific action or procedure, so they can effectively assess the risk the action may pose to their safety. While being financially capable may help, it cannot prevent financial abuse alone. Financial literacy programs aimed at IPV victim-survivors see significant increases in confidence rather than money management skills . Anna showed she was financially capable again and again, even in the face of poor advice from professionals, including her bank. On the advice to stop repaying her mortgage, Anna told me she has since heard she is not unique; it is not uncommon advice. The 'narrow' view of potential harms presented here are reactive from a financial institution's standpoint . The goal in designing policy and procedure is to ensure an existing customer's circumstances are not made worse through the bank's actions. That is, the consumer is not rendered vulnerable by the system with which they are interacting. Referring to Hill and Sharma's definition of consumer vulnerability, an informed set of policies and practices prevents harm for the victim-survivor of financial abuse arising through their interaction with the bank. That is, their ability to function in the financial marketplace is not controlled or restricted by the financial institutions and their systems. Returning to the problem inherent in re-engaging with perpetrators, changing bank policy only to require dual signatures if the outcome is not beneficial to the customer is possible, for example, the opportunity of a lower interest rate on a loan product. In this case, the bank could simply lower the interest rate and inform the customer of their action. For cases where dual signatures are required for regulatory reasons, measures can be put in place to ensure bankers have a full set of information available to help their customer keep themselves safe. However, a narrow view of systemic harms, that is, the avoidance of further harm, is unlikely to be enough in this case. Instead, a proactive or 'broad' approach, and attempt to 'make it better for all' may be required. --- A 'Broad' Banking View of Systemic Harm For customer-facing staff to appropriately serve their customers, awareness of what coercive control may look like in Fig. 1 Systemic harms approach to financial abuse in the banking context a banking setting is required. Research shows positive social reactions to IPV disclosure are important to victim-survivor's mental health and general wellbeing . Therefore, financial professionals need education and training to address direct disclosures of financial abuse or IPV and identify potential red flags. The latter is arguably where this approach becomes proactive, in line with Herzog's concept of broad systemic harm. Rather than simply providing a safe environment for victim-survivors to disclose abuse, dealing with the customer's circumstances and their potential vulnerability sensitively , there is scope for financial institutions to help consumers identify that they may be the victim of financially abusive behaviours and offer support. A broader view of banker responsibility also allows general bank messaging to actively promote healthy financial relationships between couples at key junctures, such as taking out joint debt or opening joint accounts. Banks may already be involved in community outreach, such as financial capability education in schools or sponsorship of community services. Therefore, it is not entirely beyond their remit to take a more prominent advocacy role. Industry leadership in the design of best practice guidelines for regulators to rollout industry-wide also falls under this 'attempt to make it better' response to financial abuse as a social problem requiring collective action, such as the Five Point Plan developed by The Co-Operative Bank in partnership with Refuge in the U.K. . Outside the banking context, immediate pandemic-induced need for government, industry, and business-led crisis responses have also led to innovative ways for IPV victim-survivors to report violence. Examples include the use of code words in essential retailers , providing free internet/wifi in essential stores to improve access to support services and emergency services promoting 'silent solutions' for those seeking their help but unable to safely speak . 'Safe exit' buttons on a website describing financial abuse may further protect victim-survivors from their abuser monitoring their browsing history, a technique any organisation can implement. Finally, a proactive bank or financial institution may form a specialised in-house team for customer-facing staff to refer customers to in the event of disclosure of violence or suspected violence . In Aotearoa N.Z., BNZ's Customer Assist team contains a specialist economic harm response unit . In the U.K., examples include RBS NatWest and Lloyd's Banking Group . This team may have relationships beyond the organisation with family violence support services and offer specialised banking assistance to existing customers and potential customers. Victim-survivors ending a relationship with their abuser may desire to establish a new banking relationship with another institution for a clean start, separate from their former spouse. In these cases, a proactive and 'broad' stance may be to have guidelines regarding the necessary documentation required to open a new account, sensitive to the fact a victim-survivor may not have their identification documents available to them. For existing customers, debt liabilities incurred due to violence or coercion may be partially or fully waived. Ultimately, operationalising any comprehensive consumer vulnerability practice within an organisation requires informed policies and procedures that adhere to regulatory requirements while being open enough to allow bespoke solutions to mitigate vulnerability and systemic harm. --- The Bank is Not the Entire Solution: Signposting and Referrals To appropriately address financial abuse, banking staff require clear tools and certainty over which aspects of a customer's situation they can and should deal with. These are the actions they need to take to prevent consumer vulnerability and harm arising due to the banking transaction at hand. However, not all of a customer's circumstances fall under the banker's responsibility or the care they ought to afford their customers. In the context of IPV and financial abuse, the bank does not have the expertise to support the customer beyond their immediate interactions with the financial system. Therefore, the bank requires clear referral policies allowing staff to refer their customer onto specialised services, such as family violence agencies, to receive holistic support. Further, a bank may provide a grant or financial resources to ensure customers can access support . --- Ongoing Challenges Intimate Partner Violence is an undeniably complex social problem, requiring a coordinated and collective response from various fields, including legal and justice, health, government agencies and financial institutionsindeed, any consumer-facing organisation dealing in economic resources . Economic and financial abuse is experienced by almost all victim-survivors of other forms of coercive control and violence, as one of a range of control tactics used by perpetrators of IPV . However, coercive control can be experienced without the physical violence society still tends to associate with domestic abuse. While our understanding of what constitutes family violence or domestic abuse is widening, economic and financial abuse sits within a grey area of poor awareness and is compounded by the money taboo and societal, gender and cultural norms. Here, I touch on a range of ongoing challenges for banks and financial institutions, including their joint customer relationship to both victim-survivor and perpetrator, and myths and misconceptions of staff. --- Joint Customer Responsibilities Perhaps the biggest hurdle for financial institutions in unintentionally causing further harm to a victim-survivor of financial abuse is that the perpetrator is likely also to be a customer. This joint relationship means several conflicts may arise for banks and financial institutions around regulatory constraints, including privacy and transparency. Transparency in all relationships is key, as demonstrated in Anna and Eloise's cases above, where their respective bankers offered varying levels of transparency around their actions, systems, and processes. For victim-survivors of abuse, transparency of their financial obligations and money arrangements within their intimate 'relationship' often does not exist. However, they are held liable for financial decisions made without their consent or under coercion. For bankers, awareness of the dynamics at play when presented with a case of financial abuse is critical to forming a response that satisfies best practices and reduces the potential for further harm. This hurdle is not insurmountable; however, it requires careful thought and informed policy to ensure procedures are clear for customer-facing staff. When banks are acting on behalf of a victim-survivor, privacy and information-sharing limitations restrict the protective actions the bank can take to limit perpetrator abuse, such as transaction abuse . When the perpetrator of economic harm is their customer, the bank cannot contact the victim-survivor's financial institution and disclose on their behalf, having to limit their action to a warning or, in severe cases, debanking the individual. Addressing these difficulties transcends the institution themselves, and their industry, with any shift in policy and procedure requiring regulatory change. --- Myths and Misconceptions Bankers are not immune to having the same implicit and explicit biases present in the general population, and misconceptions of what a victim-survivor of IPV looks like or the forms violence can take remain an issue globally . In general, banking and financial institutions remain male-dominated and gendered in their service provision, despite moves toward gender equity . For instance, victim-survivors may not fit the stereotypical 'victim' mould. A victim-survivor may not know they are subject to abusive behaviours and financial control. Financial abuse does not discriminate across demography, including gender and sexual orientation. For men, being the victim of IPV and financial abuse may be even more fraught . A person experiencing abuse may be from a wealthy household; they may be facing poverty. Household income or individual earning power is not always indicative of access to financial resources. A customer who is educated, put together, strong and smiling may be hiding a reality far from what they are letting the outside world see. Anna asserts she didn't present as a victim. In her words, "There's nothing victim about it. I'm a mother of four children. I had had enough of being beaten…I am just taking back my power. I had a protection order now; I didn't need to be afraid of him…". The importance of awareness, education and training cannot be over-emphasised, as individual bankers and those in the finance industry are likely to have the same misconceptions of IPV as broader society. Specific training on what financial abuse looks like, how it factors into a wider domestic environment of coercive control, and IPV more generally, can help staff identify and respond to explicit disclosures of abuse, and trigger statements that are potential red flags requiring further information. For example, a customer's throwaway comment about not having access to an account when they should, seeming surprised when presented with an account or debt, or not knowing one's own private details like a PIN. It is these red flags that customer-facing staff are likely to hear and see, and having clear guidance as to what to do next allows them to respond empathetically and work in the best interests of their customer. Despite not necessarily having an explicit policy or guidelines governing their approach, bankers can and do support their customers outside the 'usual' provision of the banking service. Similarly, other financial services, including debt advice services, can help victim-survivors rebuild their lives. Empathy goes a long way, as is evident in Eloise's story above. However, without trauma-informed guidelines to fall back on, bankers may be asked to make decisions beyond their experience and training. In this case, the consumer experience is likely to be ad hoc, inconsistent and detrimental to a wider agenda of financial inclusion and quality provision of services to those consumers who may be experiencing circumstances rendering them vulnerable to harm. --- Concluding Remarks The very question of whether financial institutions have any place beyond the direct services they offer-their core businesses of financial transactions and lending-is an inherently ethical one. Financial institutions increasingly recognise they have an active role in addressing the economic and financial harm inflicted on victim-survivors of IPV, with examples in N.Z., the U.K. and Australia. By the time you read this, further momentum will have built as international information sharing between scholars, NGOs, and industry expands via formal and informal networks. However, the place for financial institutions in responding to economic harm in the IPV context remains under-explored, both in theory and practice. This article has sought to provide a preliminary step toward filling this gap. Here, I present an ethical argument for banks and financial institutions, and consumer-facing organisations more generally, in taking a greater role in an area not traditionally thought relevant to their business: economic harm in the context of IPV. I build on work of business ethicists before me, namely consumer vulnerability at the micro-level while accepting a macro-level risk of systemic harm, to consider what an operational framework of new and improved policy might look like. Victim-survivors are rendered vulnerable by the circumstances they face-entrapped by their abuser. However, it is not until they have their financial autonomy and agency restricted or controlled in the financial marketplace by the banking systems they must interact with that the financial institution's relevance in their vulnerability comes to the forefront. By framing victim-survivors as customers experiencing vulnerability, banks and financial institutions can begin revising their policies and practices to prevent the unintentional harm existing policies may cause. Such work is already underway for other consumer vulnerabilities but requires specialised knowledge, training, and understanding. The strength shown by victims-survivors, illustrated by Anna and Eloise's stories, emphasises that it is possible to provide support within the context of their complex stories without presuming to be the entire solution. It is appropriate to provide support to a victim-survivor in a way that is practical, with immediate impact right now, within the confines of existing policy and regulation. Simultaneously, the institution may lead internal and external conversations on best practices moving forward to enact lasting change to those same problematic policies and procedures. By considering more deeply what works in addition to what causes harm, future research may expand on the suggestions and rudimentary roadmap outlined in this article to further illuminate the issue for those working in the banking industry. This article is the beginning of what I hope will be a rigorous scholarship examining financial abuse and IPV more widely, beyond the obvious systems and institutions traditionally associated as having some role to play in addressing this evasive, invasive, and significant global social problem. ---

Intimate Partner Violence (IPV) is a global social problem that includes using coercive control strategies, including financial abuse, to manage and entrap an intimate partner. Financial abuse restricts or removes another person's access to financial resources and their participation in financial decisions, forcing their financial dependence, or alternatively exploits their money and economic resources for the abuser's gain. Banks have some stake in the prevention of and response to IPV, given their unique role in household finances and growing recognition an equitable society is one inclusive of consumers with vulnerabilities. Institutional practices may unwittingly enable abusive partners' financial control as seemingly benign regulatory policy and tools of household money management exacerbate unequal power dynamics. To date, business ethicists have tended to take a broader view of banker professional responsibility, especially post-Global Financial Crisis. Little scholarship examines if, when and how a bank should respond to societal issues, such as IPV, traditionally outside their 'remit' of banking services. I extend existing understandings of 'systemic harm' to conceptualise the bank's role in addressing economic harm in the context of IPV, viewing IPV and financial abuse through a consumer vulnerability lens to translate theory into practice. Two in-depth stories of financial abuse further illustrate the active role banks can and should take in combating financial abuse.

INTRODUCTION The family planning policy is the basic state policy of China, and its strict implementation has driven a dramatic shift in China's population growth pattern. Between the 1970s and the midto late-1990s, China's total fertility rate dropped sharply, from 5.81 children per family to approximately 1.80. By 2015, the National Bureau of Statistics released data on the 1% population sample survey, which indicated that China had entered a stage of severe under-population. Once the TFR falls below 1.5, the society falls into a "low-fertility trap, " and it is difficult to return to normal levels of generational replacement over a longer period of time . When China entered an aging society in 2000, there were about 130 million people over the age of 60, accounting for 10.2% of the total population; by the end of 2019, the number of people over the age of 60 had risen to nearly 254 million, representing 18.1% of the total population. The alarming trend of couples having fewer children, combined with an aging population, has led to the disappearance of the demographic dividend, posing a significant challenge to China's economic and social development. To address the challenges of continued low fertility and an aging population, in January 2016 the Chinese government endorsed the birth of two children per couple. There have been mixed views among scholars on this policy. Proponents argue that the policy will lead to a great increase in TFR and slow down the process of population aging . However, other researchers take the opposite view, citing the high cost of raising children and sense of gender equality . Today it has become a social consensus that the two-child policy does not work well in China because the TFR did not increase as expected. With the change in fertility policy, the fertility incentives faced by families are bound to change. Therefore, it is important to clarify the factors influencing couples' willingness to have a second child so that we can promote an effective intervention policy that will optimize the demographic structure. At the same time, the study on the fertility intentions of second child can help to design the future progressive liberalization of population policy . Considering the above realities of China's socioeconomic situation, this paper seeks to examine the following core questions: What factors have important impacts on couples' willingness to have a second child in China? What are the differences between men and women in terms of their willingness to have a second child? Specifically, this paper uses data from the Chinese General Social Survey to analyze the relationships between the fertility intentions of different cohorts and the variables impacting those intentions. The paper uses three machine learning methods-artificial neural network , random forest , and XG-boost-to systematically identify the important factors; next, a logistic regression model is used to investigate the mechanisms of the factors that influence the willingness to have a second child. --- LITERATURE REVIEW Scholars have conducted many studies investigating the factors that influence fertility outcomes, and that body of research can generally be categorized into micro, meso, and macro perspectives. At the micro level, these can be further divided into psychological decision-making dimensions, such as the desire to have children ; partnership dimensions, such as marital relationships and the division of household labor ; individual socioeconomic dimensions, such as income and human capital, including education and employment ; and the influence of underlying biological and genetic traits . The meso level can be divided into the social-interaction dimension, including personal networks and social learning ; the place-of-residence dimension, which focuses on the heterogeneity of regions ; and the social capital dimension, which includes goods as well as information, money, work capacity, influence, power, and positive help . At the macro level, the dichotomy between "economy and culture" dominates and considers economic and employment trends along with values and culture . Family policies and welfare systems can strongly influence fertility outcomes. As the birthplace of the modern welfare system, European countries have implemented complementary family policies such as paid leave entitlement, childcare services, and financial transfer to avoid a continuous decline in population by extending state involvement to the family-reproduction sphere . This state involvement in the family domain has prevented a steady decline in the birth rate in recent years. Although many studies have attempted to capture the causality between patterns of demographic transitions and types of family policies and welfare regimes , encouraging fertility is a very complex, systemic ambition, and there are significant differences in the values, programs, and support of fertility policies depending on the welfare regimes, resulting in very different fertility outcomes across countries. In terms of the statistical methods used to study fertility, there are currently two focuses of research. The first is the analysis of causality and endogeneity, such as the link between the field of education and age at first birth at the micro level , the problem of correctly identifying socialinteraction effects at the meso level ), and the difficulty of distinguishing the effects of policies from other factors-observable or unobservable-at the macro level. The second focus is on the prediction of fertility outcomes by machine learning, which has been very poorly covered in the literature. Studies from the United States, India, Indonesia, and other regions have used different machine learning algorithms like regression, decision trees, k-nearest neighbors, and ANN, as well as other ensemble methods like bagging and boosting, to determine the population. This relied on the availability of historical demographic data like population, fertility, mortality, and life expectancy. However, the current studies have not focused on estimating population and TFR in a data-constrained environment and with a lack of international comparison. Since the population policy is unique to China and has been implemented since 2016, scholars have conducted extensive research on second-child fertility. Chen et al. conduct a comprehensive analysis using stepwise regression and found that household economic and health risks significantly influence the intention to have a second child. Zhou and Guo use multilevel regression to find that men, the younger and wealthier, ethnic minority, and rural populations are more likely to have a second child. Scholars have also attempted to focus their research on one perspective. Economic factors are the focus of current research, with Lan demonstrating that women with better socioeconomic status and those who were born into better-off families show a relatively strong desire to have children, and Shen and Jiang investigating highly educated women and finding that their fertility choices are the result of the intersection of state policy interventions and career choices. Urban-rural differences have captured the attention of many scholars as well. Using the Blinder-Oaxaca decomposition technique, Zhou and Guo find that both education and son preference play a prominent role in explaining the willingness to have a second child in rural areas, and Li et al. note that women face the dilemma of having children or seeking employment stability, and that this effect is stronger for urban than rural women. In China today, small families are preferred, and women also have more of a voice in family issues and decision-making , so scholars have also studied gender equality and fertility intentions. Bao et al. and Li and Jiang find that the more equal gender role attitudes are, the more women with more economic and family/social resources are more likely to have a second child, while women who work in the non-agricultural sector and have higher decision-making power in the household are less likely to have a second child. The influence of siblings has also intrigued scholars, and the number of siblings has been shown a significant predictor of women's fertility intentions . The loneliness experienced by only children during childhood and adulthood leads most of them to believe that having siblings is better than being single . However, Zhang L. et al. argue that sibship size may also have negative direct effect presumably due to sibling competition for intergenerational support. Scholars have sought to further analyze it also through qualitative analysis methods such as interviews. Using qualitative data from 53 urban parents in China, Peng suggests that the fertility decision to have a second child is an ongoing bargaining process rooted in the life course rather than an isolated family event. In addition, attitudes and behaviors toward fertility in China are rooted in Confucian philosophies and traditions of ancestor worship and may also be influenced by religious notions of fertility . Therefore, the status of fertility studies in China needs to be improved in three aspects. First, the data need to be updated. Due to the delay in data release, the existing research mainly comprised survey data through 2015. However, the data from 2017 to 2018 represent important improvements in the content of the questionnaire, with more informative content, a more reasonable design, and higher data quality. Second, the methodology needs to be improved. Most of the existing studies used a single model for empirical evidence, such as probit or logistic regression. If the relationship between the independent and dependent variables does not conform to the regression form, it will not be extracted by the model, and the important variables can be easily missed. This results in one-sided empirical results, which cannot achieve the desired comprehensive, systematic, and integrated test. Third, the generalizability of research findings needs to be improved. The existing studies focused on the local population and cannot grasp the relationship between the influencing factors and fertility from the overall population. In summary, inspired by the micro-meso-macro classification and applying it to the specific national situation of China and the characteristics of CGSS data , this study focuses on three characteristics and proposes four hypotheses. Hypothesis 1: Individual characteristics will affect fertility intentions. Hypothesis 2: Family characteristics will affect fertility intentions. Hypothesis 3: Social Characteristics will affect fertility intentions. Hypothesis 4: The above three groups of characteristics will have different effects on the fertility intentions of males and females. The subsequent portion of this paper is structured as follows: The third section introduces the machine learning models that will be used in this paper and briefly describes the data sources and classifications; the fourth section introduces the data and methodology; the fifth section describes an empirical analysis in which the importance of different factors is analyzed by machine learning and quantified and explained using logistic regression. Finally, a discussion and summary are presented for the benefit of future research. --- MATERIALS AND METHODS --- Machine Learning Methods The connection of the variable "willingness toward fertility" to different independent variables is always complex and shifting, and the relationship between variables and willingness toward fertility is non-linear. Because machine learning methods can fit non-linear information as well as linear data, this paper adopts three common machine learning methods to identify the influencing factors: ANN, RF, and XG-boost, and it adopts a logistic regression model to study the mechanism of those influencing factors of willingness toward fertility. Because this paper adopts machine learning, data mining, and data-driven research methods and ideas, and starts from data and objective reality rather than a priori assumptions, the relevant variables are selected to reflect the respondent's basic personal information, work information, and health information. This method makes the profile more detailed, the behavior measurement more accurate, and the empirical conclusion more comprehensive, systematic, and effective. At the same time, these three machine learning methods can support each other to obtain more scientific conclusion. First, from the perspective of optimization methods, the ANN is a local search-optimization method, which may fall into local extremes instead of global optima, thus affecting the overall training effect. The RF technique is a random method to build a forest, which adopts the principle of "minority follows majority" to perform integrated discrimination and ensure the overall optimum. The XG-boost is a method to grow a forest by continuously splitting feature variables to ensure the overall optimum. XG-boost grows the tree by splitting the feature variables continuously and relearns every tree generated to continuously improve the learning quality and dynamically approach the overall optimum. Second, from the viewpoint of applicability, RF has the tendency of overfitting when the data is noisy; XG-boost can effectively prevent the overfitting problem by introducing penalty terms. The ANN has strong fault tolerance and can work normally even after local damage. Finally, in terms of computational efficiency, XG-boost pre-sorts the nodes features before iteration and iterates through them to select the optimal partition point, which is a greedy algorithm that takes a longer amount of time when the data volume is large. --- Artificial Neural Network Model The ANN is an abstract computational paradigm modeled after the human brain that consists of interconnected neurons, i.e., processing units, which simulate the human brain's thinking for computational modeling. Figure 1 shows the common two-layer and three-layer ANN. The B-P ANN model used in this paper is a kind of feed-forward ANN, which has many advantages. First, it has a strong non-linear mapping capability. As many of the independent and dependent variables in the underlying data are non-linear, and the strength and form of the relationships are unknown, the ANN can solve this problem well. Second, the ANN has outstanding self-learning and self-adaptive capabilities. During the training process, the model can automatically extract the data rules between the input data and the output data and write down the learned "rules" through the weights. This allows the ANN to spontaneously learn the connections between the data and objectively reflect these connections. Third, it is highly fault-tolerant, ensuring that the global training results do not deviate significantly when local or partial neurons are damaged and that the system can still function properly even after local damage. --- Random Forest Model The RF model is an RF technique that contains several decision trees with high prediction accuracy that are weakly correlated or even uncorrelated. Based on the classification results of these decision trees, the principle of "majority rule" is applied to make a comprehensive, integrated discrimination. Random forest is a combined prediction model that contains multiple decision trees. As a common method of machine learning, RF has wide applicability due to its simple and easy-to-understand modeling principle, and it offers several advantages. First, RF can handle many input variables, i.e., it can handle high-dimensional data. Second, the RF uses error estimation for errors, which makes the model generalize well. For unbalanced data, the RF can balance the errors and maintain the accuracy of the model even when the features are missing. Finally, since the RF is composed of decision trees, and these decision trees are independent of each other, each tree can be processed in parallel, which makes the model training fast. Through the idea of integration, the integrated discriminative classification is performed based on the classification results of all trees. --- XG-Boost Classification The basic purpose of the XG-boost algorithm is to keep generating the tree and to grow the tree by continuously splitting the feature variables. Each time a tree is generated, a new function is relearned to fit the residuals of the last prediction, which improves the learning quality and approximates the actual value. Compared with the traditional gradient boosted decision tree algorithm, XG-boost has significant advantages. First, the traditional GBDT algorithm model only uses the information of first-order derivatives in the optimization process, while XGboost performs the second-order Taylor expansion of the penalty function and uses the information of first-order derivatives and second-order derivatives for optimization, which makes XGboost's optimization faster. Second, XG-boost can automatically learn the splitting direction when there are missing values in the sample, and after each iteration, the model assigns the learning rate to the leaf nodes, reduces the weight of each tree, reduces the impact of each tree, and provides better learning space for subsequent training. Finally, XG-boost also supports parallelism but unlike RF, this parallelism is at feature granularity rather than tree granularity. In sum, the XG-boost model can quickly and accurately complete the classification and prediction of data, and the addition of penalty terms can effectively prevent overfitting based on high accuracy. --- Evaluation Indicators of Results All the machine learning model results are analyzed based on the confusion matrix, as shown in Table 1 the size of the area under the ROC curve. This study also uses machine learning to rank the variables according to their importance from largest to smallest according to the F-score. --- Logistic Regression Logistic regression modeling is a multiple regression analysis method used to study the relationship between dichotomous dependent variables and their influencing variables, i.e., it assesses whether an event occurs and what the probability of occurrence is when the influencing variables take on different values. Assume a vector X = with n independent variables, and assume that the conditional probability P = p is the probability of occurrence of the event in the dependent variable when the independent variables take values. The logistic model can be expressed as: P = π = 1 1 + e -g f = 1 1+e -g is called logistic function, g = ω 1 + ω 1 x 1 +. . . + ω n x n , then the probability of not occurring under condition x is: P = 1 -P = 1 - 1 1 + e -g = 1 1 + e g Therefore, the ratio of the probability of an event occurring to the probability of it not occurring is: P P = P 1 -P e g Logistic regression models can be solved iteratively by using the gradient ascent algorithm or by using the Newton-Raphson iteration, and the values of the dependent variable are , and after modeling, the probability values of the dependent variable represent the probability of the event. The logistic regression model can predict not only in-sample but also out-of-sample data, and it can compare and test the prediction results. --- DATA Data Source The data came from the CGSS questionnaire for two consecutive years-2017 and 2018. The survey is the first nationwide, comprehensive, and continuous large-scale social survey project in China, and it includes 125 counties , 500 streets , 1,000 neighborhood committees, and 10,000 individuals in households. A total of 25,369 samples were obtained from the survey data in 2017 and 2018, and according to the characteristics of the research subjects of this paper, the sample of people of childbearing age between 20 and 50 was selected. The sample was selected according to the characteristics of the research population, and after determining the relevant variables that conform to the research content of this paper, some "missing, " "don't know, " "indifferent, " "unable to answer, " and "not applicable" were excluded. Ultimately, 15,909 valid samples were obtained. --- Variable Description In keeping with the purpose of this study, the dependent variable is whether or not respondents are willing to have a second child. Specifically, the questionnaire asks, "How many children would you like to have if there are no policy restrictions?" If the answer is fewer than two , we assign the value "0"; for those who are willing to have two or more children, we assign the value "1." The independent variables include basic individual characteristics, family characteristics, and social characteristics. Individual characteristics include age, ethnicity, health, and education level. Family characteristics include the respondent's marital status, family economic level, and the number of people living together in the family. Social characteristics include the respondent's geographic location, hukou, religious beliefs, and decision about participating in medical treatment insurance. The specific variable descriptions are shown in Table 2. --- EMPIRICAL ANALYSIS --- Identification of Factors Influencing Fertility Intention From the descriptive statistical analysis, the influence of willingness to have two children on different independent variables is complex and variable. That relationship is non-linear, and machine learning methods can fit the non-linear information well. Therefore, this paper adopts the three commonly used machine learning methods previously mentioned in attempting to identify the influencing factors. --- Artificial Neural Network Modeling A total of 15,916 samples are included in the base data of this study. The categorical independent variables are transformed into dummy variables with horizontal signs, and the ordinal independent variables are transformed into factors with numerical signs. This may cause the model to learn the characteristics of those who are willing to have a second child, but not those who are not willing to have a second child, resulting in poor classification results. Therefore, in this paper, we use a down-sampling method to randomly select 3,366 samples of people who are willing to have a second child and obtain a total of 6,732 samples for modeling. The main purpose of this paper is to separate the data into two categories: those who are willing to have a second child and those who are unwilling to have a second child. Before training, the data are randomly split into a test set and a training set in the ratio of 3:7. In determining the topology of the ANN, a trial-and-error method is used to build a four-layer ANN after trying different ANN structures. The prediction results on the test set are obtained based on the training set model, and the confusion matrix with the actual categories in the test set is shown in Table 3. It is apparent that the category "willing to have a second child" is applicable to 1,078 people, of whom 597 are judged to be "willing to have a second child, " and 481 are judged to be "unwilling to have a second child." For the category "unwilling to have a second child, " 731 people are judged to be "unwilling to have a second child, " and 348 are judged to be "unwilling to fertility." For the "willing to have a second child" category, the precision rate of the model is 597/ = 55.4%, and the recall rate is 597/ = 63.2%. This means that about two-thirds of those who are willing to fertility are judged to be correct, while the model misclassified two-fifths of the sample as "willing to have a second child." The result of classification of the ANN is acceptable. The independent variables are more effective in distinguishing those who are willing to have a second child, but the model cannot give a numerical measure of which variables are more important for classification. In practice, there are significant differences in the values of "willing to have a second child" and "unwilling to have a second child" in terms of gender, age, health status, education, number of siblings, and region. In conclusion, although the ANN can fit the non-linear relationship --- Random Forest Modeling The same extended 6,732 samples are used to build the RF model. To determine the number of decision trees in the RF, the relationship between the out-of-bag error rate and the number of decision trees is drawn, as shown in Figure 2. When the number of decision trees reaches 286, the error rate reaches its lowest point and stabilizes; therefore, an RF model with 286 decision trees is established. The model obtained in the training set is applied to the test set, and the prediction results are compared with the actual values of the sample. The classification confusion matrix is shown in Table 4; of the 1,068 people who are willing to have a second child in the test set, 647 are judged as "willing to have a second child" and 421 are judged as "unwilling to have a second child." The actual number of those "unwilling to have a second child" in the test set is 1,089, among whom 659 are judged as "unwilling to have a second child" and 430 are wrongly judged as "willing to have a second child." For the "willingness to have a second child" category, the precision rate of the model is 647/ = 60.1%, and the recall rate is 647/ = 60.6%. The RF has lower precision than the ANN, but a higher recall rate. The RF classification model can play a vital role in the classification and can prevent the omission of important variables. The results of the importance assessment of the variables are shown in Table 5. Here, the top five most important variables are age, number of siblings, health, education, and district. Age is the most important factor-neither income nor health. The second-most important variable is the number of siblings, which reveals the importance of the family of origin. --- XG-Boost Modeling The values of the XG-boost model at each sample point are obtained with the AUC indicator due to the optimization target and the maximum depth of the number of five. To classify the samples into two categories based on the values of the sample points, an optimal threshold is determined based on the ROC curve , which is used as a threshold to segment the samples to achieve the highest accuracy of the classification results. The ROC curves in Figure 3 reflect the classification accuracy obtained with different values of the classification threshold. As demonstrated, the classification results are better when the threshold value is around 0.40. The best classification result is obtained when the threshold value is 0.40, and the AUC is the largest. The samples less than 0.40 are classified as "unwilling to have a second child, " and those greater than or equal to 0.40 are classified as "willing to have a second child." The confusion matrix of classification results and actual categories is shown in Table 6. Among the 1,068 people who are actually "willing to have a second child" in the test set, 620 people are judged by the model as "willing to have a second child, " and 448 people are judged by the model as "unwilling to have a second child." Among the 1,089 people in the test set who are actually "unwilling to have a second child, " 697 are judged by the model as "unwilling to have a second child, " and the remaining 392 are wrongly judged as "willing to have a second child." The precision rate of the model is 620/ = 61.3%, and the recall rate is 620/ = 58.1% for the category of "willing to have a second child." The precision rate of the XGboost model is in between those of the ANN and RF, but the recall rate is the lowest. The XG-boost model also examines the importance of variables, and the importance histograms are shown in Figure 4. The XG-boost modeling process plays an important role in age, number of siblings, education, health, and district. Although some of the other variables changed , they are consistent with the variables identified in the RF model, and the results of the two models can be corroborated by each other. FIGURE 3 | ROC curve. --- Comparative Analysis of Empirical Machine Learning Results To more objectively compare the results of each model, the modeling method of fivefold cross-validation is used, and the data on the average precision rate and average recall rate obtained are shown in Table 7. All of the average precision and average recall rates of the three models are above 60% under the fivefold cross-validation. The ANN has the highest mean precision rate , and the RF has the highest recall rate for the category "willing to have two children." The important variables obtained from the RF and XG-boost models-age, number of siblings, health status, income, and beliefs-are basically the same, although there are very small differences in ranks. Notably, economic level, although important but ranked sixth, is not as important as the previous five items. Therefore, we can validate hypotheses 1, 2, and 3: Individual characteristics, family characteristics, and social characteristics all have an important impact on the intention to have a second child in China. --- Logistic Regression If we use the machine learning classification model, it can only be determined that these variables are significantly related to "willingness to have two children." The exact quantitative relationship and significance would need to be determined by regression models. Since the dependent variable of the sample is a dichotomous variable, a logistic regression model is used for further empirical study. --- Logistic Regression of Full Sample The correspondence parameters and variables after using the logistic regression model are shown in --- Logistic Regression by Gender The logistic regressions by gender are shown in This paper is innovative in finding that age is the most important factor influencing the intention to have a second child, and the intention becomes stronger with age. This result contradicts previous findings that suggest that second-child policies are more likely to increase fertility intentions among younger cohorts . Economic factors have been the focus of research , and this study demonstrates that higher family income increases secondchild fertility intentions, the results of the logistic regression are statistically significant. However, the research shows that family income is only the sixth most important in both RF and XG-boost models of all variables. In other words, economic factors are not decisive as expected. Another breakthrough result in this study is that the larger the number of siblings, the stronger the intention to have a second child. Although the role of siblings has been discussed previously in the literature , the results of the machine-learning model show that it ranks second and is higher than health, education, and family income. At the same time, as the current literature has been lacking a regional comparison, this paper further confirms that district has an impact on second-child fertility intentions. Like the study by Chen et al. , we paper find the same result those who are healthier or have lower health risks will have higher fertility intentions for a second child. This paper considers gender differences as well. The relationship between education and fertility has been discussed in the existing literature and the results of the model show that education is in the top five of all factors. This paper uses logistic regression to further reveal that women with high school and university degrees are less willing to have a second child. The current literature is less designed for men, but this paper finds that men who are married and have a non-agricultural hukou are with higher intention to have a second child. In addition, religious beliefs may increase women's fertility intentions. In short, this study suggests that future fertility intentions should be explained more through the interaction of individuals with their family of original and their geographical areas, rather than concentrating too much on economic factors. It also provides implications for future governmental demographic stimulus policies: traditional family policies may not have the desired effect, and policymakers should focus more on family and socio-cultural orientations, and take into account regional and gender differences. Although this paper utilizes the latest CGSS data and adopts a machine learning approach, it still has some shortcomings, and there is room for improvement in the future. First, the feature variables found in this paper may still be insufficient because many factors affect the intention to have a second child. Second, the time span is not long enough. The implementation of the second-child policy started in 2016, so the relevant data are not abundant, and most of them are cross-sectional data. Third, models of machine learning are frequently updated, and this paper only considers mainstream learning approaches. Therefore, a potential improvement in future research can be applied from these three perspectives. University of China, NSRC. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin. --- 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. --- ETHICS STATEMENT The studies involving human participants were reviewed and approved by National Survey Research Center at Renmin --- --- SUPPLEMENTARY MATERIAL The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpsyg. 2022.883317/full#supplementary-material --- Conflict of Interest: 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. 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.

Although the Chinese government has shifted from a one-child policy to a two-child policy (allowing a couple to have up to two children) since 2016 in response to the aging population, the policy results have been unsatisfactory. This is the first paper to systematically investigate the factors influencing residents' intentions to have a second child. The research focuses on the perspective of individual, family, and social characteristics based on the Chinese General Social Survey (CGSS) from 2017 to 2018. Three machine learning methods are used in conjunction with logistic regression to reveal that the intention of having a second child increases heavily with age, more siblings in the family of origin, and better health. The family income, which is currently the focus of the literature and is statistically significant, is only sixth most important. This study further reveals differences between genders: Women with a lower level of education and religious beliefs prefer to have a second child, whereas for men, nonagricultural hukou and marriage are the position factors. The results of this study also illustrate the importance of future research focusing on the relationship of individuals to their family of origin and districts.

INTRODUCTION Outstanding success stories of membership-based websites -such as online social networks and online forums -captivate the attention of the general public and research community alike, inspiring a variety of studies dedicated to the description of the mechanisms behind their rapid growth [56,4,28,34,35,29,32]. But these models only describe the growth in membership rather than the number of members that visit the website per day, henceforth denoted daily active users . This omission is surprising since the DAU is undoubtedly a better proxy of a website's social impact and revenue generation potential than its total number of members [7,9]. The DAU is used to describe the financial health of websites in their quarterly reports, e.g., Facebook [25]. Websites with large but overwhelmingly inactive member bases are often deemed "failures" [7,55] and have low market values [9]. Ultimately, even the study of the online social phenomenon [33] is incomplete without understanding the mechanisms that can describe the DAU evolution of both successful and unsuccessful websites. This study fills this gap with a model that not only captures the DAU evolution observed in high-resolution datasets -collected by a leading Internet analytics company -but also predicts their evolution into the future. Our data consists of up to six years of daily measurements of the DAUs from twenty-two membership-based websites -encompassing online social networks, grassroots movements, online forums, political news organizations, and membership-only Internet stores -totaling over 55,000 website-day observations in a sample well balanced between successful and unsuccessful websites. These websites are diverse and in a constant state of change. Our primary goal is to find a common set underlying dynamics that are immutable throughout the website's life and can roughly describe -without getting into the particularities of each website -their membership dynamics, more specifically their DAU time series. To achieve our goal we present a simple model that uses a set of reaction-diffusion-decay equations describing the attention-seeking interactions between active members, inactive members, and not-yet-members of the website. The low barrier imposed to subscribe to a website means that an individual may subscribe to multiple websites [22], showcasing the importance of capturing members attention to the website's survival. And indeed, we note that today's successful membership-based websites seem to constantly bombard us with attention-seeking messages in the hope that we remain active members, which, as this study shows, is a key factor for their long-term survival. The end result is a model that embodies Simon's 1969 visionary remarks that, in an information-rich world our attention is bound to become one of our most scarce, important, and vied-for resources [48]. This work is organized as follows. Sec. 2 presents the related work. Sec. 3 presents our proposed model and the algorithm to fit its parameters to the datasets. Sec. 4 fits the parameters to the datasets and present the fit and prediction results. Finally, Sec. 5 presents our conclusions. --- RELATED WORK Adoption models describe phenomena as diverse individuals deciding to adopt a new technology [31,37,17,6,21,18] or individuals adopting a new health habit [11,10]. These models have deeply influenced the study of online social network growth [56,10,4,28,34,35,29]. Roughly, adoption models can be classified as: network effect adoption models [20,31,37,17,40,51,26,5,49,61,41], where individual rationality and adoption costs and utilities are modeled in a game-theoretic framework. Network effect models often assume strategic adopters, i.e., the decision to adopt a product depends on whether the product will gather more adopters in the future. Adopters also face adoption and switching costs in a game-theoretic framework [17]. Garcia et al. [20] uses structural features and classical network effect theory to predict whether an user departure from Friendster triggers other users departures. Garcia et al. does not consider the role of user activity and their analysis of Friendster's popularity rests on Google search volume data for searches of "www.friendster.com", which is likely an unreliable proxy of the true DAU values. threshold adoption models [23,46], where an individual adopts if enough of his or her friends are adopters; The threshold models [23,46] describe product adoptions in social networks, where an individual adopts a product if "enough" of his or her friends are adopters. Adopters are not strategic and have only a local view of the network. The notion of how many friends are "enough" vary from individual to individual and may also be determined by the number of friends and their interconnections [56,10]. diffusion of innovation models [6,21,18], where adopters influence others to adopt through word-of-mouth; In the absence of fine-grained individual-level data these models provide demand forecasting at the aggregate level. The Bass [6,21] and the Fisher-Pry [18] models are arguably two of the best known models of diffusion of innovations. In the Bass model, the diffusion of a product is driven by two forces: word-of-mouth and marketing . Diffusion of innovation models are also known as logistic models because of the S-shaped curve of the number of adopters over time, which replicates the adoption curve of a variety of real world measurements [38,6,39]. Diffusion models have also been applied in the field of sociology to describe diffusions of social behavior . adoption models from influence and network structure [56,10,4,28,34,35,29,32], where an individual adoptions depends not only on whether his or her friends adopt but also on how these friends are connected among themselves. Threshold models have inspired a variety of empirical research [56,10,4,34]. The probability that a non-member user joins the network is known to increase linearly with the number of invitations if the number of invitations is small, suffering a diminishing return effect as the number of invitations grow [56,10,4]. These findings would have a measurable effect in our problem formulation if most susceptible users were to receive a large number of invitations prior to joining the website, which is often not quantified [56,4] or is a parameter of the experiment [10]. A variety of works also consider the relationship between community growth inside an online social network websites and their network structure [29,4,34]. These studies, however, focus on the growth of communities inside the OSN and the role of network structure disregarding whether the community is alive or dead . Other works consider the popularity of entities such as news items and videos [60,16,53,14]. These works measure the total number of news readers or video viewers and, thus, can also be considered in the realm of adoption models because news readers generally do not reread the same article or re-watch the same video. In contrast, the mechanisms that drive members to return to a website depend on the complex interactions between non-members, new members, and active and inactive members. --- Modeling DAU dynamics. The above adoption models only describe the evolution of the total number of members. Modeling the DAU evolution requires the modeling of the attention-seeking interaction between active members and other Internet users. Recent Facebook shows that the activity of our Facebook friends in the website incites us to login and become active which, in turn, incites our friends to either become active or stay active [3]. This corroborates with our understanding that active members incite inactive members into activity, which should be one of the key driving forces behind a model of DAU evolution. We, however, are interested in membershipbased website in a broader sense than just online social networks. Be it a political news website, an online social network, or a grassroots movement, the interactions between active members, inactive members, and non-members is key to the understanding of the DAU dynamics. The work that most closely resembles ours is Cauwels and Sornette [9] which focuses on describing the evolution of the Facebook DAU. Cauwels and Sornette, however, is incomplete in the sense that its time-series analysis is tailored towards successful websites and cannot capture sudden DAU drops observed in unsuccessful websites. The question of what takes to increase the DAU is of great interest to the industry. Many complex factors help determine when and why members join and leave a website such as cultural and racial trends [7]. In this work, however, we do not analyze or model complex societal interactions and trends. Rather, we opt for a population-level model which we believe can be used as a foundation to build upon, providing valuable insights into the relationship between member activity, inactivity, and website growth, as seen next. --- PROPOSED MODEL Models of interacting populations have been successfully applied in fields as diverse as mathematical biology [42], economics [19], operations research [38,39], and market-ing [6,21,45]. Our model considers a large user population and the absence of sudden unpredictable events -such as a new round of VC funding that allows the website to significantly improve its appeal or increase its media & marketing exposure, or even the presence of strong new competitors -we model the population interactions as coupled reaction, decay, and diffusion processes. Reaction, diffusion, and decay processes find applications in chemistry, physics, and applied mathematics [12,13,58,57,42]. We choose to avoid stochastic models because very little is known about the stochastic behavior of the dynamics between inactive and active members of websites and between the latter and non-members. Our model can be described as follows. Consider the interactions between an active member A, an inactive member I, and a non-member U as reactions, decays, and diffusions in a solution. The first reaction A + I → 2A, describes the reaction with constant rate α as the marginal influence of active members to attract an inactive members to become active. This constant reaction approximates two real world phenomena: the activity of website members, such as posting pictures, events, messages, sending links to their friends, writing about their activity on online forums, instant & e-mail messages, face-to-face interactions, etc., attract the attention of inactive members making them active again 1 ; and the marginal increase in website utility as it gains more active members, an effect known as network effect or network externality discussed at length in Farrell and Klemperer [17]. There are many types of network effect, but the most widely used effect in its purest form can be described the following path-dependent cumulative return rule : "more buyers" → "lower prices" → "more buyers" or, alternatively, in a reaction better suited to our scenarios: higher DAU → more advertisement revenue → better website features → less inactive members . Members do not remain active indefinitely. Active members eventually become "temporarily" inactive. The decay A → I, describes an active member spontaneously becoming inactive with constant decay rate β. The second reaction A + U → 2A happens when an active member influences a non-member to join the website with constant reaction rate γ, which can happen either through word-of-mouth or because of increased utility , two widely known phenomena in the specialized literature [6,45,17]. Finally, the diffusion U → A 1 The reaction effect does not take into consideration the impact of communication constraints between members. This effect can be easily incorporated, see our older technical report [43] for a model; Additionally, over half of the websites analyzed in this study provide no explicit connectivity among their members, reducing the utility of reactions tailored to website-specific communication constraints in our analysis. describes the diffusion of marketing and media campaigns over non-members with diffusion rate λ, which then influences the non-member to join the website. Another parameter of our model is C, the fraction of the active Internet population that is targeted by the website service. In mathematical biology C is known as the carrying capacity. In our model we partition C into active members, A, inactive members, I, and non-members, U , which are part of the target niche that have not yet joined the website at time t such that A + I + U = C. As C is fraction of the active Internet population we do not need to add extra parameters to account for Internet population growth or the natural seasonal variability in the number of active members due to national holidays or school holidays. Gathering all these effects and collecting all the terms yields the following set of equations dA dt = - 1 C A 2 γ + 1 C IA + Cλ -A -Iλ , dI dt = Aβ - 1 C IAα , that describes the dynamics of A which is the DAU as a fraction of the active Internet population at time t. In the Appendix we see that the website joining rates and the size of the inactive population prescribed by the above model are consistent with metrics observed at large membership-based website. --- N.B. I: An earlier version of our technical report [43] provides model extensions that account for multiple marketing and media campaigns, growth in the target population C, account cancellations, and the sudden appearance of predatory competition by other website . These models are left out of this paper due to space constraints. --- DAU long-term activity After a certain -possibly large -time t the website nearly exhausts its pool of non-members and growth happens only when the Internet user population grows, i.e., U ≈ 0 for all t ≥ t . After time t the DAU must be sustained through the Active ↔ Inactive dynamics. For t > t we can then approximate I ≈ C -A; substituting in we obtain dA dt ≈ -Aβ + /C)Aα. and, thus, for t > t , A ≈ e -t 1 -α C e -t , where t = t + h, with h a constant. In equilibrium lim t→∞ A = 0 if β/α ≥ 1, C if β/α < 1. Equation predicts that any website whose ratio β/α ≥ 1 will eventually decay into inactivity, regardless of how much marketing and media exposure the website gets or how strong the word-of-mouth buzz is. More importantly, we can predict the function that describes its decay. The website fades into inactivity as A ∝ e -t . In what follows we see that these predictions match surprisingly well the observed data. Note that an increase in α has a great impact on the number of active members. Not surprisingly, online social network websites have recently been targeting member inactivity. For instance, Facebook in recent years has introduced notification messages of the form: "Here's some activity you may have missed" , even if almost no member activities are reported or these emails are consistently ignored by the user. Network growth is not the only reason why online social networks want to keep their members active. The website revenue is often tied to its DAU [9], mostly through online ads. But, most importantly, our model shows that the website survival depends on whether the ratio β/α ≥ 1, creating a great incentive to increase α through "here's some activity you may have missed" reminders. The DAU signature of equation . Equation predicts two distinct DAU signatures related to the coupling of activity ↔ inactivity reactions and activity decay. Fig. 1 shows two likely DAU signatures. The self-sustaining DAU behavior in Fig. 1 is observed when β/α < 1 and the initial DAU is higher than the asymptotic DAU level of C. The curve shows a slow decay from the initial DAU towards C. The unsustainable DAU time series in Fig. 1 is observed when β/α ≥ 1 and always converges to zero irrespective of the starting DAU value. --- DAU signatures of growth Intense word-of-mouth and media & marketing effects leave distinctive signatures. From a small number of active members an intense word-of-mouth suffers exponential growth in the initial DAUs ramp up phase due to the word-of-mouth reaction process feedback higher DAU → word-ofmouth growth → higher DAU. On the other hand, media & marketing diffusions have a characteristic convex growth that is independent of the current DAU value. In websites that depend heavily on media & marketing campaign growth has λ γ while in websites that depend heavily on word-of-mouth growth has λ γ. For t small such that ∀t < t we have U ≈ 1. When λ γ equation yields dA dt ≈ )λ, ∀t < t , λ γ. Fig. 2 shows an initial DAU time series signature of websites with media & marketing intensive adoptions. Similarly, when λ γ equation can be approximated dA dt ≈ Aγ 1 - A C , ∀t < t , λ γ. --- RESULTS Our results section has two objectives, namely: 1. Verify whether our model can be fit to the DAU time series observed in our datasets and whether the DAU time series behavior predicted by our model is observed in the datasets. 2. Automatically fit the model parameters to a training set of the data consisting of between 25% and 60% of the initial DAU time series, reserving the rest of the DAU data as a test set to test the DAU predicted by our fitted model. --- Datasets The DAU data was measured from June 2007 to June 2013 as a fraction of the active Internet population of twenty-two websites. The data is provided by Amazon's Alexa web analytics company totaling more than 40,000 website-day observations. The chosen websites encompass online social networks, online forums, political news websites, membership-based retail stores, and online social movements, all ranging from outstanding successes to catastrophic failures. We choose websites that either require membership or depend on the activity of a loyal user base . Unsuccessful websites urls were collected from TechCruch's failed startup's epitaphs [54] between February and August of 2012 and 2013 while successful websites names were gathered in the press and the related literature. As standard practice we smooth out the DAU outliers using a moving median with a 31-day DAU interval centered around each day. A description of the websites and further details about the data collection procedure are found in the Appendix. It is important to note that Alexa's datasets do not include smartphone traffic. Therefore we do not consider websites that have significant smartphone-based traffic . Facebook is the only exception because our Alexa's DAU metrics seem to have suffered little from the introduction of the Facebook smartphone app, i.e., despite the smartphone app Facebook users seem to still also access the website through their desktop or laptop computers. --- Automatic parameter fit Our model has five parameters, namely, α, β, λ, γ, and C. Notwithstanding the five-dimensional parameter space, the overall DAU evolution shapes allowed by the model effectively has fewer degrees of freedom. The reduction in degrees of freedom is due to the relationship between β and α explored in Section 3.1, the initial DAU growth differences between λ and γ explored in Section 3.2, and the magnitude of λ and γ which helps dictate the DAU initial growth rate. Note that C is just a resealing parameter and does not affect the shape of the DAU time series curve. Figures 4, 4, and 5 present three of the most common DAU evolution shapes observed in our datasets. The algorithm used to fit the model to the data works as follows. Let D be the set containing all datasets used in this study. We find the an optimal parameter fit to a dataset d ∈ D using the Levenberg-Marquardt algorithm [36]. We use the first few years of DAUs data to train the model and the remaining h years data as holdout data to evaluate the model predictions . The Levenberg-Marquardt algorithm only finds a locally optimal solution starting from an initial parameter guess m0 = . Hence, the initial guess m0 may significantly influence the output of the algorithm. To make the algorithm fully automatic and robust we need to find a principled way to provide this initial guess using our datasets. We provide the initial parameter guess by feeding the Levenberg-Marquardt algorithm with other parameter examples obtained in our datasets. It is reasonable to assume that the DAU time series of a given dataset d ∈ D is similar to the DAU time series of some other datasets in D. If we knew the best parameter fit of a dataset with similar DAU time series as d we could then use it to initialize the Levenberg-Marquardt algorithm. But as we don't know which datasets in D\{d} have similar DAU, we test all possible datasets as follows. Let LM be the Levenberg-Marquardt fitted parameters using initial guess m0. Let f d be the best currently known fit to dataset d ∈ D\{d}. At this stage our algorithm produces a set of candidate fitted parameters C = {LM : d ∈ D\{d}}. Instead of selecting the best fitted parameters from C, we make our parameter selection more robust by selecting at least three fitted parameters from C. We test the quality of the fitted parameters through a model selection phase. In the model selection phase we use the first three to six months of DAU data in the holdout data to assign an L2 error to the predictions of the model with each fitted parameters in C. Using these errors, together with the actual parameter values, we cluster the elements of C using k-medoids clustering . We choose to use k-medoids instead of the widely used kmeans because k-medoids is likely more robust to noise and outliers than k-means, in the same manner that the median of a set of measurements is more robust to noise and outliers than their mean. The output of our algorithm is then the k-medoid cluster c ⊆ C that has at least three elements and the smallest average error. More precisely, our algorithm reports c and the medoid of c. The medoid of c is the vector of parameters that best represents all vectors of parameters in c. The above procedure must be bootstrapped by manually assigning parameter fits to the datasets. But once we automatically obtain the medoid parameter fit of d we can replace its manually initialized value with the automatic one and restart the process. The R source code of our algorithm is freely available online to be tested on other datasets 2 . --- Model fitting results In what follows we present the model fitting results of our algorithm and later contrasts them against the real DAU time series. The magnitude of the fitted model parameters should be compared between websites as they need to be rescaled by C, which varies from dataset to dataset. For instance, the value of C for Facebook is much larger than that of any other website in our dataset. Figure 3 shows the medoids of the fitted parameters α and β for each of the twenty-two websites. The light blue area in Figure 3 indicates the long term unsustainable region where β/α ≥ 1. The DAU of websites in this region is predicted to die in the long run. In our plots whenever the parameter values are smaller than 10 -4 we declare the parameter as N/A indicating that the required precision may be beyond the numerical capabilities of our algorithm. Interestingly, websites that are known to have a sustained loyal fan base such as Facebook, LinkedIn, Salon, DailyCaller, and TheHuffingtonPost, are unmistakably in the self-sustaining area of the α and β parameter space. On the other hand, websites of social movements such as the Tea Party websites teapartypatriots.org and teapartynation.com, along with OccupyWallSt.org rest squarely on the "unsustainable" area. Internet fads such as 12seconds.tv and also belong to the unsustainable area. Other websites such as Flixster.com, FormSpring.me, and MarriedSecrets.com are near the border between self-sustaining and unsustainable. MarriedSecrets.com and AshleyMadison.com, unlike a regular dating websites, are designed to support extramarital affairs of both genders. We choose these dating websites instead of more traditional dating websites such as OkCupid.com and Match.com as we believe that people looking for extra-marital affairs are not likely to quit the website once they find a partner, thus making their behavior more similar to our model. Fig. 3 shows the medoids of the fitted parameters γ and λ and the respective website. Observe that most websites are classified as having a significant word-of-mouth component with a comparatively small media & marketing push. Few websites such as Occupy-WallSt.org and community.babycenter.com are classified as 2 http://www.cs.cmu.edu/~ribeiro/Ribeiro_AIU.zip having both strong word-of-mouth and media & marketing exposure. Only two websites were classified as having predominantly media and marketing campaigns: ebay.com and theblaze.com. Indeed, eBay has strong visibility on search engines for searches related to electronics and hard-to-find consumer products, to which its "marketing" visibility could be attributed. The TheBlaze.com result is more mysterious. TheBlaze.com is an online conservative news venture run by conservative pundit Glenn Beck [47]. Contrasting the collected DAU time series of TheBlaze.com in Fig. 5 with the shapes of strong media & marketing intensive DAU signatures in Fig. 2 and word-of-mouth intensive DAU signatures in Fig. 2 it is clear that TheBlaze.com is correctly classified as media & marketing intensive growth with little word-of-mouth growth. Unfortunately, it is unclear why TheBlaze.com presents an intensive media & marketing growth signature while similar political news websites such as DailyCaller.com and TheHuffingtonPost.com have the opposite classification . While our model is designed to capture just the basic mechanism that drives the DAU in the absence of sudden events that may happen during the website's lifetime -e.g., deep website redesigns, extra marketing and media exposures, and arrivals of new strong competing websites vying for the attention of the website members -in what follows we observe that even in the presence of unpredictable DAU bumps and spikes our model is able to capture the DAU time series and predict its long-term trend. --- DAU fit & long-term predictions Here we present the data fit and the predictions made by our model. First, however, we introduce the most common combinations of the DAU signatures of self-sustaining v.s. unsustainable and word-of-mouth v.s. marketing & media intensive growths in our datasets. Figs. 4 and 4 use our model to show the most common DAU shapes of unsustainable websites observed in our datasets. Fig. 5 uses our model to show a common word-of-mouth intensive DAU shape of sustainable websites. Few websites are classified as sustainable with media & marketing intensive growth . One of our goals in this section is the challenging task of presenting multi-year DAU predictions into the future based on the various vectors of parameters obtained by our algorithm from the training and model selection data. It is worth noting that parameterizing the model using only DAU information has its limitations. It is only possible to classify a website as unsustainable if the website shows signs of decreasing DAU activity. Interestingly, however, is that not all websites showing signs of decreasing DAU activity are classified as unsustainable, as exemplified by the website MarriedSecrets.com. In Figs. q q q q qq q q q q q q qq q q q q qqq qq q q q q q qq qqqq qq qqq q q qq qqqqqqqqq qqqqqq q qqqq 12seconds.tv q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q ruelala.com q qqq q q q q q q q q q q q qq q q q q q q q q q q q qq q q qq q qq qq qq q q q q qqqq q q qq q q qqqqqqq true.com Fig. 4 shows the results of the twelve websites automatically classified as unsustainable by our algorithm. We observe that the model fits the data reasonably well, except for a few "DAU bumps", possibly the product of website changes or media & marketing investments. In general, however, our model is able predict the overall website trend years in advance. Three anomalies are the overestimation of the DAU of True.com and Adaptu.com. Per our Tech Report [43] the linear decay of Adaptu.com's DAU shows the characteristic signature of a strong competitor , probably Mint.com [54]. Our fitting algorithm also tends to output some parameter assignments for Ruelala.com as self-sustaining. Brand-Stack.com was acquired in late 2011 by DesignCrowd.com and has since changed its url and thus we truncate its data at the time of acquisition. The predicted DAU trend of Brand-Stack.com seems accurate if the website was allowed to continue its trajectory. The predictions of Flixter.com, Occu-pyWallSt.org, 12seconds.tv, FormSpring.me, TeaPartyNation.com, True.com, and TeaPartyPatriots.org are remarkably accurate. Interestingly, not even the renewed interest in Tea Party during the 2012 U.S. presidential election cycle was able to disturb the long-term trajectory of TeaPartyNation.com and TeaPartyPatriots.org towards what appears to be a negligible DAU. Fig. 5 shows the DAU time series of twelve websites automatically in our dataset that are classified as self-sustaining by our algorithm. One of the most remarkable predictions of our model is the long term stabilization of the DAU of self-sustaining websites. The datasets overwhelmingly confirm this prediction. While websites keep constantly changing to broaden their audience , it seems that once the website audience is determined the website %DAU trajectory towards C A Common selfsustaining + word-ofmouth DAU shape. q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q huffingtonpost.com q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q community.babycenter.com q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q linkedin.com q q q qq q q q q q qqq q qq q q q q q q q q q qq qq q q qq qq qq q q qq q q qq q q qq qq q q q qqqq q q q marriedsecrets.com q q q qq q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q theblaze.com qq qqq qqqqqqqqqqq q q qqqqq qq q qqq q q qqq q q q q q qq qq q q q q qqqqqq q q q q q q qq q netflix.com is set. One of the most remarkable examples of this observation is Facebook.com. Three out of the four outputs of our algorithm predict three years into the future -and with surprising accuracy -Facebook's final DAU size and time series. The training + model selection data has Facebook's DAU reaching up to 37% of the active Internet population and our model is able to extrapolate that Facebook would eventually reach about 45% of the %DAU and then stabilize. The DAU fit and predictions of our model for AshleyMadison.com, Facebook.com, PatientsLikeMe.com, CafeMom.com, MarriedSecrets.com, Webchat.Freenode.net, and Netflix.com worked remarkably well. Both Meetup.com and Despite the high variability of predictions at Meetup.com and Webchat.Freenode.net, these predictions seem to agree well with the DAU data . Curiously, while the curve fit for TheHuffingtonPost.com is remarkably accurate, the model has a strong tendency to stabilize the DAU while the true data shows a growing DAU until stabilization. The reason behind this apparent mismatch may be the following. At the beginning of the TheHuffingtonPost.com life its DAU growth was mostly due to word-of-mouth, which is well captured in the model parameters over the training data. However, reports show that somewhere between early 2011 [24] and late 2013 [50] the TheHuffingtonPost.com started to be indexed by the Google News service, bringing a large amount of "unpredictable" media & marketing growth to the The-HuffingtonPost DAU, thus the DAU underestimation of our model. Similar to the TheHuffingtonPost.com the DAU of LinkedIn.com and community.babycenter.com are also shown to be self-sustaining but the final DAU is underestimated. Unfortunately the reasons behind these two underestimations are unclear. The website TheBlaze.com seems to have also slightly deviated from its predicted DAU. In the first 112 days into the prediction the website DAU follows the model remarkably well. Later, however, TheBlaze.com experiences two unpredictable DAU spikes and, unfortunately, only time will tell if the model prediction is indeed showing the long-term DAU plateau of the website. --- CONCLUSIONS Our study sheds light on the mechanisms of growth and member activity and inactivity of membership-based websites. Through reaction, diffusion, and decay processes we model the dynamics of website member activity, inactivity, and growth, mostly focusing on the DAU metric. We showed that our model predicts two general DAU signatures of growth and two DAU signatures of long term stability . We proposed an algorithm to fit the model parameters to realworld DAU time series data. Finally, from the DAU time series of twenty-two websites we show that our model not only fits well the DAU data but can also predict its future evolution. This work makes a positive step not only towards modeling the dynamics of websites but possibly also towards modeling a broad range of dynamics of societal movements, such as the activity and growth of grass-root organizations. There is, however, much left to do to tailor the above reactiondiffusion-decay dynamics to specific types of websites and changing environments . --- ACKNOWLEDGMENTS The author would like to thank the members of the MoBS lab at Northeastern University for their kind hospitality and insightful discussions, Sharon Goldberg for pointing out some missing related work, Don Towsley for encouraging the generalization of an earlier version of this work, and A. Clauset for bringing to the author's attention his study on online multiplayer networks. This work was partially done while the author was at the University of Massachusetts Amherst. This work was supported by NSF grant CNS-1065133 and ARL Cooperative Agreement W911NF-09-2-0053. The views and conclusions contained in this document are those of the author and should not be interpreted as representing the official policies, either expressed or implied of the NSF, ARL, or the U.S. Government. The U.S. Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation hereon. --- APPENDIX --- Simpler model oversimplifies problem In this part of the Appendix we show that the simpler model in N.B.II cannot characterize a vanishing DAU time series. Contrast the prediction for Flixter.com made by the simpler model in Fig. 6 against the predictions of our complete model in Fig. 6. In both cases the training data used to generate these graphs is slightly smaller than the one used in our previous experiment in Fig. 4. Note that the simpler model shown in Fig. 6 incorrectly stabilizes the DAU while the data and our correct model in Fig. 6 indicate that the Flixster.com DAU converges towards zero. --- Dataset description The web traffic data used in the evaluation of our model was obtained by the commercial web analytics company Alexa.com, a subsidiary of Amazon.com. Today, Alexa provides traffic data, global rankings and other information on 30 million websites [1]. Alexa ranks sites based primarily on tracking information of users of its toolbar available for all the Internet Explorer, Firefox and Google Chrome web browsers. Since 2008 Alexa claims to remove self-selection bias -bias related to gathering data of a specific audience subgroup that is more likely to install Alexa's toolbar -by taking into account other data sources "beyond Alexa Toolbar users" [1], but the nature of such data sources and the methodology employed are not disclosed. Nonetheless, because Alexa's report is detailed and widely used in the industry, we believe that Alexa's unique subscriber daily traffic reports are a good source of data for our study. The following websites were used: q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q Incorrect simpler model prediction time % DAU 1 yr 2 yr 3 yr 4 yr 5 yr 6 yr 0 937.5 1875 2812.5 3750 x10 -6 q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q --- Model & the growth of MySpace Online social network websites rarely allow access to their subscriber activity data. To complement the analysis of the preceding sections we use a complementary source of data. This dataset records the activity of 1. Average degree y ~ 10.3 x q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q appeared [27] and in April 2008 Facebook usage overtook MySpace [59]. We start our analysis with the relationship between network growth and member activity. The member activity lifespan is defined as the period between the member join date and the member's last login date. In what follows we make the following reasonable assumption: member lifespan and member activity rate are linearly dependent; and a member active for x semesters has logged in at least once per semester. In our data we observe that members with an activity lifespan of x weeks have on average 10.3x friends . This constant rate of edge acquisition is consistent with measurements in other OSNs [62,34] and is consistent with our model. It is worth noting that Leskovec et al. [34] measures the rate of edge creation with respect to member's age in the website rather than member lifetime as we do. Leskovec et al. measurements are less correlated with the DAU than our measurements. In Fig. 8 we also see a linear relationship between member activity and the network monthly growth between the first semester of 2004 until the end of the second semester of 2006 . The green bars show the number of new subscribers of each year observed in our ran-dom sample while the black line shows the number of active subscribers . Assuming that the activity rate of a subscriber grows linearly with his or her lifespan, we conclude that between 2004 and 2006 for every active subscriber MySpace acquired a new subscriber. The above measurement is also consistent with our model. The inset of Fig. 8 shows the main plot including the years 2007 and 2008. Note that the network growth rate decreases significantly in 2007 and 2008 when compared to previous years . Another interesting aspect of the 2007/2008 decline is that -as seen later in our earlier Technical Report [43] -Facebook's competition seems to have only significantly affected MySpace by late 2008. The notion that the Facebook competition may not explain MySpace's falling numbers of new subscribers in 2007 is corroborated by estimating the lifespan distribution between subscribers that joined in 2006 and 2007 using the Kaplan-Meier [30] estimator , which reveals little change in the subscriber lifespan distribution between the years where the number of new members first dropped. Thus, the unchanging lifespan distribution and the constant activity of subscribers in 2007 both indicate that there is little Facebook effect in the dwindling growth of MySpace. This means that MySpace growth saturated near early 2008 and the fraction of non-members in the population, U , started to converge to zero over time and that the member growth is bell shaped in self-sustaining websites, which can be easily verified to be also consistent with our model. and July-December . We consider subscribers with 60 or more days of inactivity at the time of measurement to be permanently inactive.

Driven by outstanding success stories of Internet startups such as Facebook and The Huffington Post, recent studies have thoroughly described their growth. These highly visible online success stories, however, overshadow an untold number of similar ventures that fail. The study of website popularity is ultimately incomplete without general mechanisms that can describe both successes and failures. In this work we present six years of the daily number of users (DAU) of twenty-two membership-based websites -encompassing online social networks, grassroots movements, online forums, and membership-only Internet stores -well balanced between successes and failures. We then propose a combination of reaction-diffusion-decay processes whose resulting equations seem not only to describe well the observed DAU time series but also provide means to roughly predict their evolution. This model allows an approximate automatic DAUbased classification of websites into self-sustainable v.s. unsustainable and whether the startup growth is mostly driven by marketing & media campaigns or word-of-mouth adoptions.

Introduction COVID-19 has emerged as a global health emergency and posed a great threat to almost all countries and regions . It affects all segments of the population, especially the patients of COVID-19 . The impact is far beyond merely physical concerns. Previous studies have shown that the pandemic has led to psychological problems among patients, healthcare workers, and other caregivers . Patients infected with COVID-19 not only suffered from illness, but also had mental health problems due to viral infection and worries about after-effects . Perceived stigma is prevalent among COVID-19 survivors and healthcare workers in COVID-19 designated hospitals, which has an interrelated bearing on their mental health . In post pandemic era, most patients of COVID-19 have been discharged . The mental health of those who had recovered from COVID-19 and been discharged from hospital deserve more attention during their rehabilitation . These patients were isolated during treatment and had limited freedom and communication with the outside world . Thus, their negative emotions cannot be alleviated in a short period of time. RD may have a more serious sense of loneliness and repression, as well as a higher level of psychological pressure . In the aftermath and the longcovid period, they may experience depression, anxiety, fatigue, posttraumatic stress disorder, and neuropsychiatric syndromes . Poor mental health condition will impact one's social behaviors and cognitive functions. As a result, RD's mental health should be attached much importance. RD's mental health condition might affect their perceived COVID-19 stigma . Perceived stigma is one's personal feelings about the stressors and his projection of the feelings on others . From the patient's perspective, they might feel being stigmatized if their mental health condition was poor. COVID-19 RD are at high risk of PTSD, partly because of their near death experience, delirium, and ICU-related trauma during the COVID-19 experience . They might have uncontrollable thoughts about the experience and their image in others' mind, which would increase their perceived stigma. Perceived stigma might also in turn predict PTSD . Depression is another prevalent mental issue among COVID-19 RD . RD with depressive symptoms might be more sensitive and pessimistic to the negative attitudes from the community, which makes them feel more stigmatized emotions . Besides, to contain the spread, patients are required to stay in close isolation during treatment and reduce their movement after discharge, which may lead to feelings of loneliness and fear of discrimination, thus increasing their perceived stigma . Peace of mind is important for them to manage stressful situations, as well as avoid the irresistible but unwanted impulses . Resilience is not a linear path toward happiness, but a combination of behaviors that encourages individuals and communities to persevere and move forward confronting difficult situations . Higher level of resilience might decrease the risk of developing psychological distress, and suppress suicidal thoughts and insomnia . Resilience might be influenced by job stress, perceived stress, and mindfulness, and be promoted by brief resilience interventions based on positive psychology . Thus, with higher level of peace of mind and resilience, patients will control their emotions better and be less sensitive to the negative attitudes from others, which might result in lower sense of perceived stigma. From the society's perspective, low perceived social support may also lead to perceived stigma among COVID-19 RD . Perceived stigma might in turn increase the mental problems among RD and be detrimental to their mental health recovery . Therefore, the stigma among COVID-19 RD may have a certain impact on the whole population. The perceived COVID-19 stigma in RD could be evaluated by a modified 12-item HIV stigma scale, which contains 4 sub-scales to measure personalized stigma, disclosure concerns, concerns about public attitudes, and negative self-image . However, this scale has no cut-off point, which makes it hard to precisely evaluate the stigma among RD. Clinical psychiatric interviews are usually regarded as the gold standard for diagnosis and the criterion for determining cut-off points of screening tools, However, the identification and diagnosis of cases with perceived COVID-19 stigma has not reached a consensus. Additionally, the characteristics and prevalence of perceived COVID-19 stigma among RD and its psycho-social influencing factors remain elusive. Currently, most previous studies focused on the recursive effect of perceived stigma on mental health without considering the possible vicious circle between mental health and perceived stigma among RD. While according to the theory of socio-ecological model, one is not a passive recipient of life events, but a key role in constructing and modifying the living system . It is therefore important to explore the influencing factors of perceived COVID-19 stigma among RD. The specific objectives of current study are to identify the characteristics of perceived COVID-19 stigma in RD using latent profile analysis ; to explore the psycho-social influencing factors of perceived COVID-19 stigma in RD; and to determine the cut-off point of the stigma scale using ROC analysis for further evaluation and application, which may help healthcare professions and policymakers to deal with the increasing stigma and control the pandemic effectively. --- Methods --- Study design and participants The cross-sectional study was carried out among previouslyinfected COVID-19 patients in Jianghan District from June 10 to July 25, 2021. Extracted from the electronic medical records of the Jianghan District Health Bureau, a total of 3,059 COVID-19 patients met the inclusion criteria and were eligible for the study, for they were infected with the original SARS-Cov-2 strain and were diagnosed between December 10, 2019 and April 20, 2020. When they were receiving clinical re-examination, 1,601 COVID-19 survivors were invited to complete a questionnaire survey on their mental health status, and 1,541 of them who finished the survey were included in the study. All investigators and support staff in this study were trained according to the same protocol and required to have an educational background in medicine or public health. From June to July 2021, the online structured questionnaire was distributed to those who had a history of COVID-19 infection and had been discharged. All participants' digital informed consent was obtained to ensure their voluntary participation. An online survey platform Redcap was used to disseminate the self-administered electronic questionnaires and digital consent to the target population. --- Stigma The Short Version of COVID-19 Stigma Scale is a 12-item scale that is employed for evaluating the perceived stigma of patients of COVID-19 during the past 2 weeks . The scale was reviewed by several experts in the field and was approved to use in this population. Each item is scored on a Likert scale of 1-4. Higher total scores indicate greater stigmatization. In this study, the Cronbach's alpha of the instrument was 0.936. --- Post-traumatic stress disorder The Impact of Events Scale-Revised is a 22-item scale aimed at screening posttraumatic stress symptoms in adults or older people. The items of this instrument are rated on a 5-point Likert scale from 0 to 4 . The IES-R contains three dimensions measuring intrusion, avoidance, and hyperarousal. Respondents rate their degree of distress during the past 7 days after they have identified a specific stressful life event that occurred to them. A total score equal to or above 35 can be regarded as positive PTSD symptoms. This instrument has been proven valid and reliable among Chinese COVID-19 patients . In this study, the Cronbach's alpha of the instrument was 0.965. --- Anxiety The Generalized Anxiety Disorder Questionnaire consists of 7 items that are rated on a 4-point Likert scale from 0 to 3. It was developed for measuring the severity of generalized anxiety symptoms during the past 2 weeks . The scores of the instrument range from 0 to 21. A cutoff score of ≥ 5 is recommended for considering significant anxiety symptoms. This instrument has demonstrated to be reliable and valid among the Chinese population . In this study, the Cronbach's alpha of the instrument was 0.951. --- Depression The Patient Health Questionnaire is a 9-item questionnaire that is used for screening and monitoring depression of varying degrees of severity during the past 2 weeks . The items of the PHQ-9 are rated on a 4-point Likert scale ranging from 0 to 3. The total score is utilized to assess the degree of depression of participants, with scores of ≥ 5 indicating depression. This instrument has been validated among various Chinese populations . In this study, the Cronbach's alpha of the instrument was 0.914. --- Sleep disorder The Pittsburgh Sleep Quality Index consists of 18 items and is used to measure an individual's quality of sleep during the past 2 weeks . It contains seven components including subjective sleep quality, sleep latency, sleep duration, sleep efficiency, sleep disturbance, use of sleep medication, and daytime dysfunction, and each component is a 4-point Likert scaled from 0 = no difficulty to 3 = severe difficulty. The total scores range from 0 to 21 and a cutoff score of ≥ 6 is recommended for considering certain sleep disorders . This instrument has been validated among Chinese population . In this study, the Cronbach's alpha of the instrument was 0.784. --- Fatigue The Fatigue Scale-14 is a 14-item scale aiming at measuring the severity of fatigue during the past 2 weeks . The items of this instrument are rated on a 2-point scale of 0-1. The FS-14 contains two dimensions measuring physical fatigue and mental fatigue, respectively. Higher total scores of the 14 items indicate a higher level of fatigue. This instrument has been proved valid and reliable among Chinese . In this study, the Cronbach's alpha of the instrument was 0.845. --- Resilience The Resilience Style Questionnaire consists of 16 items that are rated on a 5-point Likert scaled from 1 to 5. It is used to measure the level of an individual's resilience during the past 2 weeks . Higher total scores of the 16 items indicate a greater ability to recover from negative events. This instrument was developed and validated among the Chinese rural left-behind adolescents and non-local medical workers . In this study, the Cronbach's alpha of the instrument was 0.975. --- Social support The level of perceived social support of the participants was measured by two items including emotional support and material support during the past 2 weeks . The items were: "How much support can you obtain from family/friends/colleagues when you need to talk or to obtain emotional support?" and "How much support can you obtain from family/friends/colleagues when you need material support ?" and each item was 11-point Likert scaled from 0 to 10. In this study, the Cronbach's alpha of the instrument was 0.819. --- Peace of mind The Peace of Mind Scale comprises a total of 7 items rated on a 5-point scale ranging from 1 to 5 and is used for measuring the peace of mind during the Frontiers in Public Health 04 frontiersin.org past 2 weeks . Higher total scores indicate a more peaceful mind. This instrument has been validated among Chinese population . In this study, the Cronbach's alpha of the instrument was 0.874. --- Statistical analysis Descriptive analyses were performed to describe the participants' demographic characteristics, clinical characteristics, the condition of perceived stigma, and potential influencing factors. In the absence of an accurate and precise reference standard, LPA has been widely employed to identify the symptom characteristics and to further calculate and determine optimal cut-off points of assessment instruments . LPA is a person-centered statistical method that employs latent profile model to divide population into multiple profiles, and it focuses on identifying latent subpopulations within a population based on a set of continuous variables . Despite the possible arbitrariness for LPA in determining the number of class members due to its semi-subjective properties, the misclassification rate is relatively low, and it could produce more reasonable results compared with some other classification approaches . Generally, in LPA, individuals assigned to the latent profile that represents the lowest level of symptoms or risks are regarded as "non-cases, " and others are considered "cases" . Hence, LPA was conducted to identify the characteristics of perceived COVID-19 stigma among RD. Robust maximum likelihood estimation was employed to estimate the parameters. The Lo-Mendell-Rubin and the bootstrap likelihood ratio test were performed to compare the model fit improvement between models with k classes and k-1 classes, significant p values indicated a better model fit with k classes. The optimal number of classes was evaluated by the entropy, Akaike Information Criterion , Bayesian Information Criterion , the adjusted Bayesian Information Criterion , and the interpretability and definition of classifications, where an entropy value≥0.80 represented adequate quality of classification, lower AIC, BIC, and aBIC values indicate better model fit, and the "turning point" of the scree plot for the aBIC could suggest an appropriate number of classes. After the selection of optimal model and definition of classifications, Chi-square began with the full set of demographic and clinical characteristics, PTSD, anxiety, depression, sleep disorder, fatigue, resilience, social support, and peace of mind, to evaluate their associations with different characteristics of perceived COVID-19 stigma. Statistically significant variables in the univariate analysis were further used for stepwise multinomial logistic regression analysis. Adjusted odds ratio and the corresponding 95% confidence intervals were calculated to assess the regression model results. Receiver operating characteristic analysis was conducted to determine the optimal cut-off value for the CSS-S. The area under the ROC curve , sensitivity, specificity, and Youden's index value were employed to evaluate the performance of classifiers, and Youden's index value was used to identify the optimal cut-off value. SAS9.4 and Mplus8.3 were utilized to conduct all the analyses with level of significance determined at a 0.05 value of p. --- Results --- Demographic characteristics Among the 1,541 people who finished the survey questions, 1,297 questionnaires were enrolled in the data analysis. As illustrated in Table 1, over half of the participants were male and were less than or equal to 60 years old . The majority of the participants were from urban areas and married . Most of the participants had an income for 2020 less than 60,000 China Yuan, and had an education level as senior high school or below . A small percentage of participants lived alone , used alcohol no less than 2 times per week , and were current smokers . The COVID-19 patients were clinically classified into four categories: asymptomatic , mild , moderate , critically severe . A significant proportion of the participants had no experience at ICU , had never received psychological or emotional counseling during hospitalization , and had never received psychological or emotional counseling before infection . Just under a half of participants stayed over 20 days in hospital , and had no complication . Most of the patients perceived good or moderate mental health status during hospitalization. --- Stigma and related psychological factors The 12-item CSS-S's total scores range from 12 to 48 with higher scores indicating a more stigmatizing attitude. The mean score in this study was 28.04 . The mean scores of fatigue, peace of mind, resilience, and social support were 6.38 , 24.70 , 56.82 , 14.25 , respectively. The prevalence of PTSD, anxiety, depression, and sleep disorder were 16.5, 28.8, 37.9, and 47.1%, respectively . --- Latent profile analysis Latent profile models with one-to-five-class solutions were specified, and the fit indices of the 5 models are displayed in Table 3. The entropies of all classifications were above 0.9. The LMR and BLRT test were all statistically significant. The AIC, BIC and aBIC decreased with the increase of class number, and the scree plot of aBIC flattened out after the 3-class model . Taken together, considering the model fit, parsimoniousness, and interpretability of the classes, the 3-class model was selected as the optimal model for the current sample, the distribution and conditional means of items of CSS-S on each class in the 3-class model are illustrated in Figure 2 and Table 4. In the 3-class model, the average latent class probabilities for most likely latent class membership demonstrate reasonable classification and good distinction . Given the conditional means of items on each class, we define Class1 as "low perceived COVID-19 stigma" group, Class2 (n = 663, --- Influencing factors of perceived COVID-19 stigma of RD The result of univariate analysis showed that female , older age , being married , low family income , living with other people , low education level , having complication , perceiving worse mental health status during hospitalization , PTSD , anxiety , depression , sleep disorder , and fatigue were positively associated with perceived COVID-19 stigma, while resilience , social support , and peace of mind were negatively associated with perceived COVID-19 stigma among RD . These variables were further employed in stepwise multinomial logistic regression analysis with the "low perceived COVID-19 stigma" group as a reference. The result of stepwise multinomial logistic regression analysis showed that older age , living with other people , anxiety , and sleep disorder were positively associated with moderate perceived COVID-19 stigma, while higher educational level was negatively associated with moderate perceived COVID-19 stigma; Female , older age , living with other people , anxiety , and sleep disorder were positively associated with severe perceived COVID-19 stigma, while higher educational level , social support , and peace of mind were negatively associated with severe perceived COVID-19 stigma among RD . --- Receiver operating characteristic analysis To identify the optimal cut-off value of CSS-S for screening perceived COVID-19 stigma among RD, participants assigned to the "low perceived COVID-19 stigma" group in LPA were defined Scree plot of change trend of adjusted Bayesian Information Criterion . as "non-cases" , and those assigned in "moderate perceived COVID-19 stigma" and "severe perceived COVID-19 stigma" groups were defined as "cases" . The ROC curve was then plotted for the total score of CSS-S using the binary outcome, with an AUC value of 99.96% , indicating a good predictive capacity for perceived COVID-19 stigma . The diagnostic criteria and indices are illustrated in Table 8. The optimal cut-off value was ≥ 20, where the sensitivity, specificity, and Youden's index value were 0.996, 0.982, and 0.978, respectively. --- Discussion The cross-sectional study employs LPA to assess the characteristics of perceived COVID-19 stigma among RD and analyzes its psychosocio contributing factors. Perceived stigma of RD was divided into three categories in this study. We measured the demographic characteristics and some possible psychological predictors of perceived COVID-19 stigma. Generally, older age, living with other people, anxiety, and sleep disorder were positively associated with moderate perceived COVID-19 stigma, while higher educational level was negatively associated with moderate perceived COVID-19 stigma; female, older age, living with other people, anxiety, and sleep disorder were positively associated with severe perceived COVID-19 stigma, while higher educational level, social support, and peace of mind were negatively associated with severe perceived COVID-19 stigma among RD. The cut-off point of the stigma scale was determined at 20 using ROC analysis. This study classified COVID-19 RD into three groups according to the stigma level: "low perceived COVID-19 stigma, " "moderate perceived COVID-19 stigma, " and "severe perceived COVID-19 stigma" group. Only 12.8% of RD were categorized into the "low perceived COVID-19 stigma" group, which indicated the lowest levels of stigma and reported the lowest level of psychological risk factors. The majority belonged to the "moderate perceived COVID-19 stigma" . Compared with the "low perceived COVID-19 stigma" group, anxiety and sleep disorder were positively associated with moderate perceived stigma. Similar to previously published studies, anxiety was a major risk factor for stigma. In a study that evaluated the depression and anxiety symptoms among 174 patients who recovered from symptomatic COVID-19 infection in Saudi Arabia, the stigma scores were significantly associated with higher scores on anxiety . Some other studies on people living with epilepsy, dementia, and cancer patients also demonstrated that anxiety is one of the psychosocial determinants of perceived stigma . Therefore, mitigating the anxiety symptoms is essential to decrease the stigma among RD. Emotional regulation, mindfulness, and experiential techniques are possible solutions to improve social anxiety disorder symptoms . RD could also try exercise, yoga, and meditation, which were proven to have modest positive effect on assisting their anxiety alleviation . Hospitals and communities should assess the anxiety level of COVID-19 RD to detect anxiety as early as possible. For RD . The society should be less hostile to RD. It is necessary for social media to refute false information, strengthen the information guidance of social media, and output positive information, so as to avoid the anxiety mood in origin. Our study also found that sleep disorders is a determinant of moderate perceived stigma in RD. Previous studies showed that 29.5% of the COVID-19 hospitalized patients had sleep disorders . Poor sleep quality was associated with stigma . Cognitive behavior therapy is aimed at treating insomnia by avoiding behaviors and thoughts that might develop into sleep disorders . RD with sleep disorders could use this method on their own to improve their sleep quality. Effective programs based on the therapy could also be embedded in smartphones to assist their sleep promotion process . In addition, progressive muscular relaxation is an effective way to help COVID-19 patients feel less anxious and have better quality sleep . The receiver operating characteristic curve of the CSS-S for screening perceived COVID-19 stigma. The "severe perceived COVID-19 stigma" group reported three more risk factors compared with "moderate perceived COVID-19 stigma" group, including female gender, insufficient social support and peace of mind. Female gender is a risk factor of "long-covid" syndrome and tend to have a higher proportion of physical and psychological symptoms than male . Because of the more severe illness and torment they suffered, they might find it difficult to maintain a good mentality toward the stigmatized attitudes. A low perceived level of social support prevailed during the pandemic due to the shutdown of many places, like schools, markets, and workplaces to avoid transmission of the virus . RD facing such conditions may arouse a sense of isolation and vulnerability, which would cause severe stigma. Perceived social support and use of adaptive coping strategies were found to affect individuals' psychological adjustment and resilience . Interventions like in-person interview, supportive psychotherapy, and positive attention would improve their social support and could be considered widely promoted . Peace of mind might increase one's self awareness and attitude toward the surroundings, and indirectly reduce the sense of being stigmatized. A previous study on female patients with schizophrenia also identified that enhancing peace of mind will help reduce stigma level . Our study determined 20 as the cut-off score for CSS-S by LPA and ROC analysis, which may guide future epidemiological studies on COVID-19 stigma. The cut-off value is instructive for clinical practice in COVID-19 RD mental health promotion. Hospitals are suggested to collect stigma information of discharged patients and carry out relevant psychological intervention for patients whose scores exceed 20. Although our team have analyzed the same population in advance and explored the prevalence and influencing factors of anxiety and depression in RD , a further analysis in this study provided insightful observations from a different perspective. This study enriched our knowledge on the association between mental health and perceived stigma among RD, and provided possible suggestions for the authorities and the society to reduce perceived COVID-19 stigma in the future. However, it has several limitations. First, this crosssectional study has its inherent limitations, for it contains no dimension of time to support a causal relationship. Second, the study was conducted more than 18 months after the COVID-19 patients were discharged, which may cause recall bias. Third, convenience sampling may decrease the representativeness of the population. Fourth, stigma contains two factors, namely "public stigma" and "selfperceived stigma. " In this study, we only mention the latter. Further studies should measure stigma more comprehensively in a representative sample. --- Conclusion This study provides an insightful result of the prevalence and influencing factors of perceived stigma among RD in Wuhan. Stigma among COVID-19 RD could be divided into 3 groups: "low perceived COVID-19 stigma, " "moderate perceived COVID-19 stigma, " and "severe perceived COVID-19 stigma" group. Based on the cut-off value we explored, the high proportion of perceived stigma level highlights the importance of solving the stigma and discrimination problem, for its impact on personal and community well-being. Therefore, it is essential to mitigate the psychological problems and reduce the perceived stigma level of RD as part of the response toward the COVID-19 pandemic. Psychological interventions on anxiety, sleep disorder, and social support are suggested to alleviate mental health problems and stigma among this population. Additionally, this study discovered the precise cut-off value for CSS-S, which provides a valuable tool for screening perceived stigma among future COVID-19 patients and can be used to identify the patients in noosed of tailored interventions. --- 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 the Ethics Review Board of the Institute of Pathogen Biology, Chinese Academy of Medical Sciences , and the Research Ethics Committee of the hospital . The patients/participants provided their written informed consent to participate in this study. --- --- --- 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: Perceived stigma has greatly influenced the life quality of the COVID-19 patients who recovered and were discharged (RD hereafter). It is essential to understand COVID-19 stigma of RD and its related risk factors. The current study aims to identify the characteristics of perceived COVID-19 stigma in RD using latent profile analysis (LPA), to explore its psycho-social influencing factors, and to determine the cut-off point of the stigma scale using receiver operating characteristic (ROC) analysis.

INTRODUCTION Similar to gay men, studies have found that bisexual men experience violence, discrimination, and negative attitudes as a result of their sexual orientation . However, attitudes towards gay men and bisexual men may not necessarily be the same . Whereas the attitudes towards gay men have been researched extensively, little is known about specific social attitudes towards bisexual men. Only a few studies have empirically examined the prevalence of negative attitudes towards bisexual men and their influence on social evaluations . The aim of this study was to contribute to this literature. --- Social Stereotypes and Public Invisibility Several qualitative studies have claimed that bisexual men experience two unique social experiences: social stereotypes and public invisibility . Social stereotypes relate to specific biased evaluations that resonate in prejudicial behavior. These studies have reported that bisexual men face very specific social stereotypes; namely, that bisexual men are believed to be: confused and indecisive regarding their sexual and romantic preferences, untrustworthy, less inclined towards monogamous relationships, and therefore less likely to maintain long-term relationships and more likely to cheat on their partners, sexually promiscuous and open minded and open to new experiences . These findings thus suggest that bisexual men may frequently encounter unfavorable evaluations stemming from these stereotypes. The second unique experience reported by these studies is public invisibility; i.e., that bisexual men have little political and public visibility . Hence the typical experiences of bisexual men remain relatively unknown to the general public, including professional psychologists and sexologists . Public invisibility deserves attention, as several authors have connected it to higher rates of various health issues . In terms of social attitudes, it is unclear how public invisibility can co-exist with common social stereotypes. In other words, if people have no knowledge of bisexual men, how can they have specific biased beliefs about them? Two mutually exclusive explanations seem plausible. On the one hand, stereotype-based prejudice towards bisexual men may be an uncommon social phenomenon. This can be dubbed the "Weak Stereotypes Account". For example, Eliason conducted a survey among heterosexual college students and found that although a general negative attitude was prevalent, most participants could not report whether or not they thought bisexuals fit the stereotypes stated above. Alternatively, stereotype-based prejudice towards bisexual men may be common and still co-exist with public invisibility if it is not dependent on explicit a-priori knowledge about bisexual men. A similar idea was examined by Goff, Eberhardt, Williams, and Jackson , who studied the effects of implicit stereotypes on social evaluation. They found that participants had an implicit, but not an explicit, association between Blacks and apes which affected evaluations, thus indicating that stereotypical evaluations were possible even without relevant knowledge. The main goal of the current study was to assess the strength of these accounts by empirically examining two aspects of the social stereotypes of bisexual men. Study 1 examined factors that influence knowledge of stereotypes , and Study 2empirically examined the inclination to evaluate bisexual individuals stereotypically . --- Stereotype Knowledge Stereotype knowledge is the awareness that certain traits are stereotypically associated with a specific group. In the presence of a member or symbol of the group, stereotype knowledge should immediately become available . Stereotype knowledge is considered a necessary component in the implementation of stereotypes in social evaluations and behaviors . That is, it is claimed that if a certain individual does not have knowledge concerning a specific stereotype, he or she cannot implement the stereotype and behave prejudicially. Therefore, it is important to establish a clearer understanding of knowledge regarding the stereotypes of bisexual men. The first goal of Study 1 was to provide a descriptive measure of stereotype knowledge regarding the stereotypes of bisexual men. We compared these descriptive results to those of other studies, which explored stereotype knowledge of other social groups. Stereotype knowledge is thought to be constructed on the basis of well-learned associations in a particular cultural context. Devine reasoned that individuals who share a social context should have an equal amount of stereotype knowledge of a specific group. Several studies have found that low-prejudiced and high-prejudiced individuals should show an equal amount of stereotype knowledge. . By contrast, other studies have shown that individuals' personal beliefs are correlated with the extent of their stereotype knowledge . Krueger argued that individuals perceive their beliefs as reflecting their broader social beliefs, and concluded that high-prejudiced individuals should therefore report more stereotype knowledge than lowprejudiced individuals, simply because they have more stereotypical beliefs. It should be noted that the social stereotypes examined in these studies pertained to well-known racial minorities in the participants' communities. For example, Australian participants were asked about Aborigines , whereas Dutch participants were asked about Moroccan and Surinamese people in The Netherlands . --- Stereotype Implementation Stereotype implementation includes any behavior guided by social stereotypes. One expression of stereotype implementation can be found in stereotyped evaluationthe judgment of an individual's traits on the basis of his or her social group. In the only experimental study to date on bisexuals, Spalding and Peplau used an explored the implementations of stereotypes in the evaluation of bisexuals. They presented heterosexual college students with descriptions of couples with different sexual orientations. Participants evaluated one of the partners on a variety of scales. They hypothesized that if participants had biased beliefs regarding bisexuals, they would evaluate the bisexual targets less favorably on a number of specific traits. The findings indicated the presence of stereotype implementation in participants' evaluations of the targets. First, compared to heterosexual targets, bisexual targets in a relationship with heterosexual partners were evaluated as more likely to cheat, transmit a sexually transmitted disease and sexually satisfy their partners. Second, compared to lesbians or gay targets, bisexual targets in a relationship with lesbian or gay partners were evaluated as more likely to transmit STDs, and less likely to sexually satisfy their partners. Furthermore, Spalding and Peplau found that the evaluation of the bisexual targets depended on the target's current partner: bisexual targets in a relationship with heterosexual partners were seen as more likely to cheat and sexually satisfy their partners compared to bisexual targets in a relationship with lesbian or gay partners. However, contrary to Spalding and Peplau's hypothesis, bisexual targets were not evaluated as less trustworthy than the other targets. This result suggests that the content of the bisexual stereotype includes specific sexual conduct, but not untrustworthiness although this stereotype was cited in subjective accounts of bisexuals and postulated by others . Furthermore, Spalding and Peplau's study did not cover the evaluation of stereotypical traits such as confusion and indecisiveness regarding sexual orientation, lack of inclination towards monogamy, inability to maintain long term relationships, and openness to new experiences. Thus, as was the case for Eliason , the results found by Spalding and Peplau can only support a weak and incomplete account of the stereotypes of bisexual men. However, as Spalding and Peplau noted, the study had one important limitation. The characters in the study were individuals in functioning long-term relationships . The word "relationship" in itself is associated with notions of trust, stability, and commitment . Therefore, an individual in a functioning relationship is probably automatically evaluated as trustworthy and committed, at least in the context of the specific relationship. It may be useful to examine another context, such as an initial engagement with a potential partner, wherein specific traits are not inferred in advance . Study 2 explored the implementation of the stereotype of bisexual men, as measured by the evaluation of specific personal traits, in the context of a first date. Evaluation of bisexual targets was compared to that of heterosexual and gay targets. If the Weak Stereotypes Account is correct, we would expect to find: little or no evidence of specific stereotypical evaluations, or negative evaluations regardless of stereotypical content. On the other hand, if people do have specific stereotypical beliefs regarding bisexual men, we would expect to replicate Spalding and Peplau's results and find additional stereotypical evaluations of bisexual men as compared to heterosexual or gay men. Specifically, in accordance with previous studies , we hypothesized that bisexual men would be evaluated as less inclined to monogamy and less able to maintain a long term relationship than non-bisexuals. Also, in accordance with the literature on the stereotypes of bisexuals , we hypothesized that bisexuals would be evaluated as more confused, untrustworthy, and open to new experiences than non-bisexuals. Moreover, in order to rule out biased evaluations due to a general negative attitude towards bisexual men, we also explored whether any differences would be found in other traits as represented by scales of the Big Five Factor Inventory . We hypothesized that differences in evaluation between bisexual and non-bisexual targets should appear for stereotypical items, but not for non-stereotypical items. Study 1 and Study 2 were designed separately to investigate two closely related concepts: "stereotype knowledge" and "stereotype implementation." Though markedly different in method, these two studies are complementary. The combined findings should provide a clearer understanding of how stereotype-based prejudice and public invisibility can co-exist, and point to a novel way of thinking about these stereotypes. Psychological Research on the Net website, an international website for publishing links to online academic studies. Participants volunteered for a 1 in 30 chance to win $25 in gift certificates. The study was limited to heterosexuals who reside in English speaking countries . Participants who reported a different sexual orientation or a different place of residence were omitted from the study, and their results were not analyzed. 31.7% of the participants who agreed to participate dropped out before the end of the study, but such high drop-out rates are common in internet studies . The final sample included 88 participants: 58 women and 30 men. Participant age ranged from 17 to 63 years with an average age of 31.47 . 65 were residents of the U.S., 18 were residents of the U.K., and 5 were residents of Canada. 51 participants were college or university students. --- Procedure Participants entered the online questionnaire, gave their voluntary consent and read written instructions. The instructions stated that the purpose of the study was to better understand social attitudes and social stereotypes toward bisexual men. Further instructions were identical to Devine's original instructions; that is, participants were asked to list, in free-form, the content of the cultural beliefs and stereotypes of bisexual men, to the best of their knowledge. They were also told that the researchers were not interested in the participants' personal opinions, but in the views they thought people held regarding bisexual men. In other words, the measure called for the retrieval of all known stereotypical associations, without tapping the participants' personal views. The coding procedure was based on the stereotype knowledge literature . First, categories were created by the researchers in accordance with previous studies . Second, two independent judges, blind to the participants' answers to the Attitudes Regarding Bisexuality Scale-Male version , were presented with the categories and asked to code in individual responses. The judges were not obligated to use the categories and were allowed to add additional categories if they felt they were appropriate for more than 5% of the participants. For example, the category "Feminine" was added, since more than 5% of the responses included items such as "girly", and an additional category was added for participants who indicated they did not know of any social stereotype regarding bisexual men . Finally, if a response did not match any of the categories, the judges were allowed to code a response under three dummy categories: a "Negative-Miscellaneous" category , a "Positive-Miscellaneous" category , and a "Neutral Descriptions" category . These categories were not analyzed, as they provided no information about the participants' stereotype knowledge. Overall, 9 categories and 3 dummy categories were created . Stereotype knowledge was indicated by a person's indication of a specific category. Thus, for each participant, multiple responses referring to the same category were counted once. The judges agreed on 94% of the responses coded into the regular categories. After initial coding, disagreements were resolved through discussion. --- Measures --- ARBS-M After finishing the free-form task, participants completed the ARBS-M. This questionnaire contains 12 items measuring two dimensions of attitudes regarding bisexuality: Tolerance-the degree to which male bisexuality is viewed as a tolerable, moral sexual orientation and Stability-the extent to which bisexuality is viewed as a legitimate, stable sexual orientation . Responses on the ARBS-M were obtained on a 5-point Likert-type scale with anchor points of 1 and 5 . The original alpha coefficients for the scales were .83-.92 for the Tolerance scale and .83-.90 for the Stability scale . For this study, the coefficients were .92 for Tolerance and .91 for Stability. --- Additional Information After completing the ARSBS-M, participants were asked to indicate their acquaintance with bisexual individuals, on a 5-point Likert-type scale with anchor points of "I'm not acquainted with bisexual individuals at all" and "I'm well acquainted with bisexual individuals." Finally, participants reported their gender, age, educational status, and sexual orientation. --- RESULTS The average score on the ARBS-M was 3.21 for the Stability scale and 3.53 for the Tolerance scale . The scales were strongly correlated, r = .67, p < .001. Furthermore, a Spearman correlation test showed that the participants' acquaintance with bisexual individuals was significantly correlated with the Tolerance scale, rs = .22, p = .03, but not with the stability scale, rs = .08, p = .45. Since the ARBS-M was not constructed with normative values, the participants were assigned to either a relative lowprejudice group or a relative high-prejudice group , according to the median point of the answers on the Tolerance scale of the ARBS-M. These groups did not differ in distribution of gender, χ 2 < 1, age, t = 1.60, or educational status, χ 2 = 2.50. The frequencies of reporting each category were calculated. The overall frequency of specific stereotype knowledge ranged from 6.1% to 34.1% . Furthermore, 20.5% of the participants reported not knowing any stereotypes, and an additional 5% of the participants were only able to report descriptive terms, which were classified in the dummy categories. Next, the frequencies of each category were calculated separately for each group. For example, 41.3% of the low-prejudiced participants reported that bisexual men were stereotypically believed to be promiscuous whereas, this figure was only 16.7% among the high-prejudiced participants . To analyze differences in frequency of reporting stereotype knowledge, a series of chi-square tests were conducted for each of the individual categories 1 . Four comparisons stood out. First, the difference in frequency of reporting the category "Sexually Promiscuous/Unable to Commit" was found to be significant, χ 2 = 6.40, p = .01. Surprisingly, contrary to previous stereotype knowledge studies, the low-prejudice group reported this category with higher frequency. The same trend was observed for the category of "Closeted Homosexuals," χ 2 = 4.92, p = .03, and "Indecisive/Confused," χ 2 = 6.40, p = .01. The other comparisons did not reach significance, all ps > .05. Finally, the comparison between groups for the category "Don't Know Any Stereotype" was significant, χ 2 = 8.19, p = .004. Similar to the previous comparisons, the high-1 To avoid possible problems from dichotomizing continuous data we conducted a series of bivariate correlations between tolerance scores and stereotype knowledge parallel to the series of chi-squares. All reported results were replicated. Specifically, the likelihood of reporting the following stereotypes increased with a person's tolerance score: Sexually Promiscuous/Unable to Commit, rpb = .31, p = .003; Closeted Homosexuals, rpb = .28, p = .009; Indecisive/Confused, rpb = .25, p = .02; also, the likelihood of not knowing any stereotype decreased, rpb = -.46, p < .001. None of the other correlations were significant. prejudice group more often reported lacking any stereotype knowledge. Thus, the lowprejudice group generally showed more knowledge regarding the stereotype. In order to investigate this result further, we examined the correlation between tolerance and stereotype knowledge. To do so, we calculated a new continuous variable that represented stereotype knowledge by averaging the individual frequency for reporting only previously described stereotypical categories . For instance, a participant who reported all of these stereotypes received a score of 1, whereas a participant who reported only 3 categories received a score of 0.5. We performed a Pearson correlation between this variable and the subject's score on the Tolerance scale of the ARBS-M. This analysis also yielded a highly significant effect, r = .33, p = .002, which confirmed that low-prejudiced individuals had more stereotype knowledge regarding bisexual men than high-prejudiced individuals. In order to investigate possible gender differences, an additional set of chi square comparisons was conducted for the individual categories, but none of the comparisons was significant , all ps > .05 . --- DISCUSSION The first objective of Study 1 was to provide a descriptive measure of stereotype knowledge. The key feature of the data was the overall low response frequency for all categories, as well as the large percentage of participants who could not name a single stereotype of bisexual men. Previous studies that used this free-form paradigm regarding other social groups reported higher proportions of stereotype knowledge, ranging from .25 to .80 in Devine and from .11 to .87 in Lepore and Brown ; the current sample yielded much lower proportions, ranging from .04 to .41 . Furthermore, 20.5% of all the participants claimed they did not know of the stereotypes in question, whereas previous studies reported no such phenomenon. In previous studies, only Lepore and Brown reported of subjects having no stereotype knowledge whatsoever. Despite these marked differences in response rates, our ability to compare stereotype knowledge towards bisexual men and other groups is limited, unless stereotype knowledge for all groups is gauged with similar participants in a similar environment. For example, one might argue that, as our study was conducted online, rather than in a lab, low response rates could be attributed to the participants' low engagement and lack of commitment . However, results from recent studies make this possibility unlikely: even though participants in internet-based questionnaires show higher drop-out rates, these studies reveal no significant differences from questionnaires completed in a lab environment . The second objective of Study 1 was to examine whether there were any differences between stereotype knowledge of low-prejudiced individuals and that of high-prejudiced individuals. The results strongly indicate that the low-prejudice group had more knowledge of the stereotypes of bisexual men than the high-prejudice group. This result is incompatible with the reasoning that stereotype knowledge is essential for prejudicial attitudes , or that people who are more prejudiced have more stereotype knowledge . A possible interpretation of our results fits the Weak Stereotype Account. That is, the sample of high-prejudiced participants may have had general negative attitudes towards bisexual men , but not any specific stereotypical beliefs. However this cannot explain why high-prejudiced participants were also less inclined to believe that male bisexuality is a stable and viable sexual orientation , a notion derived from the stereotype that bisexual men are actually closeted gay men. A different explanation to our results is that people hold knowledge that they do not consider to be stereotypical. Note that in the free-form task, participants were asked "what are the social stereotypes and social beliefs concerning bisexual men?" This question might have tapped two unrelated factors: the knowledge of stereotypical categories regarding bisexual men and the acknowledgement that this knowledge is considered stereotypical . However, heterosexuals might simply have prejudicial beliefs that they think are true and not stereotypical at all. For example, one participant wrote: "I'm not familiar with any specific stereotypes of bisexual males. I do sometimes feel that they are actually homosexuals, but are afraid to identify as such do [sic] to social stigma." In this example, the participant reported having specific stereotype knowledge , despite being unaware that the notion is actually stereotypical. Following this reasoning, low-prejudiced individuals may simply be more explicitly aware of the stereotypical nature of common beliefs regarding bisexual men. Indeed, the low-prejudiced participants were more personally familiar with bisexual individuals . This reasoning points in a surprising direction; namely that lack of certain knowledge regarding a social group can actually encourage prejudicial behavior. Kunda and Spencer argue that the motivation to avoid prejudicial behavior rely on the desire to comply with egalitarian values. However, if people do not know that a certain belief regarding bisexual men is, in fact, stereotypical , they cannot know their behavior can be construed as prejudicial and should therefore have less motivation to suppress it. To summarize, in Study 1, the findings suggest that stereotypes of bisexual men are not well known. Nevertheless, knowledge of stereotypes was not a prerequisite for prejudice. Thus, a lack of explicit knowledge regarding bisexuals may co-exist with stereotypical evaluations of bisexual men. However, we did not measure stereotypical evaluation per se. In Study 2, we examined stereotypical evaluations directly in a contextualized evaluation task. Unlike surveys that measure attitudes towards a social group as a whole , the contextual paradigm in Study 2 probed attitudes towards hypothetical individuals . This paradigm has two main advantages. First, the evaluation is closer to an everyday, real life situation. Second, since the sexual orientation of the target is presented as a single detail among many, the purpose of the study was less obvious, and social desirability biases are reduced. These advantages make the evaluation task far more implicit than surveys and require less explicit conceptions of stereotypes. Thus, probing for evaluations of hypothetical individuals should clearly reveal even implicit biases. --- Study 2: Stereotype Implementation --- METHOD --- --- Procedure The experimental conditions were manipulated in a between-subjects design. Participants were assigned to one of the four experimental conditions describing the target and the non-target partners: bisexual man dating a woman, bisexual man dating a man, heterosexual man dating a woman, and gay man dating a man. The main interest was the encapsulated in two comparisons: evaluations of bisexuals versus non-bisexuals, and evaluations of a bisexual man dating a woman and a bisexual man dating a man. To allow for sufficient statistical power per comparison, participants were assigned to the different experimental conditions so as to create a ratio of approximately 1.5:1 in favor of the bisexual target conditions. We also sampled at least 20 men in each group to allow for a meaningful analysis of gender differences. The final sample for each condition was as follows: bisexual man dating a woman, n = 79 ; bisexual man dating a man, n = 63 ; heterosexual man dating a woman, n = 45 ; and a gay man dating a man, n = 47 . Participants entered the online questionnaire, gave voluntary consent and read a description of a non-target and a separate description of the target . Finally, participants were instructed to read all the descriptions and then evaluate the target on a list of items. To reduce knowledge of the purpose of the experiment, participants were told that all dates were selected at random from a larger pool and that they would receive descriptions of one to three dates, although all participants received only one date description. --- Materials --- Non-target description Participants read one of two possible descriptions of a single person: Suzanne, a heterosexual woman, or Scott, a gay man. The only additional information was that Suzanne/Scott is "a 24-year-old college student, who is dating for the purpose of finding a steady, long-term relationship." --- Target description Participants read a series of questions that Suzanne/Scott asked the target, James, on their date as a basis for their evaluation. The set of questions was written by the authors, specifically for this study, and included neutral questions such as: "What is your favorite color?" and short neutral answers such as "red" . The questions, with the exception of the question concerning the target's sexual orientation, were pre-tested to ensure that they indeed produced a neutral description of the target. It was important to show that the answer to the neutral questions would not produce a high or low evaluation on any of the items by themselves. We presented the set of questions and answers without indicating the target's sexual orientation, and responses were collected on the same scales used in the study. We conducted a series of t-tests and compared the results to the middle point on the scale . The pre-test showed that the set of questions and answers did not produce any significant evaluation, positive or negative, for any of the neutral items . The sexual orientation of the target was manipulated by the answer to the question: "Do you ever find yourself attracted to men?" if asked by the heterosexual woman non-target or "Do you ever find yourself attracted to women?" if asked by the gay man non-target. The answer of the bisexual targets was always, "I'm bisexual, so yes." The answer of the heterosexual and gay targets was "No, only women" or "No, only men," respectively. Note that the question was embedded among other questions, to further reduce the participant's awareness of the study goals. --- Illustrations In addition to the descriptions, participants were presented with illustrations depicting the couple on their date 2 . All illustrations were drawn based on stock photos of real models, and were constructed so that poses, eye level and distances between individuals would be the same for all couples. All individuals in the illustrations were portrayed smiling. The target character illustrations were pre-tested in a fashion similar to the textual description, to ensure that the illustrations did not produce a high or low evaluation for any of the items measuring the dependent variables. No significant effects were found for any of the items . --- Measures --- Non-stereotypical traits: Ten-Item Personality Inventory After reading the description of the target, participants rated the target on the Ten-Item Personality Inventory , a very brief measure of the Big-Five personality domains . Each item included a pair of descriptive characteristics that corresponded to a specific trait. Instructions were given to rate the target to the extent to which the participants agreed that the pair of characteristics applied to their evaluation of James, even if one characteristic applied more strongly than the other. All items were rated on a 7-point Likert scale with anchor points of strongly disagree and strongly agree . The TIPI includes the following paired items: "Extraverted, enthusiastic" and "Reserved, quiet" ; "Sympathetic, warm" and "Critical, quarrelsome" ; "Dependable, self-disciplined" and "Disorganized, careless" ; "Calm, emotionally stable" and "Anxious, easily upset" ; "Conventional, uncreative" and "Open to new experiences, complex" (subscales of Openness to New 2 The illustrations are available from the corresponding author upon request. Experiences). The last item was used to test the stereotype of open-mindedness. Although not as reliable as the longer versions of the Big-Five inventories, the TIPI has an acceptable test-retest reliability of .72 and was found to converge with widely used instruments . As the TIPI included only one item for each subscale, it was not possible to calculate internal consistency. However, in Gosling et al.'s original study, each pair of items was found to be significantly correlated, which indicates that both items measured the same overall trait. --- Stereotypical traits and overall match ratings In a similar fashion to the TIPI, participants then evaluated the target on an item termed "Indecisive, confused" designed to test the stereotype of confusion. Afterwards, participants were requested to answer a short set of questions, specifically designed to assess stereotypes of bisexual men, namely: trustworthiness , inclination to monogamy , and ability to maintain a long term relationship , and the overall match between the target and non-target characters . All these questions were evaluated on a 7-point Likert scale with anchor points of strongly disagree and strongly agree . --- RESULTS Prior to the main analysis, we calculated the correlations between the subsets of items that comprise the 5 traits of the TIPI. Four out of the five correlations were highly significant: "Extraverted, enthusiastic" and "Reserved, quiet", r = -.34, p < .001; "Dependable, selfdisciplined" and "Disorganized, careless", r = -.38, p < .001; "Calm, emotionally stable" and "Anxious, easily upset", r = -.39, p < .001; and "Open to new experiences, complex" and "Conventional, uncreative", r = -.32, p < .001. The correlation between "Sympathetic, warm" and "Critical, quarrelsome" did not reach significance, r = -.09, p = .08. Following this analysis, subscale pairs were averaged to form the traits of the TIPI. "Open to new experiences, complex" and "Conventional, uncreative", were not averaged to a single trait, as the former item was considered a stereotypical trait item, whereas the latter was considered a non-stereotypical trait item. Next, the reliability of the items pertaining to the measurement of the stereotypical evaluation was estimated. The reliability of the overall sample was rather low . However, separate analyses for each of the experimental groups revealed that reliability was somewhat better for the condition of a bisexual man dating a woman and much worse for a bisexual man dating a man, heterosexual man dating a woman, and gay man dating a man . It is important to note that each question measured a different aspect of the stereotype and high reliability scores were therefore not expected. The analysis was then divided into two main groups: 1) non-stereotypical traits ; and 2) stereotypical traits. A family-wise alpha of .05 was used for each of the two groups. Items representing stereotypical traits and non-stereotypical traits were analyzed using a Multivariate Analysis of Variance test with the evaluations of traits as dependent variables and experimental condition as the independent variable. Each significant analysis was followed by a series of separate Analysis of Variance tests for individual items . Finally, the main research questions were represented by two planned contrasts tests per item. The first contrasted the evaluation of bisexuals to that of non-bisexuals and the second contrasted the evaluation of bisexual man dating a woman to that of bisexual man dating a man. As each MANOVA included 5 items, a Bonferroni correction was used and alpha levels were reduced to .01 per test3 . In addition, the item measuring the overall match was analyzed separately, using a family-wise alpha of .05. Similar to the analysis detailed above, the item was analyzed using an ANOVA, followed by the two planned contrasts. The mean evaluation scores for each item and the results of all comparisons are summarized in table 3. --- Non-stereotypical traits Preliminary analysis revealed that the evaluators' gender did not yield any main effects or interactions in the evaluation of stereotypical traits. Therefore, the following analysis did not include it as an independent factor. The MANOVA for non-stereotypical traits did not yield a significant effect, F = 1.06, p = .39; Wilk's Λ = .931. Even though the MANOVA method is sufficient to show a lack of significant differences between the four experimental groups , it cannot show whether there were significant differences between the groups , as described in the pre-planned contrasts. Therefore, we conducted the planned contrasts between bisexuals and non-bisexuals and between a bisexual man dating a women and a bisexual man dating a man. However, none of these comparisons yielded significant results . --- Stereotypical traits As was the case for the evaluation of non-stereotypical traits, preliminary analysis did not show any influence of participants' gender on evaluations. Therefore, the following analysis did not include it as an independent factor. The MANOVA for stereotypical traits revealed a statistically significant difference in participants' evaluations of stereotypical traits based on the experimental condition, F was evaluated as less trustworthy, F = 11.53, p < .001 and less likely to maintain a long term relationship, F = 16.47, p < .001. A bisexual man dating a man , was evaluated as more likely to have had many previous romantic relationships, F = 6.75, p = .01. The other two planned comparisons did not yield a significant effect, both ps > .10. As an additional secondary analysis, all of the one-way ANOVAs and contrast analyses were conducted with the evaluation of non-stereotypical traits as covariates. All the significant results reported here remained significant after adding these control variables. --- Match between target and non-target characters An initial analysis revealed that gender was a contributing factor in the evaluation of the overall match between target and non-target. Therefore, evaluations of overall match were entered as the dependent variable to a 2 X 4 ANOVA. A main effect was found for gender: men gave overall higher match ratings than women, regardless of the target condition, F = 3.94, p = .048. The main effect of Experimental Condition and the interaction between the two factors were not significant . However, the planned comparison revealed a surprising significant effect: bisexual targets were evaluated as better matches than non-bisexuals, F = 3.89, p = .049. The second planned comparison did not yield a significant effect, F < 1. --- DISCUSSION The results of Study 2 go beyond previous studies and show that heterosexuals have specific prejudicial beliefs concerning bisexual men. These beliefs conform to the stereotype described in the literature; namely, that bisexual men are confused and indecisive, less inclined to monogamy, unable to maintain longterm relationships, untrustworthy and open to new experiences as compared to heterosexual or gay men. This result provides the most extensive empirical evidence to date that heterosexuals do in fact implement very specific social stereotypes of bisexual men, as described by subjective reports from bisexual individuals . We also found that participants did not differ in their evaluation of target characters on the non-stereotypical traits . Indeed, adding these items as control variables showed that they were unrelated to the evaluation of bisexuals on stereotypical traits. Taken together, these results cannot support the Weak Stereotype Account, a conclusion we return to in the General Discussion. Furthermore, the results indicated that heterosexuals took into account the current potential romantic partner of bisexual men in their evaluation: a bisexual man dating a gay man was evaluated as more likely to have many previous romantic relationships than a bisexual man dating a heterosexual woman. Also, a bisexual man dating a heterosexual woman was evaluated as less likely to maintain a long term relationship, and less trustworthy than a bisexual man dating a gay man. Moreover, as the higher reliability score indicated, while the evaluation of a bisexual man dating a heterosexual woman was rather consistent , the evaluation of a bisexual man dating a gay man was not. How can such results be explained? Before we proceed, the reader should recall that no hypotheses were made regarding these effects, and therefore these explanations are post-hoc and should be regarded as such. With that in mind, the pattern of these results seems to fit the Integrated Threat Theory . According to this theory, prejudice and stereotype implementation stem from perceived threats to in-group members. Why should bisexuality be perceived as a threat? As bisexual men are believed to be untrustworthy romantic partners, they are perceived to pose an emotional threat to women, and even a physical threat in the form of STDs. Also, because bisexuals can potentially have same-sex relationships and are believed to universally reject monogamous relationships , they are perceived to pose a symbolic threat to normative relationship-related values. In the current study, the stereotype of bisexual men may have been implemented in the evaluation of bisexual men dating a heterosexual woman simply because they seemed to pose a greater threat to in-groups members; namely, other heterosexuals. It is interesting to note that this explanation produces a testable prediction: the results should be reversed if the evaluators were lesbians and gay men. Otherwise, if the stereotype was not influenced by the evaluator's feeling of threat, no differences should be expected between the evaluations of heterosexual, lesbian and gay participants. One surprising result was the participants' evaluations of the match between the target and non-target. Despite the evaluations regarding relationship-relevant traits, bisexual men were regarded as better matches for the non-target character than the heterosexual or gay men. This result is confusing, mainly because it was initially stated that the non-target characters were looking for a steady long term relationship. It is important to note that this effect was produced mainly due to a relatively low estimate of the heterosexual target on the match item . That is, heterosexual evaluators, when evaluating the overall match, were more critical of the heterosexual target, but not of the other targets . It is likely that for this general item, a heterosexual target seemed more relevant to the lives of the heterosexual participants. Relevancy is one factor that is known to make evaluations more deliberate, less heuristic and consequently more critical . This explanation also resonates in the finding that men gave higher match evaluations than women. This account, although tentative, also produces a testable prediction: gay and lesbian participants should evaluate the match of gay targets less favorably than heterosexual targets. Future studies could sample gay and lesbian participants and further investigate this effect. --- General Discussion This study replicated earlier studies, and also produced a number of novel findings. In line with the idea that bisexual men are an "invisible" sexual minority, the findings showed that few people have explicit knowledge concerning the stereotypes of bisexual men. Furthermore, low-prejudiced individuals had more knowledge about the stereotype of bisexual men than high-prejudiced individuals. Nevertheless, the contextual evaluation task produced significant results, indicating that heterosexuals, when prompted, do in fact evaluate bisexual men stereotypically . Overall, these studies strongly support the notion that the implementation of specific stereotypes of bisexual men is prevalent, whereas stereotype knowledge regarding bisexual men is not. In this study, it was assumed that public invisibility and common social stereotypes can co-exist. If stereotype knowledge is a crucial aspect of stereotype implementation as has been claimed , the findings seem to suggest that heterosexuals do in fact have such knowledge. However, this knowledge might not be recognized as stereotypical or offensive. Where would such common knowledge stem from? As bisexual men have little public visibility, it is unlikely that heterosexuals can gather much knowledge from media representations or direct contact with bisexual individuals. Therefore, it is possible that heterosexuals draw their knowledge from indirect sources. One such source is common knowledge and beliefs regarding human sexuality in general. For example, the identity of a bisexual man might be questioned due to the more common knowledge that gay men often experience bisexuality as a transitory stage . Rust claimed that heterosexuality and homosexuality are commonly perceived as separate and inverted forms of sexual attractions. Similarly, as male and female are perceived as "opposite sexes", sexual attraction to males and females can be viewed as two separate and inverted attractions . If bisexuality is perceived as a mixture of two unique attractions , a likely deduction is that bisexuality inherently entails a persistent conflict. This can explain why bisexuals are perceived as never being satisfied with a single partner and therefore as unfaithful romantic partners. This account places the genesis of the stereotypes of bisexual men in a broader belief system regarding gender and sexual orientation rather than in direct contact with bisexuals or media representation of bisexuals per se. This perspective raises new and important questions regarding these stereotypes, and perhaps other stereotypes as well. There are a number of limitations to these studies which should be noted regarding the lack of relationship between stereotype knowledge and stereotype implementation. First, stereotype knowledge and stereotype implementation were assessed in two separate studies. Thus, the sample in Study 2 may have had more stereotype knowledge regarding bisexual men. However, assessing both factors in a single study would have entailed methodological problems, since one procedure would probably affect the results of the other. A study that can resolve this problem may reach more concrete conclusions regarding a possible connection between the two factors. Second, it is possible that participants in Study 2 only produced biased evaluations because they were probed for such evaluations . Note, however, that several precautionary measures were implemented to disguise the purposes of Study 2, and that the questions were not directed towards bisexual men in general, but towards an individual target . Also, direct probing regarding stereotypes should entail more cognitive control and thus reduce stereotypical evaluations . Hence the current results are likely to reflect a real bias in evaluation. Finally, in order to explore the involvement of general knowledge structures regarding sexuality in evaluation, future research should explore the possibility of a connection between compliance to sexual norms, and common beliefs regarding sexuality and the evaluation of bisexual men. Some support for this account was found in a recent study where priming participants with traditional gender roles increased negative attitudes towards bisexuals in participants that did not know bisexual individuals. The importance of further studies on this subject is underscored by the implications of the findings: if people do not acknowledge that their beliefs are stereotypical, they should have less motivation to suppress their stereotypical-related prejudicial behavior . In fact, they might not consider certain behaviors as prejudicial at all. This may contribute to accounting for unanswered questions, such as why bisexuals tend to disclose their identity less than lesbians or gay men and choose to endorse complex selective disclosure strategies , since even sympathetic individuals might behave prejudicially towards them. Enhancing scientific and social knowledge regarding bisexuality should improve understanding and acceptance of bisexuality as a valid sexual orientation, which should consequently reduce prejudice and social stress experienced by bisexual individuals.

We thank William Fisher and Dominique Lamy for their useful feedbacks on earlier versions of the article.

INTRODUCTION Residential segregation is often analyzed using aggregate data and descriptive measures, e.g., the Duncan Index or indices of exposure or isolation. However, trends of segregation at the level of cities or other spatial units result from micro-processes, namely from relocation decisions. This micro-foundation of segregation is taken for granted since Th. Schelling introduced his model of "micromotives and macrobehavior" , but there are only few studies analyzing these micro-processes empirically in a longitudinal perspective . Following the early work of Rossi , we assume that many residential relocations are adaptations to events of family formation and family extension during the life course. Such events trigger a considerable part of residential relocations. Neighborhoods become re-evaluated after family formation, now as contexts for children's socialization and development. We argue that the perceived quality of schools and neighborhoods is an important condition of choosing a particular destination. It is yet an open question whether relocations during family formation of migrant and native families are influenced by the same preferences and neighborhood evaluations. Following the literature on "ethnic colonies" and immigrant-native residential segregation, it can be concluded that immigrants differ from natives in their neighborhood preferences and relocation patterns . If relocations of migrant and native families do indeed occur mainly during family formation and family enhancement , and if relocations are at the same time outcomes of different preferences, the micro-dynamics of young families' adaptation of housing conditions might have a considerable impact on segregation. To find out whether natives and migrants differ in their residential preferences, we analyse in a first step whether migrant families value indicators of social embeddedness and neighborhood diversity higher than non-migrant families do. Our neighborhood indicators are subjective evaluations of proximity to relatives, perceived diversity, proximity to religious institutions and proximity to a desired primary school. Instead of primarily focusing on socio-economic factors, such as the mismatch of household income and housing prices, we test in a second step whether indicators of "ethnic social capital" do better explain residential moves of migrants during family formation than the proximity to the desired primary school. If the educational infrastructure in the neighborhood was more important than the local "ethnic social capital, " immigrants' relocation patterns would be more in line with "moving for the kids" then "moving for diversity." Hence, the aim of our paper is to obtain a better understanding of relocation decisions of native and migrant families, which leads to a better understanding of the underlying "micromotives" of residential segregation. In the empirical part of our paper, we use unique data from the "Moving for the Kids" project , in which more than 6,000 parents of 2nd and 3rd grade kids in elementary schools where interviewed about neighborhood perception and past relocations in a self-administered survey. To test whether evaluations of neighborhoods differ between immigrants and natives, we use an ordered probit-regression, which controls for self-selection into a respective neighborhoodcondition. Furthermore, we apply event history models to test whether the evaluation of neighborhood characteristics related to diversity and social embeddedness influence residential relocations, how relevant these factors are compared with the educational infrastructure and whether the effects differ between immigrant and non-immigrant families. The structure of the paper is as follows: In section Theory and Research on Ethnic Residential Preferences, Relocations, and Segregation, we will start with a short overview on theory and research on residential segregation with particular reference to residential preferences of migrants. In the third part, we will give a short introduction into our measurements and statistical methods. Empirical results will be presented in section Result, which is divided into two parts: First, we analyse whether migrant families evaluate indicators of social embeddedness and neighborhood diversity higher than non-migrant families. Second, we test whether migrant families tend to different relocation decisions, given their evaluations. In the fifth and last section, we will summarize our findings and draw a conclusion regarding our research questions. --- THEORY AND RESEARCH ON ETHNIC RESIDENTIAL PREFERENCES, RELOCATIONS AND SEGREGATION According to Th. Schellings theoretical model , the interdependence of preferences at the micro-level and inherent system dynamics at the macro-level tends to perfect segregation if changes in ethnic neighborhood composition trigger cascades of relocations to neighborhoods where households can realize their preference of not being in a small local minority. As an unintended result of micro-level behavior , the macro-level outcome of strong segregation can be regarded as a "perverse effect" , which means that the outcome is in sharp contrast to the rather inclusive "taste for diversity" in both groups. In contrast to Schelling's "taste-for-diversity" assumption, early Chicago School sociologists explained segregation patterns with socio-economic inequalities, processes and practices of exclusion, but also with own-ethnic preferences. Ethnic communities can provide social support especially for newly arriving immigrants. Often, members of the ethnic community already assist in planning the emigration . Since immigrants' "social capital" usually emerges at the local level , they tend to spatial concentration. In the long run, growing ethnic communities reduce incentives to invest into receiving-context cultural and social capital also in the 2nd generation . Concerns over potentially disintegrative effects of rigid ethniccultural boundaries crystallize in the terms "ethnic colonies" or "parallel societies" , which highlight the separation of ethnic communities from the majority population and describe a situation similar to "institutional completeness" . Ethnic groups do not only tend to spatial clustering and dense strong-tie networks within their own communities, but they also create their own ethnic institutions and organizations, namely businesses, schools, or even legal institutions for different religious groups . Regarding this mode of ethnic integration, P. Collier's distinction between emigrants and settlers challenges optimistic views about multiculturalism. While emigrants are willing to change group membership and adapt to norms of the receiving group , settlers bring their own institutions, norms and taken-for-granted knowledge and try to install their own "social model" in the acquired territory . Social network ties within "settler" communities create "bonding" instead of "bridging" social capital , and thereby reinforce ethnic boundaries . Within the liberal-democratic framework of most host countries, institutions of an ethnic colony develop a new self-understanding and perform many more functions than they would in the country of origin. For example, mosques are no longer just spiritual places, but become important places of self-help and socio-cultural exchange . Describing "ethnic colonies" in total as "parallel societies" is, according to Ceylan , inappropriate because of their different social segments with various including and excluding functions that satisfy the social, cultural and economic needs of the colony's inhabitants. Additionally, the taste for own culture and segregation assumed in the concept of "parallel societies' has been criticized for neglecting immigrants" disadvantages on housing markets, but also for normative reasons . If immigrants' residential choices were driven by preferences for own-group neighbors, they would likely end up in homogenous ethnic minority neighborhoods. Given that in Germany ethnically homogeneous residential areas are not common , in contrast to spatial patterns in the U.S., they might have at least some "taste for diversity" . Whether they are interested in own-ethnic cultural and social capital or not, they tend to live in ethnically diverse neighborhoods. The spatial clustering of ethnic minorities can emerge for various reasons-even without immigrants' preference for embeddedness into local own-group networks, e.g., due to stratified housing markets and discriminatory practices in the provision of housing . Inequality in the access to housing markets becomes also obvious by the fact that many immigrants cannot realize their preferences with respect to proximity to urban green spaces . Furthermore, ethnic residential segregation might be a result of "white flight" processes , which is a self-reinforcing outflow of better educated higher-status families from neighborhoods with a high concentration of ethnic minorities. This form of selective mobility might drain off resources and social capital from the local community and thereby increase neighborhood disadvantage . Increasing disadvantage leads to a decrease in housing prices, which cause a selective inflow of poorer people, who are often migrants. The "ethnic colony" hypothesis assumes that immigrants are particularly interested in local ethnic or religious social capital, and therefore prefer to live with their co-ethnics. Alternatively, immigrants could have such a preference, but are unable to realize their preferences by residential relocations if "competing" benefits outweigh the utility of ethnic local capital. For instance, immigrants might prefer to live close to co-ethnics and relatives, but also prefer neighborhoods with good primary schools for their children-which they would often find in other neighborhoods. Especially for long-term residents or 2nd or 3rd generation migrants, processes of structural assimilation might also lead to spatial assimilation and therefore cause a demand for better housing conditions . Accordingly, the longer migrants stay in Germany, the more likely they will adapt their housing needs to the native population. Therefore, we assume that migrants' family formation changes the evaluation of the neighborhood to more child-related aspects, e.g., proximity to a desired primary school. Since we know from research that migrant parents have high educational aspirations for their children , we expect that they, like native parents, also evaluate the spatial educational infrastructure when choosing a new dwelling. Qualitative studies observe selective relocations of Turkish middle class families to neighborhoods with a lower share of ethnic minorities. For this group, moving to a new dwelling is motivated by the parents' desire to realize access to highquality educational infrastructure. Concerns about the extent of school segregation and low achievement levels in adjacent schools motivate Turkish families to leave ethnic neighborhoods . Preferences for proximity to ethnic infrastructure, such as grocery stores or mosques, as well as to ethnic social networks, seem to be of secondary importance, especially for young parents . Even if integration into the ethnic community and access to ethnic infrastructure were important factors for migrant families, this preference does not necessarily require physical proximity to ethnic neighborhoods in times of modern transportation and communication technologies . "An ethnic grocer across town can easily be reached by bus once a week; friends or family members can be called every few days; and important community gatherings can be attended anywhere in the region on occasion" . Qualitative interviews conducted by Wiesemann with Turkish immigrants in Germany show that the ethnic character of a neighborhood plays an important role when choosing a location. However, ethnic preferences are in opposite directions: Whilst some households in his study preferred to live in areas with predominantly German natives, others chose to live in neighborhoods with large numbers of Turkish migrants, either because of the intra-ethnic contact opportunities or due to financial constraints. Taken together, these qualitative studies underscore that immigrants seem to have at least one important motive in common with natives, namely the preference for a "good" environment for their children, which is characterized by the absence of neighborhood disorder and the presence of high quality educational institutions. During family formation the evaluation of the neighborhood and the decision where to relocate might be similar compared with natives: in the end, it might be "moving for the kids, " rather than "moving for diversity." To test whether the local educational infrastructure is related to relocation behavior, we include the spatial proximity to a desired primary school in our analysis. Controlling for the evaluation of neighborhood characteristics, a strong effect of the absence of the desired primary school on residential moves would be an indicator of status attainment-motives. It is yet an empirical question whether patterns of residential relocations during family formation are either more in line with the "ethnic colony" hypothesis, or with spatial assimilation, motivated by better conditions for educational attainment for the children. --- DATA AND METHODS In our survey conducted in 2017 and 2018 in the federal states of Bremen, Lower Saxony, and North Rhine-Westphalia we asked mothers of children in 2nd and 3rd primary school grades about their residential biography, including their perceptions of neighborhood characteristics. Our window of observation begins with the date of moving into the dwelling where the female respondent became pregnant with the first child. Since respondents were required to recall neighborhood characteristics retrospectively, the survey instrument strongly benefitted from the idea of "cognitive anchors" . Even events that occurred rather distant in the past can be remembered if they relate to a significant other event, such as pregnancy or childbirth. To prevent distorted or erroneous memories, it is important that the interviewee has lived together with the child since birth, which is why we asked the biological mothers to complete the questionnaire. In 91.50% of cases this requirement was met, in 7.5% of cases the questionnaire was completed by the fathers, in 0.25 and 0.7% of cases by the stepmother/nursing mother and stepfather, respectively . The survey question for the first residential episode of interest was the following: "First of all, please think back to the time shortly before the pregnancy with the oldest child, i.e., with the 1st child in your household. Where did you live at that time? Please tell us the name of the town or city, the district or part of the town or city, and the time when you moved there." For each dwelling the respondents were then asked to provide information on year and month they moved in, on characteristics of the living environment and to rate these characteristics on a five-point scale. Figure 1 gives an example of how we measured the presence of specific neighborhood characteristics and their evaluation. Regarding the neighborhood characteristics, we included the perception of relatives and migrant families living nearby, the perception of having a house of worship of the respective religion nearby as well as the proximity to a desired primary school. Since we assume that both the evaluation of neighborhood characteristics and the relocation rate might depend on perceived neighborhood disorder, we built a scale of neighborhood disorder by using factor analysis based on tetrachoric correlations among five binary items, which we show in Table A2 in the Appendix. The higher the value, the higher is the level of perceived disorder. Detailed descriptive statistics for the independent variables are shown in Table A3 in the Appendix. Most studies lack information on the subjective assessment of neighborhood characteristics . Our procedure allows combining the information of whether a characteristic existed in the respective neighborhood or not with the respondent's evaluation of this characteristic from her point of view. By doing so, we measure the subjective assessment of the neighborhood characteristics. For instance, respondents can live either with or without many immigrant neighbors and can evaluate the situation as it is. We rescaled the "positive" vs. "negative" continuum by centering each scale on its mean value. Mean-centering the scale does not change its interpretation: the higher the value, the more positively a respective characteristic is evaluated . We distinguished two categories of migration background, namely "Turkish, Arabic and African" and "other migrants." If a respondent reported that she was not born in Germany, or does not communicate with the child predominantly in German language, or if she completed the questionnaire in Turkish or Arabic language, we assume a respective migration background. We are well aware that our classification is very simple and that the category "other migrants" suggests a homogeneity, which is in reality inexistent. By considering also the language predominantly spoken at home, however, we capture an important indicator of ethnic background, which is usually ignored by categorizations applied in official statistics 1 . Moreover, residential mobility is a rare event. Since the computation of the hazard rate results from the number of events divided by the "time-atrisk" for each response-pattern in the explanatory variables, applying a more fine-grained distinction of ethnic groups is not possible. For the analysis of the evaluation of neighborhood characteristics we use an ordered probit Heckman model . The ordinal outcome of this model is the evaluation of neighborhood characteristics which we estimate for the selective subsample of those respondents who have the respective characteristic nearby in their neighborhood. In other words, the Heckman model takes into account the respondent's selection into a particular neighborhood. This selection process precedes the evaluation of neighborhood characteristics. If the selection into certain neighborhoods is not taken into account, the positive evaluation of a given state would indicate the appreciation of either the presence or absence of a particular characteristic, which is uninformative. The Heckman model accounts for the selection by weighting the effects of explanatory variables x i on the positive evaluation by the process of selection into the respective state, which is influenced by the covariate vector z i . While the ordered probit model estimates the probability of a particular value v h on the ordinal scale, which is the probability that x j b + u 1j falls between the cutpoints defined by θ , the term s j in the binary probit selection equation equals 1 when the respective characteristic exists in the respondent's neighborhood, and zero otherwise. Pr = Pr s j = 1 The two error terms are assumed to have a bivariate normal distribution with zero mean, a variance of 1 and a covariance of ρ. Having analyzed the evaluation of neighborhood characteristics conditional on the selection process, we investigate in a subsequent step the effects of these evaluations on residential relocations. Residential relocations are events occurring after some waiting-time, beginning with moving into a dwelling and ending with the event of moving out. Since the information on the residential biography has been collected on a monthly basis, the resulting time-to-event data allows to predict the transition from the initial state "not moved" into the destination state "moved". We thus apply event history analysis. Here, the hazard rate r is the outcome of interest, which is the relation of the number of events f in the nominator and time at risk of those who have not yet experienced the event of relocation G in the denominator, formally r = f/G. At each point in time t, right censored observations without an event contribute to the denominator of the ratio of events to risk time . Since the event of interest is a residential relocation during the period of family formation, we will refer to "relocation rates." These relocation rates are the outcomes in the event history regression models. However, we are also interested in relocations triggered by particular motives, namely by improving the social context or for family reasons. Accounting for different motives of relocations is important in our study since we are interested in particular preferences of migrants and natives. We regard these different motives as competing risks. Accordingly, a relocation can be motivated either by improving the social context or by family reasons, and these two outcome-events "compete" against each other for occurrence. We defined relocations motivated by the "social context" by a set of items where respondents reported the reasons for a relocation. In order to identify the "improvement of social context" as a motive, we combined the following statements: the respondent wished to live "in a better social environment, " "with lower cultural diversity" and "nearby the desired school." In addition, we enhanced this measurement with information from an open-ended category where respondents reported their motives in their own words, e.g., saying that they lived in an "unsafe, " "bad" or "noisy" neighborhood or with many "non-German citizens." We defined the destination state "relocation, family" by reasons related to marriage and divorce and added information from open-ended questions on e.g., "parents, " "relatives, " and "family." In a competing risk analyses we get different coefficient vectors for each competing risk . Our observations are clustered in residential areas, namely in 585 different localities, that is, towns, cities, and villages. In order to account for the statistical non-independence of observations in these localities, we apply multilevel Weibull models of event history analysis, which enhances the standard Weibull model with a random intercept u j . The term u j is constant within the contexts, and varies between contexts. r = r 0 • exp In the Weibull model , the hazard rate r estimates hazard ratios relative to a baseline hazard r 0 . These hazard ratios might depend on unobserved characteristics of residential places , captured by the random effect u j . In the Heckman model and in the event history model we do not control for income, but for high education, unemployment and home ownership. It is hard to get reliable information on income in a self-administered survey. Moreover, respondents' cognitions and subjective perceptions correlate with education rather than with income, which is why we do not necessarily need the income variable. We also include the squared value of perceived disorder into our models in order to allow for non-linear affects, e.g., declining effects at higher values of disorder. --- RESULTS In the first part of our empirical study we test whether immigrants and natives evaluate particular neighborhood characteristics differently. Subsequently, we analyse the effects of these evaluations on residential relocations. Table 1 shows the two components of the Heckman model: the effects of the ordered probit model on the evaluation of neighborhood characteristics and the effects of the binary probit selection model . The upper part of the model does not control for many confounders because we assume that economic factors do more account for the location in a respective neighborhood, rather than for the cognitive process of evaluating its characteristics . Results show that migrants of Turkish, Arabic or African origin tend to evaluate proximity to a house of worship of their religion more positively than the reference group of nonmigrants. The same is true for the category "Migrant, other, " albeit the effect is smaller in magnitude. We find a similar pattern in the evaluation of living with many Non-Germans nearby. While the effect of "Turkish, Arabic or African origin" is significant and positive, it is significant only at the 10% level for other migrants. Regarding the evaluation of living with relatives nearby, FIGURE 2 | Average marginal effects of the immigrant categories, results from Table 1. we do not find any difference between non-immigrants and immigrant groups: on average, proximity to relatives is evaluated similarly in all three groups. High educational aspiration, i.e., the expectation that the child will graduate from high school with the Abitur2 , increases the positive evaluation of relatives nearby and desired school nearby. Since our sample is biased with respect to educational attainment , we cannot rule out that this effect also results from dual-earner families with high educational aspirations, where employed parents appreciate the proximity to e.g., their children's grandparents who regularly care for their children. Overall, the basic pattern of covariate effects on positively evaluating the desired school nearby is more or less similar to the pattern found for relatives nearby: again, there is no significant difference between non-immigrants and our two immigrant groups. Furthermore, while families living in residential property tend to appreciate a house of worship nearby, they tend to deprecate living with many Non-Germans in the neighborhood, but the latter effect is significant at the 10% level only. Similarly, living in residential property reduces the positive evaluation of having the desired school nearby. Although the effect is significant at the 10% level only, it is rather counterintuitive, since property has a robust positive effect of selection into such neighborhoods. In the selection part of the model, results show that respondents in both immigrant categories tend to live significantly more often in neighborhoods where they perceive many Non-Germans nearby, so these effects reflect microlevel manifestations of immigrant residential segregation. The same is true for the positive effect of unemployment in the household on living with many Non-Germans, which is in line with findings showing strong correlation of high shares of immigrants and socio-economic deprivation at the aggregate level of neighborhoods . This interpretation corresponds well with the very strong effect of perceived neighborhood disorder on the propensity to live with many Non-Germans nearby. Due to the simultaneous inclusion of its squared value , the effect is positive in particular at lower values of disorder, but dampens at higher values. Interestingly, there is a significant positive effect of respondents' higher education on living with many Non-Germans nearby, which possibly results from the fact that higher educated respondents tend more to live in urban areas, where the exposure to ethnic and cultural diversity in their neighborhoods is higher. Additionally, it is more unlikely for respondents with university degree to live nearby relatives. This is not surprising, since academics tend more to long-distance relocations for job reasons, which is often accompanied with a higher spatial distance to other family members. Figure 2 shows average marginal effects of the immigrant categories on the positive evaluation of living with many Non-Germans nearby and on living nearby a house of worship. The vertical line represents the non-immigrant reference group . For each category of the ordinal outcome variable, the error bar represents the AME of the respective immigrant category. Regarding living nearby a house of worship, migrants of Turkish, Arabic and African origin show a significantly reduced probability of categories 2 and 3 of the dependent variable, while the probabilities of categories 4 and 5 are significantly increased: they systematically tend to more positive evaluations of having a house of worship nearby. Similarly, the probability of category 1 for the evaluation of living with many Non-Germans nearby is reduced for migrants of Turkish, Arabic and African origin , also the probability of category 2 , whereas the probabilities of categories three and higher are increased . Regarding the group of "other migrants, " we also find a tendency toward a more positive evaluation of many Non-Germans in the neighborhood, but the effects differ less strongly from the reference group of native persons . Overall, first and later immigrant generations of Turkish, Arabic or African origin seems to have a positive attitude toward ethnic-religious cultural and social capital , and a preference for diversity rather than to spatial assimilation . To sum up, we do find differences in the evaluation of neighborhood characteristics between immigrants and natives. However, these differences are limited to the evaluation of having a house of worship nearby and living with many Non-Germans nearby, whereas we do not find systematic group differences between the evaluation of living with relatives nearby and having the desired school nearby. In the next step, we are interested in whether these differences in neighborhood evaluation influence residential relocations. In Table 2, the multilevel event history Model shows the overall patterns, whereas the second Model applies interaction terms to disentangle the effects of neighborhood evaluation between natives and immigrants. Both models show negative effects of age at family formation on the relocation rate . In line with common expectations, families who live in their residential property are much less mobile than families in the reference group . In addition, both models indicate that respondents of our two immigrant categories tend to lower relocation rates. In contrast, respondents with higher education and higher educational aspiration for their children are more inclined to relocate during family formation. Unsurprisingly, the effects of time-varying covariates "±6 months before and after giving birth of a subsequent child" and "± 3 months before and after starting a new job" affect relocation rates positive and are significant. In addition, the absence of the desired school in the neighborhood considerably increases the relocation rate. This finding provides clear evidence that the local educational infrastructure has an effect on relocation decisions. Both models include three effects of neighborhood evaluation: having a house of worship nearby, living with many Non-Germans, and proximity to relatives. In Model the main effect of a positive evaluation of proximity to relatives has a significantly negative effect on relocations, whereas the other two effects are insignificant. We find an interaction effect in Model in opposite direction to the negative main effect for immigrants of Turkish, Arabic and African origin. This is in contrast to our expectation: if this particular group were more interested in living close to their own ethnic relatives than natives are, this "bonding social capital" should have resulted in a negative interaction effect, so that these immigrants would have been even more immobile than natives when they appreciate proximity to relatives. Similarly, the interaction "eval. many Non-Germans X oth. Migrants" is positive, but significant just at the 10% level. In both cases it seems that the interaction effects result from relocations which are not in line with the assumption that immigrants were particularly interested in "bonding social capital" to their ethnic group or their family . Models in Table 3 estimate effects of neighborhood evaluation on relocation rates in a competing risk perspective. Models and show effects on relocations aiming at improving the social context, Models and estimate determinants of relocations for family reasons. Overall, respondents of both immigrant categories seem to be less mobile, which corroborates results from Model 1 in Table 2, but the hazard ratio is significantly < 1 only for other migrants aiming at improving the social context in Model . Similarly, a higher educational aspiration for the children increases the rate of relocations for improving the social context , but not for family reasons, whereas the intervening events of giving birth to a new child and changing the job tend to positive effects in all four models. This also holds for perceived neighborhood disorder, even though the effects seem to be stronger and more robust for improving the social context than for family reasons. A similar pattern results for the absence of a desired primary school: in all four models we find a significant and positive effect on the relocation rate, whereby this influence is clearly stronger for relocations aiming at improving the social context. Again, living in residential property has a consistently negative effect on all competing risks, whereas proximity to the workplace has a negative effect only on relocations motivated by relocations for improvement of the social context. Interestingly, the positive evaluation of many Non-Germans in the neighborhood points in the opposite direction for other migrants. While the effect is negative for non-immigrants , it is even positive for other migrants . In other words, even though other migrants appreciate the presence of many Non-Germans in the neighborhood, they show an increased tendency to relocate in order to change their neighborhood context. We find a similar pattern for relocations for family reasons, but the positive effect of "eval. many Non-Germans X oth. migrants" is only significant at the 10% level. At the same time, the insignificant main effect is close to 1, which means, there is no effect on the transition into the state "moved". The relocation rate decreases in all four models the more positive respondents evaluate the proximity to relatives. If they appreciate the presence of many Non-Germans, the relocation rate decreases when motivated by improvement of the social context, whereas the effect is insignificant with respect to relocations for family reasons. Accordingly, even though we found differences between immigrants and natives in the evaluation of neighborhood characteristics, we do not find corresponding relocation patterns. Although evaluations of neighborhood characteristics play a role for the decision to relocate, e.g., by a consistently negative effect of appreciating that relatives live nearby, effects of these evaluations on actual residential moves do not differ between immigrants and natives3 . Moreover, being aware of the strong effect of the absence of a desired primary school, respondents might evaluate their neighborhood during family formation primarily with respect to the socialization of their children. It is thus interesting to compare the strength of these two effects on relocations motivated by improving the social context: first, the effect of appreciating relatives nearby, secondly, the effect of not having the desired school in the neighborhood. Which of these effects is stronger and thus more relevant ? Figure 3 shows post-estimation results from the single level Weibull model of relocations aiming to improve the social context. Instead of focusing on significance, predicted survivor functions provide a clear insight into the relevance of the effects on interest. In Figure 3, the survivor functions G indicate for each month the share of episodes without an event of relocation . For the prediction we held all control variables constant at their mean values. Technically, the prediction results from the antilog of the negative accumulation of the hazard rate over time . G = exp   - t 0 rdτ   In the predicted scenario the evaluation of relatives living nearby is either positive or negative . The visualization in Figure 3 highlights that the effect of not having the desired school in the neighborhood is much stronger than the positive evaluation of living with relatives nearby . The thin black lines represent the situation when the desired school is nearby, the bold gray lines a scenario where the desired school is absent. When the desired school is nearby, after 120 months 91.5% still live in the current location when they appreciate the presence of relatives nearby, and 90.7% who do not appreciate relatives nearby. When the school is not nearby, in contrast, the overall share of stayers is considerably lower: 76.8% of those who appreciate the presence of relatives nearby still live in the neighborhood after 120 months, and 74.8% of those who do not appreciate relatives in their neighborhood. Accordingly, the effect of the desired school is very strong, whereas the effect of social capital provided by family networks on relocations is comparatively small. In summary, our results show that immigrants, particularly those of Turkish, Arabic or African origin, show a more positive evaluation of living nearby a house of worship of their religion and of living with many Non-Germans. Thus, regarding the evaluations there seems to be a preference for high diversity. However, during family formation these preferences do not transform into relocations conducted to realize these preferences: we found that preferences do indeed have an effect on actual rates of relocation, but the effects of these preferences do not differ between immigrants and natives in the way assumed according to the "ethnic colony" hypothesis. Following this hypothesis, a strong preference for ethnic or religious capital, such as houses of worship or ethnic and cultural diversity in the neighborhood, should decrease rates of residential relocations in particular for immigrants-which is empirically not the case. Surely, our results should be interpreted against the background of a considerable sampling bias toward respondents with higher education . Moreover, if residential segregation in combination with increasingly unequal housing markets were very strong, meaning that in general, relocations of immigrants across different types of neighborhoods rarely occur and immigrants mostly stay in highly diverse and often deprived neighborhoods before and after family formation, "ethnic colonies" would exist independently of the relocation dynamics we observe in our data. --- SUMMARY AND CONCLUSION Many studies describe patterns and trends of segregation at the aggregate level, although Th. Schelling's macro-micro-macro explanatory mechanism is based on individuals' or households' behavior. In our study, we followed Schelling's analytic shift toward the micro-level. First, we analyzed respondents' evaluations of particular neighborhood characteristics. Second, we analyzed the potential effects of these evaluations on actual residential relocations. Following to the classic work of P. H. Rossi, who identified family formation and family enhancement as crucial drivers of residential relocations between different neighborhoods , we were interested to find out whether different residential preferences in the phase of family formation and extension account for different relocation decisions between native and migrant families. In the theoretical part of our study, we discussed the emergence of "ethnic colonies" or "parallel societies, " which assume that migrants had a preference for living close to other members of their ethnic community. Contrary to this theoretical argument our results show that during family formation immigrants' residential relocations do not indicate that "bonding social capital" within the own ethnic community or other immigrant groups is a basic driver of these relocations. On the one hand, immigrants and natives tend to evaluate characteristics of their neighborhoods differently, as we have shown in the first part of our empirical analysis. We found that particularly respondents of Turkish, Arabic or African origin highly appreciate living nearby a house of worship and also living with many Non-Germans. On the other hand, our analysis of how these evaluations transform into residential relocations did not show any differences between immigrants and natives. Evaluations or preferences during family formation do not trigger relocations that result in "ethnic colonies" at the macro level. First and foremost, both migrants and nonmigrants seem to be sensitive to the educational infrastructure in their neighborhood. Aside from preferences toward ethnic "bonding social capital, " there are competing factors, for instance, whether the desired school exists in the neighborhood or not. From the immigrants' or the ethnic minorities' perspective, the issue of appreciating the educational infrastructure is related to investments into educational attainment and, into the process of intergenerational integration into the host society . As we could show by comparing effect sizes, absence of the desired school has a much stronger effect on relocations than the positive evaluation of proximity to relatives-a result which is indifferent toward immigrant origin. Relocations during family formation result from the same pattern of covariates in all three groups. According to an earlier study , residential segregation between immigrants and natives is also an outcome of different destinations where households relocate, that is, immigrants and natives are sensitive to neighborhood disorder and the absence of the desired school in the neighborhood and relocate, but immigrants end up again in neighborhoods where the situation is rather similar to the previous one. Although we achieve robust and clear effects, we should also address the limitations of our study, which primarily result from the field access. Even though great importance was attached to the simplicity and clarity of the survey instrument, written surveys are particularly susceptible to measurement error due to the uncontrollability of the survey situation. In addition, retrospective information is not free from measurement error even if our instrument applies cognitive anchors. Furthermore, despite knowing the individual place of residence, we didn't include any objective characteristics of the city or village, for example the population size. Even if we assume that subjective perceptions of the neighborhood are predominantly relevant for relocation decisions, we should keep in mind that these subjective perceptions are related to objective residential attributes. For example, the perception of neighborhood diversity or disorder is probably higher in urban areas compared with rural areas. In order to gain a better understanding of how subjective perceptions differ by regional contexts, objective information about the residential spaces should be included in the analysis. To sum up, the most important result of our study is that immigrants seem to evaluate neighborhood characteristics related to "ethnic colonies" differently from non-migrants, but they do not systematically translate these evaluations into specific relocation patterns. Educational infrastructure and proximity to relatives is important for migrants and non-migrants. Results also show that the absence of the desired school nearby has a much stronger effect on relocations than the preference of living close to relatives. Even though our micro-level analyses show clear differences between immigrants and natives in the evaluation of neighborhood characteristics, and also explain the overall process of relocation, they do not systematically explain patterns of residential segregation. Our results indicate that relocations during family formation do not entail "ethnic colonies" at the aggregate level. Nevertheless, there are considerable degrees of segregation at the macro-level, which is not just a result of socio-economic inequality between immigrants and natives . In addition, while the causes of moving out of a particular neighborhood do not overwhelmingly vary between immigrants and non-immigrants, recent results show that the quality of the destination seems to differ, whereby this quality is measured by indicators of neighborhood disorder . Future research should investigate in detail the micromechanisms of residential segregation in Germany, including the migrant and the native perspective, especially since residential segregation is related to processes of social integration. Same ethnic preferences and ethnic homophily with respect to social support and friendship choice are indicators of ethnic boundaries . If these boundaries contribute to the reproduction of group differences over time, they will also reproduce group-differences in language, norms, taken-for-granted-knowledge, and culture in general. Presumably, cultural differences between groups will correspond with differences in status attainment if culture is utilized as "capital" . If cultural capital matters for social mobility it will be rather unlikely that cultural diversity is unrelated to unequal chances of status attainment. In the end, differences in cultural capital can result in intergroup conflicts . In this regard, understanding ethnic residential segregation and segregation of social networks, as potential promoters for ethnic boundaries, will be important topics for future research. --- DATA AVAILABILITY STATEMENT The datasets generated for this study will not be made publicly available, due to data privacy legislation. The data collection involves schools and has been approved by the federal school authorities, conditional that the data is not accessible to 3rd persons. However, the data can be reanalyzed in our institute. --- ETHICS STATEMENT Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The patients/participants provided their written informed consent to participate in this study. --- --- SUPPLEMENTARY MATERIAL The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fsoc.2020. 538946/full#supplementary-material --- Conflict of Interest: 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.

Family formation is a crucial event in the life course and generates a major part of residential relocations. After family formation, neighborhoods become re-evaluated, now as contexts for children's development and socialization. We argue that the perceived or assumed quality of schools and neighborhoods is an important condition of choosing a destination. However, as the literature on "ethnic colonies" and immigrant-native residential segregation suggests, immigrants differ from natives in their neighborhood preferences and relocation patterns. If relocations of migrant and native families to particular destinations do indeed occur basically during family formation and family enhancement, and if they are at the same time outcomes of different preferences, the micro-dynamics of young families' adaptation of housing conditions might have a considerable impact on segregation. Results of our ordered Heckman probit and event history models show that on the one hand, immigrants and natives tend to different evaluations of characteristics in their neighborhoods. Particularly respondents of Turkish, Arabic or African origin highly appreciate living nearby a house of worship and also with many Non-Germans. On the other hand, our analysis of how these evaluations transform into residential relocations did not show any differences between immigrants and natives. Results thus suggest that evaluations or preferences during family formation do not trigger relocations which result in "ethnic colonies" at the macro level.

INTRODUCTION An individual's sexual orientation is related to the sense of sexual desire, whether with people of the opposite sex, the same sex or both. Gender identity, on the other hand, is understood as a sense of oneself as a man, woman or something outside these categories. Biological sex is marked by physiological issues and attributed to birth, classifying beings as male or female . The Federal Council of Medicine considers that transsexual women are those born with the male sex who identify themselves as women, i.e., the person is psychologically of one sex and anatomically of another. Transvestites are people who were born with one sex, identifying themselves and presenting phenotypically in the other gender, but accepting their genitalia . The categories trans and transvestite women permeate medicalpsychiatric discourses, but it is necessary to add philosophical, epistemological and social parameters, aiming at comprehensiveness of care . The visibility and understanding of these concepts, as well as their social acceptability, are still recent. Transgender people suffer the greatest prejudice and discrimination in the family and social environment and, by extension, in health services, either due to transphobia or discrimination linked to poverty, race/color, physical appearance or lack of specific health services. Consequently, being outside the heteronormative pattern still configures a risk situation in which rights violations are committed frequently and for various reasons . There is a scenario of slavery in which trans and transvestites are subject to different shades represented by the figure of the State as those that reproduce social norms of behavior considered appropriate or pseudo normal, striking them, in every way, through a heteronormative, excluding, totalitarian and authoritarian policy, making them feel abject beings of this restrictive thought, suffering from the moralistic, discriminatory and stigmatizing terrorism of an established and validated social policy that seeks to socially exclude them from public spaces . Therefore, transand transvestite women, as an escape from acts of violence and discrimination, avoid walking in public spaces, such as the street, the market, the bakery, the pharmacy, etc. Few attend schools or are inserted in the formal labor market and often stop looking for health services, living in a condition of invisibility . This evasion, especially from the health service, causes harm to trans people. Health professionals should provide support regarding emotional challenges related to gender identity, discussing clinical options for gender affirmation, as well as support to family members, who may also need health care . Studies indicate that trans people describe the Unified Health System as a system with little capacity to provide adequate care and assistance, cataloging discriminatory events and difficulties in relation to the use and respect for their social name. Even with policies that guarantee this right ,they continue to be pointed out as those that face more difficulties to access health services, from primary care to high complexity, among the entire population of lesbians, bisexuals, transvestites and transgender people . By directing our gaze to the specificity of the meaning attributed to the so-called trans or transvestite woman, we hope to contribute to ruptures and fissures in the reductionist, hygienist and eugenicist framework present in most health services. Heidegger's phenomenological perspective assists in this understanding, because research using phenomenology allows health professionals to make sense of their experiences and work activities, allowing them to return to attention and reflection on the reality and the way of being with others in the world of care, in the attempt to provide respectful, qualified care, with respect to universal right to health. --- OBJECTIVE To understand the meanings of being a transgender or transvestite woman in the care provided by SUS health professionals. --- METHODS --- Ethical aspects To ensure anonymity, participants were coded using the term PT, regardless of whether they declared themselves transvestites or transsexuals. In compliance with Resolution 466/2012 of the Brazilian National Health Council, the research was approved by the Research Ethics Committee of the Universidade Federal de Juiz de Fora. --- Theoretical-methodological framework The theoretical, methodological and philosophical foundation was the Heideggerian Phenomenology , with a view to approaching the understanding of the phenomenon of being a trans or transvestite woman in the care provided by SUS professionals, carried out in a municipality in Zona da Mata in 2018. --- Study design This is a qualitative study, in which we use the COnsolidated criteria for REporting Qualitative research to guide, sustain and guide the methodology of this study . --- Methodological procedures To reveal the meanings of being a trans or transvestite woman in the care provided by SUS health professionals, interviews were carried out with ten participants, of whom six declared themselves to be trans women, and four, transvestites. For inclusion in this study, the criteria used were residing in the municipality, self-declaring as a trans or transvestite woman, aged 18 years or older, regardless of color, religion or sexual orientation. Those who did not use at least one of the SUS health services available in the municipality were excluded. Before starting the research, interviews were conducted with two trans women, in order to adapt the interlocution script to the proposed objectives. This moment was configured as a methodological strategy that allowed assessing the research script applicability before contacting the participants defined for the study, aiming at the elaboration of the understanding . of The experience of trans or transvestite women in accessing public health services Oliveira GS, Salimena AMO, Penna LHG, Paraíso AF, Ramos CM, Alves MS, et al. --- Study setting The municipality belongs to southeastern Minas Gerais state, in the mesoregion of Zona da Mata, with an estimated population of 75,942 inhabitants and Human Development Index of 0.751 . It presents as institutions linked to sus a general hospital, 21 Basic Health Units with Family Health Strategy, a polyclinic thatprovides secondary care in angiology, cardiology, dermatology, physiotherapy, gastroenterology, gynecology, pulmonology, orthopedics, ophthalmology and urology. The other medical specialties that the municipality does not provide are referenced to municipalities such as Juiz de Fora and Muriaé, among others. The SUS municipal network also has clinical analysis laboratories, a municipal center for radiology services, a municipal vaccination center, a testing and counseling service, a reference center in oncology and a center for dental specialties. --- Data source For the selection of participants, the snowball method was chosen, as it is an applicable method when the object of study is composed of groups that are difficult to access or when the study seeks to access private matters. Through this method, chains of references are formed, built from people who share or know other people who have some characteristics that are of interest to the study. In this regard, the number of participants can increase with each interview. Regarding the inconveniences, it is worth mentioning the lack of control over the constitution and number of participants . --- Data collection and organization To obtain the reports with the ten participants included, a phenomenological interview was used, which is characterized by an existential approach, which aims to narrate the phenomenon by those who live the existential facticity to be unveiled. The interviewer's prejudices and judgments were put aside, and listening was carried out attentively, which made it possible to approach the person in front of him. Thus, it was possible to identify the meanings of the experiences presented by the speeches in the course of the narratives . To collect the information, an open interview was used, establishing an empathic relationship with participant, aiming to reveal the being's experience through the understanding of meanings. Each meeting lasted an average of 60 minutes, and was held in the environment chosen by the interviewees . The fact that the environment for the meeting was chosen allowed the trans and transvestites women interviewed to feel comfortable during the meeting. The following questions were used: how do you live your day as a trans or transvestite woman? When you have a health problem, where do you seek care? What do you think about the health care provided to you? Would you like to say anything else? The interviews were ended when the researcher did not come across any new elements and the addition of new information was no longer necessary, as it did not change the understanding of the phenomenon studied . During the collection of information, participants' speeches were recorded in an audio device . A field diary was also used, which allowed the recording of behaviors, body expression, expressed emotions, silences, pauses in speech, looks, tears, laughter and gestures, which represents phenomenal dimension . --- Data analysis The testimonies were analyzed from two methodical moments in Heidegger : Vague and Average Understanding, aiming at understanding the meaning that founds the existential analytics of being-in-the-world. Thus, it started with the transcription of participants' testimonies, putting prejudices in suspension, seeking to understand the phenomenon, that is, what was shown as the deponent's way of being. The hermeneutic analytical movement focused on understanding the meanings of being a trans or transvestite woman in the care provided by SUS health professionals. The analysis of the statements culminated in four units of meaning , namely: UM1 -The construction and maintenance of the feminine; UM2 -The importance of the social name: being recognized for who they are; UM3 -Prejudice present in health services; and UM4 -Counseling by laypeople and health professionals'unpreparedness: their influences on self-medication. Subsequently, starting from the UM and using the researcher's own intuition movement, we sought to understand the meaning of individuals' experiences, opening the horizon to unveil the facets of the investigated phenomenon. --- RESULTS Among the ten participants, participants' age ranged from 26 to 42 years. Regarding education, seven reported having attended high school, and three, elementary school. Among the participants, six self-declared to be white, three, black, and one, brown. As regards the situation on the labor market, only one declared herself unemployed. Three presented themselves as sex workers, two claimed to be self-employed, one as a decorator, two as hairdressers, and one as an office assistant. As for family income in minimum wages, a variation of one to five was found. Only one participant claimed to be married, and the others were single. The age to start female construction ranged from 13 to 28 years old. Regarding the presence of diseases and follow-up, one said problems with varicose veins, and another, with rhinitis, performing follow-up in the private sector, since the medical specialties for these conditions are not offered by the SUS network in the municipality.The two interviewees who declared themselves to be people living with the human immunodeficiency virus stated that they were followed up by the SUS specialized care service. One of them, with bradycardia, stated that she did not follow up for the health problem. Five denied health problems. The results showed that the meanings of being a trans or transvestite woman in the face of care provided by SUS health professionals permeate factors that interfere with their care. --- The construction and maintenance of the feminine It was evidenced that the mental construction of the feminine leads to the creation of the feminine being, being complete (or In their testimonies, regarding what it is like to 'be who they are' , the participants expressed that they think and act like women, according to social standards constructed and accepted, emphasizing acts referring to behavior, tone of voice and female submission, reproducing concepts that point out the female being as delicate, submissive. They seek to maintain feminine standards and go unnoticed in society as trans women. They relate hormonization as one of the markers for becoming women, bringing out the sensitivity. --- The importance of the social name: being recognized for who they are The social name emerges as a great achievement for trans people, leaving in the past the identification with the biological sex assigned at birth and presenting health professionals with the gender identity chosen by them, according to the following fragments: The social name is very important, because using it doesn't make me uncomfortable. Create an impact? That when we hand over the SUS card, the person thinks they will read another name than the one on there. So, I arrive, I have a peaceful conversation and say, "I want to be called by my social name, this has to be respected". I've had a problem with the name, so much so that today I managed to change the name on all my documents. I was very embarrassed. The physician kept calling the male name and I said, "I'm here!" and he kept calling, because he saw a woman there. It was embarrassing for me, for him and for the public. That's when I thought "enough, this is bothering me a lot!". I changed everything.. --- . as if a new person was born. --- The social name is everything to me! Is it hard to see that the X [says his given name] no longer exists? Now I'm someone else! My name is beautiful... you better not call me by my old name! The use and acceptance of social name by professionals were pointed out by participants as relevant factors for acceptance of being female, ensuring respect and avoiding embarrassing situations when using health services. The social name helps constructing the recognition of the female being and is pointed out as one of the best attitudes taken by them. Even with the legislation in force throughout the national territory, it was noticed in the testimonies that, on several occasions, using the form of treatment must be requested by the user. --- Prejudice present in health services It is noteworthy that, sometimes, individual prejudice goes beyond the limits of ethics and professional morals, materializing in prejudiced acts, including transphobia, according to the fragments: It was possible to identify that prejudice is part of trans women's daily life when using health services, pointing out that not only health professionals can act disrespectfully. I don't Situations of experience of veiled prejudice, through looks, supposition of diagnoses by other users and the prejudice experienced also emerged in this UM. --- Counseling by laypeople and health professionals' unpreparedness: their influences on self-medication The participating trans women highlight the importance of their social network to help in hormonal treatment prescription for constructing a true body. The previous experience of their peers proved to be indicative of successful treatment. The imminent desire for transformation leads to self-medication and can generate evasion of health services as a guarantee for achieving of Sometimes, self-medication is due to the fact that they feel ashamed to seek health services. Others point out that they are faced with health professionals' unpreparedness for hormonization and that, at times, they have experienced refusal of hormone prescription by medical professionals.They also reveal that they feel the need for health professionals to seek to expand their professional knowledge to meet this demand, highlighting the need for specialized professionals to provide care. --- DISCUSSION Starting from the interpretation of participants' statements, the meaning that founds the existential movement of being was sought, directing the researcher to the interpretative analysis, the understanding of the meaning of this veiled daily life, in a hermeneutic analysis supported by the work "Being and Time" by Martin Heidegger. Heidegger's hermeneutics is based on the understanding of meanings and interpretation of the meaning that presents human beings in their ways of being in the world, enabling the search for the essence of presence in their daily lives, which is, above all, the way of being of presence . Participants' way of being was revealed when they expressed aspects that support the construction and maintenance of the feminine in the face of socially imposed standards, the importance of social nameuse and acceptance by health professionals, their everyday experiences when using health services, when they testified about the importance of their social networks for the transformation process and about health professionals'unpreparedness to support this desire. Overcoming cisnormativity patterns and building their own gender identity goes through stages that reveal themselves as facticity, since they cannot predict, avoid or abstain from the reality of the desire to identify with the female gender or even to fall ill and need health care. In their statements, participants point out that they saw themselves with gender identity opposite to that designated at birth since childhood or already feeling feminine or not identifying with male groups. A study conducted with trans adult people revealed the awareness of their gender identification from an early age, coinciding with the period of their lives where, systematically, there is the division of children between boys and girls, whether in the school environment or in social life . In the daily life of transition to the female gender identity, it is revealed that the interviewees are historical beings, unfinished and in this temporality they are carried out in the present, however, they signify the past and have future desires and plans. In their statements, it was possible to identify that the construction process is constant and that, they do not know when it will end. We perceive different opinions when the subject is inherent to the conceptions of health/disease related to transformations experienced by the trans population. Still present, but not hegemonic, the biomedical model, guided by anatomical and pathophysiological patterns, tends to be guided by the medicalization process, pathologizing bodies and lived experiences. When we immerse ourselves in the experience of those who go through the transformation, we realize that the processes are different, in a search for health, seeking to remodel their body appearance as a potential for life, mediated by standards and ideals of beauty. Esthetic standards are of great importance in contemporary society. Just as cisgender people invest in esthetic procedures to achieve their personal desire, the trans women who build on the feminine also do so. As the models of beauty change in our society, the construction of the desirable body also becomes constant, always aiming to reach the current signs of beauty . As they are released into the world, they show themselves as female beings who present themselves according to standards imposed by society. According to the statements, the female being who is worthy of respect needs to present characteristics such as emotionality, sensitivity, education, not being a vulgar person, being dignified, without hair, with a thin voice, docile and with breasts. Prejudice contributes to consolidating health professionals' unpreparedness to address issues such as sexual diversity, given that they are not taught, during their training, to approach the issue in an open and prejudice-free way. As a result of this formation, health professionals are unprepared, providing barriers in the relationships of care production . When revealing their daily routine with the nursing team regarding the intramuscular hormone administration, one participant expressed that, for not presenting a medical prescription, the nurse is not able to perform the administration, which leads her to perform the procedure on her own, which can cause damage to her physical integrity. of The experience of trans or transvestite women in accessing public health services Oliveira GS, Salimena AMO, Penna LHG, Paraíso AF, Ramos CM, Alves MS, et al. The establishment of a health care space for trans and transvestite women is seen as a link for the accomplishment of citizenship of this population, not only guaranteed by access to the health sector, but also by the characteristics of the offer, its formalization, maintenance and institutionalization of care for them . It is in everyday life that human beings are protagonists of lived experiences, existing in facticity as being-in-the-world, we are always in the reference of a context, within everyday life, because, as an occupation, being-in-the-world is "taken by the world that is occupied", understanding the world as a set of references in which one is already inserted . Existing in a daily life that still segregates, imposes standards and behaviors of normality and acceptance, being-there trans people is led to inauthenticity, where they identify themselves as different from heteronormative standards and try to give a new meaning to the feminine. Participants expressed the importance of being-with, especially when pointing out their peers in occupation and pre-occupation when indicating the possibilities for building the desired body, and as they exposed there is professional unpreparedness to achieve this objective. Trans women experience disrespect for their gender identity during the reception process in Primary Health Care units, as well as experience health professionals' unpreparedness to meet certain demands in their care, which can lead to resistance from trans people in seeking health services for fear of discrimination . They revealed in their speeches that there are still health professionals who do not respect the social name, even though this treatment is guaranteed by Ordinance 1,820 of August 13, 2009 , which ensures the use of users' preferred name by all health professionals in public or private spheres . International studies indicate that transgender and genderdiverse individuals face barriers when trying to access healthcare, from discrimination to lack of access and experienced professionals, indicating the need for additional education for health professionals, especially those providing primary care, on how to provide care with knowledge, affirmation, and intersectionality. Veiled prejudice was also perceived during the analysis of the statements because participants revealed that they feel judged when searching for health services. This pre-trial is related to infectious diseases or health problems considered more serious. Eliminating transphobia in sexual health care can help improve access to diagnostic testing to reduce infection rates and support trans' and transvestite women's overall sexual health and well-being . Transsexuality and transvestility are undoubtedly identity experiences marked by intense conflict with social gender norms, psychological suffering and experiences of violence resulting from prejudice and discrimination. When prejudice and discrimination come from health professionals, from those who are expected to be welcomed, it is certainly even more harmful for them. We need to be aware of the various demands that trans women bring to us when they come to health services, without ever forgetting to see the other as a unique being full of possibilities, living in the facticity of their existence, which must be understood to be respected. --- Study limitations The present study has as a limitation the fact that it was carried out with transvestite and transsexual users of the public health system in a certain region, which may not represent other realities. We cannot forget that the results presented here cannot be generalized, since they reveal a singular experience that is unique to being-in-the-world. --- Contributions to nursing, health, and public policies The realization of this study points to the need to carry out further research approaching health professionals so that it is possible to perceive the difficulties experienced by them that may hinder the care of trans and transvestitewomen in health services, leaving no room for omissions, accommodations or alienation on our part, given the seriousness of the problems they encounter. Understanding these meanings reveals the need to implement actions that seek the formal and continuing education of health professionals who welcome trans and transvestitewomen in health services, as well as review the training of such professionals, offering an approach to the theme since their training. --- FINAL CONSIDERATIONS By understanding the meanings of being a trans or transvestite woman in the face of SUS care, it appears that these women, in the interaction with health professionals, face prejudiced attitudes arising from culturally and historically rooted social roles and gender stereotypes. Possibly, these attitudes are perpetuated as a result of social standards, the lack of knowledge of these professionals regarding the specific care of these people's needs, prejudices and lack of familiarity with public policies regarding this population group. Trans or transvestite women, faced with these difficulties, need to follow heteronormative standards when seeking care in health services, making their real needs invisible. of The experience of trans or transvestite women in accessing public health services Oliveira GS, Salimena AMO, Penna LHG, Paraíso AF, Ramos CM, Alves MS, et al. --- SUPPLEMENTARY MATERIAL Master's thesis entitled "O vivido de transexuais e travestis nos atendimentos à saúde: compreender para melhor assistir", available at: https://repositorio.ufjf.br/jspui/bitstream/ufjf/7809/1/ guilhermesachetooliveira.pdf.

Objective: to understand the meanings of being a trans or transvestite woman in the care provided by Unified Health System health professionals. Methods: qualitative research, guided by Heidegger's phenomenology, with 10 trans or transvestitewomen residing and using the Unified Health System in a municipality in Minas Gerais. Fieldwork was carried out by interviews. Results: trans or transvestitewomen reproduce the social patterns constructed and accepted by the female, with the search for hormonization being common, and, when it is difficult to obtain a prescription, they resort to self-medication. Social name use and acceptance by health professionals promote recognition. Trans or transvestitewomen experience prejudice on a daily basis, not only by professionals, but also because of the assumption of diagnoses by other users. Final considerations: transphobia promotes withdrawal from health services, due to fear, shame, knowledge about professionals' unpreparedness, triggering illness, social exclusion and violence.

the mid-year population, decreased for all age groupsbut at different rates. It decreased from 28.6 to 12.8 for age 65 and from 42.7 to 18.2 for age 70. Over the same period, the death rate decreased only slightly from 0.5 to 0.1 for age 10 and from 0.9 to 0.3 for age 15. Figures 1 and2 show the data. To find the impact of these death rate changes on the US age distribution, we run two counterfactuals. First, we hold the death rate fixed at the 1950 level for ages 60-79. Second, we hold the death rate fixed at the 1950 level for ages 0-19. In these counterfactuals, if holding the death rate fixed for a specific age group does not result in a significant change in the proportion of people 65 years of age and older, then we can conclude that the decline in mortality of the specific age group did not contribute to the increased average age of the US population. We use the United Nations World Population Prospects database from 2022 to run these counterfactual experiments. This database does not include the age distribution of migrants, so we run counterfactuals excluding net migration. However, we focus on the population 65+ for which net migration does not significantly impact the age distri- bution. The proportion of people 65+ in 2019 is 15.8% in data with net migration and 15.9% in our calculations without net migration. We calculate the yearly population in each age group using data from the previous year. For instance, we start with the age distribution and population in 1950. The number of people 0-years-old in 1951 is the number of births in 1950. The number of people 1-year-old in 1951 is the number of people 0-years-old in 1950 minus the number of deaths in that age group in 1950. We use the same process to calculate the number of people for those up to 100 years of age. Next, we run the first counterfactual, holding the death rate fixed at the 1950 level for ages 60-79. Each year, we use this death rate multiplied by the population to find the number of deaths, which we use to find the population for each age in the following year, as noted above. If death rates for older age groups had remained at 1950 levels, the proportion of people 65+ would be 3.2 percentage points less, changing from 15.9% of the population to 12.7%. Thus, part of the age distribution shift to an older population has been a result of the decrease in death rates for older people. We run a similar counterfactual, holding the death rate fixed at the 1950 level for ages 0-19, and find that the proportion of people 65+ is almost unchanged, from 15.9% to --- Age group

I n 1950, the proportion of people 65 years of age and over in the US was 8.2%. By 2019, that proportion had increased to 15.8%. In 2017, the US Census projected that by 2034 the population of people 65+ will be larger than the population of people under 18. 1 US Commerce Secretary Gina Raimondo commented that aging demographics were going to hit the country "like a ton of bricks." 2 Vespa (2018) predicts greater demand for healthcare, in-home caregiving, and assisted living facilities, as well as problems for social security. The age distribution of the population is affected by the birth rate, death rate, and net migration rate. If the inflow of immigrants exceeds the outflow, and if immigrants are on average younger than citizens, then the population would get younger. Similarly, higher birth rates imply that the population will get younger. Much of the discussion on US population aging has been on birth rates and immigration (Howard, 2019;Murray, 2021;and Williams, 2020). We focus on the effect of death rates on the increased average age of the US population. From 1950 to 2019, the death rate, calculated as the number of people per 1,000 who die each year divided by

Introduction Pre-exposure prophylaxis , a daily oral antiretroviral medication , is an effective biobehavioral tool for preventing HIV seroconversion following exposure [1,2]. PrEP marketing efforts have heavily focused on sexual minority men [3,4] and transgender women [3,5] as populations that remain disproportionately burdened by the HIV epidemic yet many within these communities continue to face a multitude of barriers that prevent PrEP access and utilization [6,7]. Thus, despite the effectiveness of PrEP for HIV prevention, only a small percentage of people who could benefit from the drug have been prescribed by a medical provider [8]. Recent surveillance suggests that PrEP adoption is rising among young SMM, albeit at slower rates among racial and ethnic minority communities [9]. This has been attributed to experiences of homophobia and racism in healthcare contexts, as well as to broader structural racism, socioeconomic inequality and homophobia [10][11][12][13]. Successful navigation of health systems and medical care for young SMM and transgender women, especially in the contexts of sexual health care, demand heightened attention not only to physical qualities of access , but also to social qualities that foster trusting relationships between providers and patients, such as non-judgmental interactions and shared medical decision-making [14][15][16]. Another considerable barrier to PrEP utilization is healthcare providers' lack of knowledge and awareness. With the exception of California, initiating PrEP requires a prescription from a physician in the United States, rendering clinicians as gatekeepers of the medication [17]. Studies have found that even providers who have moderate-to-high knowledge of PrEP have demonstrated limited integration of PrEP discussions into their HIV prevention communication strategies, suggesting that approaches to engage patients and serve as an effective HIV prevention resource are not being maximized, especially for patients who could benefit the most from PrEP [18][19][20][21][22]. In addition to a lack of knowledge, physicians have voiced concerns regarding the potential side effects and cost imposed on their patients [23,24]. The extant literature also suggests that many clinicians have concerns about patients engaging in risk compensation . These concerns may inhibit providers' willingness to prescribe PrEP, and may negatively impact SMM's interest in or perceptions of PrEP [25]. Researchers have observed inconsistent results when assessing the relationship between PrEP uptake and risk compensation; specifically, some studies with young SMM have provided evidence for risk compensation post-PrEP initiation [26,27], while others indicate that only a small subset of participants reported stopping or intending to stop using condoms should they or their partners initiate PrEP [28][29][30][31]. These inconsistent findings suggest that providers' conversations with patients are a critical opportunity to assess their patients' motivations for PrEP use, and to address combination HIV prevention approaches and risk reduction strategies. Researchers have begun to prioritize the potential role of medical mistrust imposed on PrEP rollout strategies [32][33][34]. In health studies with SMM, medical mistrust contributes to longer gaps between medical visits, decreased willingness to disclose sexual partner status to providers, and decreased acceptability of new medications [13,19,32,35,36]. Prior studies have also begun to elucidate medical mistrust's attenuating impact on engaging SMM and transgender women to seek and initiate PrEP. One qualitative study found that while both Black and White sexual minority men shared skepticism and concerns about PrEP's effectiveness and potential negative consequences, Black men's beliefs were more grounded in medical mistrust and experiences of homophobic and HIV-related stigma [13]. Conversely, a recent study on barriers and facilitators to PrEP use among transgender women found that Black women who reported lower educational attainment were less likely than their counterparts to endorse medical mistrust and more likely to have received PrEP education from healthcare providers [6]. While these studies provide important contributions to understanding the implications of medical mistrust and beliefs about PrEP among racial/ ethnic, sexual, and gender minority populations, to our knowledge, none have assessed how and to what extent such beliefs are associated with PrEP uptake. New York City's high HIV prevalence has solidified itself as a prioritized epicenter for informing implementation efforts that address the pillars of the Ending the HIV Epidemic initiative. To address these gaps and to expand PrEP research in a high-priority region, we sought to further elucidate the associations that healthcare experiences, PrEP-related beliefs, and medical mistrust have with current PrEP use in a cohort of young SMM and transgender women in New York City [37]. --- Methods --- Study Design The mixed-methods Health-Related Beliefs Study supplemented the P18 Cohort Study, a prospective cohort study which explored HIV, substance use, and mental health in a racially/ ethnically and socioeconomically diverse sample of young adult men and transgender women who have sex with men; parent study details have been published elsewhere [37][38][39]. From February 2018 to February 2019, participants enrolled in Phase 2 were invited to enroll in this sub-study during their biannual P18 study visit, by phone, or by email. Those with an HIV-negative serostatus, confirmed by HIV antibody testing during their P18 study visit, were eligible to participate in this study. Eligible participants completed a brief, computer-based survey which included questions about experiences with healthcare providers, thoughts about the healthcare system, and beliefs about sexual health-related issues such as HIV, other STIs, and PrEP. The sample of sub-study participants did not differ significantly from the Phase 2 parent study sample based on key sociodemographic factors including race, gender, sexual orientation, and self-rated health. All participants provided written informed consent for this study, and all activities were approved by the New York University Institutional Review Board. --- Measures Sociodemographic Variables-Race/ethnicity was measured with two separate items asking participants to identify their ethnicity and race . For analytic purposes, these were re-coded as a single variable, with categories collapsed into Hispanic/Latinx , Black non Hispanic, Other non-Hispanic, and White non-Hispanic. Gender was coded as "cisgender male" or "transgender woman/non-binary". Additionally, participants were asked about their preferred anal sex positioning, which was recoded as "top" for those who were prefer exclusively or primarily being the insertive partner, "bottom" for those who prefer exclusively or primarily being the receptive partner, and "vers" for those who indicated a versatile position preference or no strong preference for acting as the insertive or receptive partner. Health Care-Participants' current health insurance status was dichotomized as "yes" or "no". Those who indicated being insured were asked about the type of insurance, and these responses were collapsed to "parents' insurance", "employer insurance", "Medicare/ Medicaid", or "other". Participants were also asked to identify the place where they typically receive medical care-with options collapsed to "primary care provider", "public or community health clinic", "ER or urgent care", and "I do not receive medical care"-and whether it is difficult for them to pay their health care costs-with responses trichotomized to "somewhat or extremely difficult", "neither easy nor difficult", and "somewhat or extremely easy". Additionally, participants self-rated their health as "extremely good", "somewhat good", "neither good nor bad", "somewhat bad", or "extremely bad", which was trichotomized for further analysis. Medical Mistrust-Several measures were used to assess medical mistrust, "conspiracy beliefs", and associated constructs. All measures used a 5-point Likert scale with responses ranging from "strongly disagree" to "strongly agree" and, after items were re-coded as appropriate, were aggregated into a mean score with higher scores indicating higher levels of mistrust or greater endorsement of "conspiracy beliefs". Additionally, we calculated Cronbach's alpha for each measure to assess internal consistency of each measure within the sample. An HIV conspiracy beliefs scale, based in part on Bogart's 2005 HIV/AIDS Conspiracy Belief Scale [40], utilized nine of Bogart's original 14 items and included an additional 18 items that the research team developed to assess government and pharmaceutical industry-related conspiracy beliefs . The nine items were selected based on the items deemed most relevant to the population based on the extant literature . The additional 18 items were created based on recent previous studies that found that HIV-related conspiracy beliefs related to the role of the government and pharmaceutical industry were common among participants [94]. The 12-item Group-Based Medical Mistrust Scale [41], assessing medical mistrust based on minority group status, was modified to create two separate scales for sexual and gender minority group and racial/ethnic group . The Trust in Physician Scale [42], with the 11 items modified to replace "doctor" with "healthcare provider", was used to measure provider-specific mistrust . Due to the low Cronbach's alpha score, only descriptive data are presented, and the scale was not used in further analyses. A modified version of the Medical Mistrust Index [43], retaining 13 of 17 items, measured medical mistrust specific to healthcare organizations . Four items were removed from the MMI based on relevance to the purpose of the study . The Beliefs about Medicines Questionnaire-General [44] was used to assess general beliefs about medications, including beliefs about their overuse and harm, with two items added and one removed to create a 19-item measure . The two additional items included "People are prescribed too many medications these days" and "People shouldn't take a medication until they really need it". Pre-exposure Prophylaxis -This study took place prior to the approval of Descovy for PrEP in October 2019. As Truvada was the only option for PrEP during our study period, PrEP and Truvada were used interchangeably in our survey. Participants were asked if they had ever heard of PrEP/Truvada, and were able to respond yes or no. Those who responded yes were asked a series of questions about current and lifetime PrEP use, motivations for using or not using PrEP, sources of information about PrEP, anticipated PrEP stigma, and other PrEP beliefs related to medication and condom use, all of which were created for this study. Current PrEP use was a dichotomous yes/no question, and lifetime PrEP use was calculated by combining this with the item asking those not currently taking PrEP if they had ever previously done so. Those who were aware of PrEP were asked where they have received information about PrEP and could select all applicable response options including "healthcare provider", "friend", "family member", and "social media". Regardless of current PrEP use, all who were aware of PrEP were asked if they anticipated experiencing stigma from a number of individuals/groups and indicated anticipated beliefs that others might have if they knew the participant was taking PrEP. These items had a 5-item Likert scale response from "strongly disagree" to "strongly agree", which was dichotomized for analyses to focus on the "somewhat or strongly agree" responses. The anticipated beliefs based on a participant's hypothetical PrEP use were coded as having negative or positive connotations , and two additional variables were created to describe the number of items endorsed in each category. Additionally, participants were asked to indicate agreement on a 5-point Likert scale with six items on attitudes towards PrEP adoption, five items on how PrEP might affect their condom use, and one item on whether cost was a barrier to PrEP access; these variables were dichotomized for analyses to focus on the "somewhat or strongly agree" responses. --- Analytic Plan All analyses, unless otherwise noted, were conducted using RStudio v1.2.1335. First, descriptive analyses were conducted to calculate frequency distributions for sociodemographic sample characteristics, means and standard for medical mistrust and associated measures in the sample, and Cronbach's alphas for internal consistency of mistrust and associated measures. Next, bivariate analyses were conducted to assess associations of current PrEP use with key sociodemographic characteristics, medical mistrust, and PrEP beliefs, using Pearson's chi-square tests of independence, t-tests, and one-way ANOVAs, as appropriate. Finally, multivariable binary logistic regressions were constructed to model sociodemographic, psychosocial, and structural covariates that predict current PrEP use, including key variables identified in literature and through bivariate analyses . Adjusted regressions were stratified by race/ethnicity and were constructed using the stepAIC function to select the best fit model based on Akaike information criterion . --- Results --- Sample Characteristics The sample of n = 202 sexual minority young adults had an average age of 26.15 and predominately consisted of cisgender men . In terms of race/ethnicity, 15.3% were Hispanic/Latinx , 33.2% were Black non-Hispanic , 28.2% were White non-Hispanic , and 23.3% were other non-Hispanic . Additionally, near-equal proportion indicated that their preferred anal sex position was exclusively or primarily "top" versus exclusively or primarily "bottom" , with an additional 20.0% indicating that they were exclusively "vers" or had no preference. A majority of the sample was insured, with 12.9% uninsured and 2.0% unsure of their health insurance status . Of the 172 insured participants, 19.8% were on their parents' insurance , 36.0% were on an employer-sponsored plan , 34.9% were on Medicare or Medicaid , and 9.3% had another type of insurance . More than half primarily received healthcare from a primary care provider , followed by public or community health clinic and ER or urgent care ; 10.9% of the sample indicated that they do not receive medical care. When asked about difficulty affording healthcare costs, 29.9% reported it was somewhat or extremely difficult , 25.7% reported it was neither difficult nor easy , and 44.3% indicated it was somewhat or extremely easy . Most participants rated their health as somewhat or extremely good , with 7.9% self-rating as neither good nor bad and 3.5% self-rating as somewhat or extremely bad . --- Medical Mistrust and HIV Beliefs As presented in Table 2, participant responses indicated low to moderate levels of medical mistrust and endorsement of "conspiracy"-related beliefs. --- Pre-exposure Prophylaxis Uptake, Motivations, and Beliefs PrEP Awareness and Uptake-A majority of the sample had ever heard of PrEP/Truvada at the time of the assessment. Among PrEP-aware participants, 41.4% had ever taken PrEP, and 22.8% were currently taking PrEP. PrEP Knowledge Sources-Of the 198 participants who were aware of PrEP , the most common sources of information were healthcare providers , the Internet , and friends . PrEP Anticipated Stigma-One-third of the PrEP-aware participants somewhat or strongly agreed that there is stigma associated with PrEP use . Most participants did not indicate that they anticipated experiencing stigma from people in their lives if they took PrEP. The most common person or group from whom participants expected to experience stigma was family , followed by society at large and sexual partners . Participants were more likely to indicate anticipating positive beliefs from others if they knew the participant took PrEP, reporting an average of 4.74 positive items and 1.66 negative items. The most common anticipated PrEP beliefs were "I take care of my health" , "I am prepared" , and "I am safe" , with those significantly associated with differences in PrEP uptake displayed in Table 3. PrEP Beliefs-When asked if they think everyone should be on PrEP, the sample was fairly evenly split, with 28.3% saying yes , 38.9% saying no , and 32.8% saying they don't know . Most said they would be willing to take PrEP every day , and less than a quarter agreed somewhat or strongly that the cost of PrEP limited them from accessing PrEP . Less than a quarter agreed that they are taking PrEP because they prefer condomless sex and more than half agreed that they would acquire other STIs if their condom use were to decrease because of their PrEP use . Regarding "somewhat or strongly" agreeing with beliefs about PrEP adoption, 42.6% don't like taking pills when not sick , 34.7% have concerns about remembering to take a daily pill , 52.5% were interested in a long-acting injectable PrEP , and 50.5% prefer a long-acting injectable over a daily pill . --- Bivariate Associations with Current PrEP Use Current PrEP users were slightly older than those who reported not taking PrEP, t = -1.738, p = 0.078). There were no other significant differences in PrEP use based on sociodemographic characteristics, preferred anal sex position, health insurance status or type, health care utilization, or self-rated health . Higher scores on the HIV Conspiracy Beliefs Scale = 2.53, p = 0.012), Medical Mistrust Index = 1.84, p = 0.067), and Beliefs about Medicines Questionnaire [t = 2.86, p = 0.005] were associated with lower likelihood of current PrEP use . Current PrEP use was also more common among those who had received information about PrEP from their healthcare providers [χ 2 3). Those with negative beliefs around PrEP adoption, including those who reported they do not like taking pills if they are not sick [χ 2 = 11.48, p < 0.001] and those who reported concerns about remembering to take a daily pill [χ 2 = 10.63, p = 0.001], were less likely to report current PrEP use . --- Multivariable Models Predicting Current PrEP Use Unadjusted binary logistic regression models were run on significant variables identified in bivariate analyses, i.e., those with p < 0.10, and key variables from the relevant literature . Adjusted binary logistic regressions were run stepwise, using the stepAIC function in R's MASS package, to select the best fit model, for the overall sample, and were then stratified dichotomously by race/ethnicity for White participants and participants of color. In the adjusted regression model for the sample overall, there were significantly greater odds of current PrEP use among those who received PrEP information from a healthcare providers, those who felt they would contract other STIs if PrEP use decreased their condom use, and those with an interest in long-acting injectable PrEP; there was also significantly lower odds of current PrEP use among those who were unsure if everyone should be on PrEP, who reported that they do not like taking pills when not sick, and who had concerns about remembering to take a daily pill. In the stratified adjusted regression model for White participants, current PrEP use was significantly associated with age, such that older participants were more likely to be current PrEP users , and interest in long-acting injectable PrEP . Current PrEP use in White participants was less likely among those who do not like taking pills when not sick and those who reported concerns about remembering to take a daily pill . In the stratified adjusted regression model for participants of color, which included Hispanic/ Latinx, Black non-Hispanic, and other non-Hispanic participants, current PrEP use was associated with increased odds of believing there is stigma associated with taking PrEP , endorsing a greater number of positive beliefs about PrEP , receiving information about PrEP from a healthcare provider , and believing their STI risk would increase if PrEP uptake reduced their condom use . Additionally, current PrEP use among participants of color was less likely among those who scored higher on the Medical Mistrust Index , were unsure if everyone should be on PrEP , reported not liking taking pills when not sick , and were concerned about remembering to take a daily pill . --- Discussion and Implications for Clinical Practice Our study sought to better understand the contributions of PrEP attitudes, provider informational support on PrEP, PrEP-related stigma, and medical mistrust on PrEP behaviors among YSMM and transgender women. Notably, only 23.2% of the sample reported currently being on PrEP at the time of the study, despite all participants being PrEP-eligible. This finding is concerning for several reasons. PrEP was FDA approved in 2012, yet almost a decade later, this extremely effective biomedical prevention strategy is still widely underutilized [9]. The study presented here took place in NYC, where sexual health care is largely accessible due to a well-resourced public health infrastructure, effective public transit system, and significant programming focused on the needs of LGBTQ communities in the five boroughs. Given the extensive efforts to educate and provide PrEP in NYC, our study suggests that there are still formidable barriers to PrEP uptake, even in the most favorable conditions . In the context of our study in particular, virtually all participants were aware and even had at least some knowledge about PrEP, as they were recruited from a larger parent investigation on HIV prevention, mental health and substance use that provided some PrEP education throughout the duration of the longitudinal study. Yet, slightly less than a quarter of participants in the study presented here reported being on PrEP, suggesting that despite material access, YSMM and transgender women may continue to have concerns about PrEP. Broadly, our findings advocate for the development of culturally sensitive, LGBTQ health competent, and accessible strategies to assuage the concerns of YSMM and transgender women, two distinct communities under the LBGTQ umbrella. We found that, overall, participants were less likely to be currently taking PrEP if they endorsed negative attitudes concerning having to take medications when not sick and having to remember to take medications daily. These findings underscore a continued need to further cultivate PrEP buy-in among young SMM and transgender women, as the behavioral mechanics that drive PrEP's effectiveness remain a persistent barrier. Our findings align with prior studies that highlight concerns among SMM and transgender women around taking a pill every day, especially when not sick, as barriers to PrEP uptake [18,45,46]. A recent study on PrEP use among SMM found that individuals believed they could only manage taking a daily pill if they perceived their risk to be high, even those who preferred non-daily dosing methods had difficulty remembering doses [47]. Similarly, researchers have found that trans women have also expressed difficulty with taking pills, suggesting there is substantial interest in alternative modalities for PrEP consumption, specifically PrEP delivered via long-term injectable and other HIV prevention options [45,48]. Taken together, these various findings suggest that communication around preventative health specifically must be bolstered in order to help increase PrEP uptake, especially among those who are concerned with having to take a daily pill. Historically, the focus on young men's preventative health, especially sexual and reproductive health, has been largely absent [87,88]. For example, recent research has found that young sexual minority men are largely not aware that boys and men are also candidates for the HPV vaccine, representing missed opportunities to prevent HPV-related cancers [89]. Moreover, young men think of themselves as essentially healthy, and thus without a clear need for preventative healthcare [88,90], which our findings also indicate. Similarly, transgender women may not have been socialized to care for their health in the same way that many cisgender girls and women are from an early age , suggesting that preventative health care must also be emphasized for trans women in order to increase PrEP uptake. When stratified by race and ethnicity, however, White participants experienced different barriers and facilitators to PrEP uptake than did participants of color. While White participants' PrEP uptake was greater with increased age and less likely among those with negative medication-related beliefs about PrEP, PrEP uptake among participants of color was more informed by medical mistrust, PrEP stigma, sources of PrEP information, beliefs about PrEP risk compensation, and medication-related beliefs about PrEP. Our study suggests that for participants of color, but not for White participants, endorsement of risk compensation beliefs related to increased STI risk resulting from decreased condom use is associated with increased PrEP uptake. The literature is mixed on whether PrEP use is associated with an increase in STIs [49][50][51][52][53][54], but it is important to note that increased incidence of STIs following PrEP initiation could be due to more frequent and consistent screening, as STI testing is part of the recommended quarterly bloodwork for PrEP patients [55,56]. These findings suggest it is critical for public health to reframe PrEP-HIV risk messaging to emphasize PrEP as a tool for agency in preventative health care, rather than focusing on "high risk" categories and deficit-centric communication about disease prevention. Focusing on the "high risk" aspect serves only to reinforce PrEP stigmatizing attitudes that contribute to limited PrEP engagement [83][84][85]. Furthermore, attending to concerns about risk compensation [86] can be reframed positively, i.e., to position PrEP as a strategy to monitor one's STI status consistently. Health communication strategies in HIV prevention have capitalized on use of counter-narratives that attend to prioritized populations' negative beliefs about a sexual health promotion behavior, and positively frame messages to simultaneously minimize stigma toward the behavior while eliciting sexual health empowerment. For PrEP, it would be important to position PrEP as a means to increase feelings of protection and control over personal health. This may improve attitudes among those who attribute having to take medications to sickness or sick persons, and who have limited self-efficacy for daily PrEP adherence. Similarly, providers should also pay attention to patients' positive associations with PrEP, including HIV prevention preparedness, safety, and caring for one's health, as a majority of participants in this study endorsed such beliefs. Providers' continual reinforcement of these emergent positive associations with PrEP may help to increase uptake on a broader, community based level. This affirming, comprehensive, and prevention-oriented approach may help assuage concerns about increased STIs, lay the foundation for successful patient-provider relationships, and normalize sexual and gender minority health needs. In doing so, both clinicians and SMM may be more likely to consider PrEP a beneficial tool for HIV prevention. Additionally, roughly one-third of our sample reported never having received any PrEP information from a healthcare provider, supporting the argument that PrEP rollout efforts are not being maximized for HIV priority populations. Research continues to show that clinicians are largely unprepared to offer and prescribe PrEP due to a variety of reasons, including limited awareness and accurate knowledge [57,58], concerns about side effects and development of antiretroviral resistance [59], discomfort testing for HIV infection, discomfort prescribing PrEP, concerns about insurance coverage, and the time needed to provide risk-reduction counseling [57,60]. Moreover, previous research has shown that primary care providers perceive infectious disease physicians as being better suited to offer PrEP [23,61]. Yet, primary care providers prescribe birth control, which could be perceived as being more appropriately prescribed by a gynecologist, indicating it is possible to integrate PrEP into primary care clinical practice. Given PrEP's effectiveness in preventing HIV infection, implementation efforts must prioritize strategies to raise buy-in among providers about PrEP, including training on incorporating PrEP as part of their HIV prevention-related service provision and methods. Providers' capacity to successfully expand their HIV prevention services to include PrEP and scaling up PrEP-related provider communication requires having accurate PrEP knowledge and cultural sensitivity and LGBTQ health competence to engage young SMM and transgender women [18,74,75] and regularly screen patients' sexual histories. In prior studies, researchers have observed that many clinicians do not regularly take sexual histories [76,77] and that young SMM and transgender women often fear or anticipate stigma from their healthcare providers, thereby avoiding discussions about their sexual health [32,78]. This places the onus on clinicians to initiate non-judgmental conversations with patients about HIV prevention in order to determine if PrEP is a beneficial strategy. Therefore, providers should be trained on skills to facilitate an open, non-judgmental environment where their patients can discuss their gender and sexual identities, sexual behaviors, attitudes about PrEP, and PrEP-related concerns [79][80][81]. We also observed that YSMM and transgender women who reported receiving PrEP information from a healthcare provider were more likely to be currently taking PrEP compared to those who did not receive any information. These findings closely align with prior research that observed YSMM who reported HIV prevention discussions with a healthcare provider were more likely to increase sexual health promotion behaviors compared to those who reported no discussions [62]. Furthermore, this was a significant driver of current PrEP use for participants of color, but not for White participants, suggesting that PrEP education from healthcare providers may be particularly important to facilitating PrEP uptake for SMM and trans women of color, who face disproportionate burdens of HIV incidence and prevalence [5]. For providers treating trans women specifically, it is also important to be aware of the unique concerns this population may have, such as concerns related to hormone medications [63,73]. In this sense, providers must broadly be competent in many facets of LGBTQ health, and not only around SMM's needs. Ultimately, these findings support the significant role that healthcare providers have in promoting sexual risk reduction strategies, which is ever more critical in facilitating the adoption of PrEP in our sample given that in New York City, it can only be accessed upon seeing a health care provider and securing a prescription. While medical mistrust was not significantly associated with PrEP uptake in the overall sample, YSMM and trans women of color were less likely to report current PrEP use when they scored higher on the Medical Mistrust Index. While not a strongly significant result, this finding does align with several studies suggesting that mistrust may inhibit PrEP uptake, particularly among people of color [64,65]. These findings suggest that barriers to engaging in PrEP may differ for different racial/ethnic groups and across different levels of the PrEP continuum, and warrant further exploration. Providers should be aware of mistrust around PrEP, particularly for marginalized populations for whom this mistrust may be driven by historical and ongoing inequities [66], as having open and affirming conversations about these beliefs may help to assuage anxiety about biomedical interventions like PrEP. Previous research has also documented transphobia-related medical mistrust among trans women specifically [73], suggesting the need for tailored strategies to increase trans health competence. Addressing mistrust among both SMM and trans women may especially be important when long-acting injectable PrEP becomes available, as may very soon be the case [48,67,68]. Finally, PrEP-related stigma was not a significant predictor for White participants, but trended toward significance for participants of color. For the latter group, PrEP stigma was associated with increased odds of PrEP use, suggesting more research needs to be done that further disentangles the complicated relationship between stigma and PrEP use among distinct subgroups of SMM, particularly communities of color. Overall, the findings presented here echo previous calls for implementation strategies that are not "risk" based, as risk-based approaches exacerbate stigma, miss opportunities to incorporate sex-positive messaging, and ultimately hinder PrEP uptake [69,70]. --- Limitations Though our findings contribute important insight into PrEP behaviors among young SMM and transgender women, our study has several limitations. First, regarding receipt of PrEP information from a provider, we cannot ascertain whether this positive association is further distinguished by patient-solicited versus provider-initiated discussions. Specifically, participants who were actively engaged or motivated to seek PrEP may have been more inclined to have discussions about PrEP with their provider. Irrespective of this limitation, our findings underscore the tangible benefit of clinicians who are comfortable with discussing and prescribing PrEP. Through our study, we were unable to assess participants' PrEP indication status. Though this would assist our understanding of PrEP uptake among those who are PrEP-eligible, prior studies have suggested that PrEP indications are not time stable [28,71]. In other words, while all participants were PrEP-indicated at the baseline of the parent study, that may not have remained true at the time of data collection for our study; therefore, assessing beliefs about PrEP, PrEP stigma, PrEP resources, and medical mistrust remain salient for the current sample. This study is likely susceptible to selection bias. Participants were recruited from a parent study which regularly conducts HIV testing and counselling, including PrEP education, thereby potentiating a skew in the proportion of participants engaged along the PrEP continuum . The cross-sectional study also does not provide evidence for a causal relationship between the examined factors and PrEP uptake. Further longitudinal investigation would be necessary to explore whether factors such as healthcare provider recommendation or positive PrEP messages increase PrEP uptake. Additionally, there were analytic limitations due to the study's sample size. While this study included transgender and nonbinary participants, it did not enroll a sufficient number to draw meaningful conclusions or recommendations about gender minority people. The perspectives of gender minority people matter and should be given research priority that does not group them in with and conflate their experiences and needs with those of sexual minority men . There also was not a sufficient number of participants in certain racial/ethnic groups, including Asian and Native American individuals, to allow for assessment of their unique experiences and health disparities. Finally, this study took place in NYC, where participants generally have increased access to LGBTQ-competent healthcare providers and city-and -state funded resources to obtain PrEP low-cost or free in comparison to people in other parts of the US. The wide availability of resources specifically designed to serve the needs of sexual and gender minority populations may explain why we did not find racial, financial, or healthcare access-related disparities in PrEP uptake in our sample, as we had expected. Thus, while these findings may not be generalizable to parts of the country with less robust public health infrastructure, it may be that these concerns are amplified in lower resource settings. --- Future Research Directions Further study is needed to elucidate the role of PrEP in perceived risk and subsequent changes in sexual behavior [26,91]. It would be highly informative to assess the factors that promote, or the contexts that facilitate, the use of combination HIV prevention, especially among young SMM and transgender women who fear PrEP will lead to decreased condom use. Related, though provider informational support was associated with current PrEP use, we were unable to elucidate the types of information that were useful in facilitating uptake. Exploring the types of PrEP information that are most useful for young SMM and transgender women will raise the impact of PrEP communications efforts on these communities' PrEP behaviors. In particular, it is critical that more research looks specifically at the unique concerns and challenges facing trans women, as much of the literature includes both SMM and transgender women in the same sample. For example, a recent study found that trans women were concerned about an "explicit bias against transgender women who take PrEP", suggesting that more research is needed to examine trans women may experience stigma and other PrEP-related barriers differently than SMM [45]. Research to monitor changes in PrEP beliefs and identifying optimal strategies for service provision must account for diversity in perspectives among populations prioritized for HIV prevention efforts. The current study's findings were derived from a sub-study within the P18 Cohort Study in which enrollment criteria included being assigned male at birth. Across data collection waves of the P18 Cohort Study, some participants transitioned and reported their gender as female or non-binary. The perspectives of gender minority people matter and should be given research priority that does not group them in with and conflate their experiences and needs with those of sexual minority men. Further research is also needed to explore how young SMM may experience mistrust, as the literature suggests mistrust may be a barrier. --- Conclusion Monitoring PrEP perceptions and attitudes as the availability of new modalities emerge will facilitate a better understanding of community buy-in among YSMM and transgender women and will assist in scaling-up marketing strategies that make PrEP awareness, access, and adoption more equitable for those who seek to benefit from PrEP. The findings from our study further support HIV expert consensus that young SMM and transgender women underutilize preventative healthcare [92,93]. Although Truvada was approved as PrEP for HIV prevention in 2012, PrEP uptake continues to lag among communities who would benefit the most from widespread adoption. The onus is on public health and medical professionals to expand PrEP use among young SMM and transgender women, and to do so with approaches that recognize the unique needs of both SMM and trans communities. PrEP-related health communications provide an important platform to scale up PrEP implementation efforts; public health media and clinic-delivered communications around PrEP need to attend to the local salient barriers experienced by young SMM and transgender women and promote equitable access to, and support for, PrEP adoption. If appropriate and culturally-sensitive messages are implemented effectively, these health communication strategies possess potential to reduce the burden of the HIV epidemic that disproportionately affects young SMM and transgender women. --- Funding The Health-Related Beliefs Sub-Study was funded by NIMH through the HIV Research Education Institute for Diverse Scholars at Yale University . Dr. Jaiswal was supported by BST NIDA T32DA007233. The P18 Cohort Study was funded by the National Institute of Drug Abuse . Dr. Halkitis and Caleb LoSchiavo are supported by NJ ACTS . --- Appendix 1 HIV Conspiracy Beliefs --- 1. A lot of information about AIDS is being held back from the public. --- 2. HIV is a manmade virus. --- 3. There is a cure for AIDS, but it is being withheld from the poor. --- 4. The government is telling the truth about HIV. --- 5. HIV is a form of genocide against racial and ethnic minority people. --- 6. HIV was created by the government to control racial and ethnic minority people. --- 7. The medicine used to treat HIV causes people to get AIDS. --- 8. The medicine that doctors prescribe to treat HIV is poison. --- 9. People who take the new medicines for HIV are human guinea pigs for the government. --- 10. The government is deceiving people about the origins of HIV. --- 11. The pharmaceutical industry is deceiving people about the origins of HIV. --- 12. The government is deceiving people about treatment for HIV. --- 13. The pharmaceutical industry is deceiving people about treatment for HIV. --- 14. The government and pharmaceutical industry are working together to deceive people about treatment for HIV. --- 15. The government wants to keep people sick because they make a lot of money from ART. --- 16. The pharmaceutical industry wants to keep people sick because they make a lot of money from ART. --- 17. The cure for HIV is being withheld so more money can be made off of HIV medication. --- 18. Rich people have access to better HIV medication. --- 19. Rich people have access to the cure for HIV. --- 20. Pharmaceutical companies developed PrEP to make more profits. --- 21. HIV was developed by the government to target drug users. --- 22. HIV was developed by the government to target sexual and gender minority populations. --- 23. HIV was developed by the government to target Black/African American groups. --- 24. HIV was developed by the government to target Asian groups. --- 25. HIV was developed by the government to target Native American groups. --- 26. HIV was developed by the government to target Latino/a groups. --- 27. HIV was developed by the government to target White groups. --- Group-Based Medical Mistrust Scale -Racial/Ethnic Group --- 1. Doctors and health care workers sometimes hide information from my racial/ ethnic group. --- 2. Doctors have the best interests of people of my racial/ethnic group in mind --- 3. People of my racial/ethnic group should not confide in doctors and health care workers because it will be used against them. --- 4. People of my racial/ethnic group should be suspicious of information from doctors and health care workers. --- 5. People of my racial/ethnic group cannot trust doctors and health care workers. --- 6. People of my racial/ethnic group should be suspicious of modern medications. --- 7. Doctors and health care workers treat people of my racial/ethnic group like "guinea pigs." --- 8. People of my racial/ethnic group receive the same medical care from doctors and health care workers as people from other groups. --- 9. Doctors and health care workers do not take the medical complaints of people of my racial/ethnic group seriously. --- 10. People of my racial/ethnic group are treated the same as people of other groups by doctors and health care workers. --- 11. In most hospitals, people of different racial/ethnic groups receive the same kind of care. --- 12. I have personally been treated poorly or unfairly by doctors or health care workers because of my racial/ethnic group. --- Group-Based Medical Mistrust Scale -Sexual/Gender Minority --- Group --- 1. Doctors and health care workers sometimes hide information from my sexual/ gender minority group. --- 2. Doctors have the best interests of people of my sexual/gender minority group in mind --- 3. People of my sexual/gender minority group should not confide in doctors and health care workers because it will be used against them. --- 4. People of my sexual/gender minority group should be suspicious of information from doctors and health care workers. --- 5. People of my sexual/gender minority group cannot trust doctors and health care workers. --- 6. People of my v should be suspicious of modern medications. --- 7. Doctors and health care workers treat people of my sexual/gender minority group like "guinea pigs." --- 8. People of my sexual/gender minority group receive the same medical care from doctors and health care workers as people from other groups. --- 9. Doctors and health care workers do not take the medical complaints of people of my sexual/gender minority group seriously. --- 10. People of my sexual/gender minority group are treated the same as people of other groups by doctors and health care workers. --- 11. In most hospitals, people of different sexual/gender minority groups receive the same kind of care. 12. I have personally been treated poorly or unfairly by doctors or health care workers because of my sexual/gender minority group. --- Beliefs about Medicines Questionnaire --- 1. Newer medications are more effective than older ones. --- 2. Most medications are addictive. --- 3. People who take medications should stop their treatment for a while every now and again. --- 4. Medications only work if they are taken regularly. --- 5. Medications do more harm than good. --- 6. Medications are not natural remedies. --- 7. All medications are poisons. --- 8. It is better to do without medications. --- 9. Natural remedies are safer than medications. --- 10. Stronger medications are more dangerous than weaker medications. --- 11. Medications are a necessary evil. --- 12. Doctors place too much trust in medications. --- 13. If doctors had more time with patients, they would prescribe fewer medications --- 14. There is a big different between a medication and a drug. --- 15. The medication you get is more important than the doctor you see. --- 16. Doctors use too many medications. --- 17. Most medications are safe. --- 18. People are prescribed too many medications these days. --- 19. People shouldn't take a medication until they really need it. --- Trust in Physician Scale 1. I doubt that my doctor really cares about me as a person. --- 2. My doctor is usually considerate of my needs and puts them first --- 3. I trust my doctor so much that i always try to follow his/her advice --- 4. If my doctor tells me something is so, then it must be true. --- 5. I sometimes distrust my doctor's opinion and would like a second one. --- 6. I trust my doctor's judgment about my medical care. --- 7. I feel my doctor does not do everything he/she should for my medical care. --- 8. I trust my doctor to put my medical needs above all other considerations when treating my medical problems --- 9. My doctor is a real expert in taking care of medical problems like mine. --- 10. I trust my doctor to tell me if a mistake was made about my treatment. --- 11. I sometimes worry that my doctor may not keep the information we discuss totally private. --- Trust in Healthcare System --- 1. How much do you trust the healthcare system? --- 2. How willing are you to put your life in the hands of the healthcare system? --- 3. How confident are you in the healthcare system's ability to care for your health? --- 4. How much do you trust the healthcare system to give you the best possible care? Medical Mistrust Index --- 1. Patients have sometimes been deceived or misled by health care organizations. --- 2. When health care organizations make mistakes, they usually cover it up. --- 3. Health care organizations have sometimes done harmful experiments on patients without their knowledge. --- 4. Health care organizations don't always keep your information totally private. --- 5. Mistakes are common in health care organizations. --- 6. I trust that health care organizations will tell me if a mistake is made about my treatment. --- 7. The patient's medical needs come before other considerations at health care organizations. --- 8. Health care organizations are more concerned about making money than taking care of people. --- 9. Health care organizations put the patient's health first. --- 10. Patients should always follow the advice given to them at health care organizations. --- 11. I typically get a second opinion when I am told something about my health. --- 12. They know what they are doing at health care organizations. --- 13. I trust that health care organizations keep up with the latest medical information. Responses indicate those who reported "somewhat agree" or "strongly agree" for the statement, compared to those who indicated "somewhat disagree", "strongly disagree", or "neither agree nor disagree" --- --- ---

Pre-exposure prophylaxis (PrEP) is an effective form of HIV prevention, but young sexual minority men face myriad barriers to PrEP uptake. Participants (n = 202) completed a survey on healthcare experiences and beliefs about HIV and PrEP. While 98% of the sample knew about PrEP, only 23.2% reported currently taking PrEP. Participants were more likely to be taking PrEP if they received PrEP information from a healthcare provider and endorsed STI-related risk compensation. Conversely, PrEP uptake was less likely among those with concerns about medication use and adherence. While there were no racial/ethnic differences in PrEP uptake, there were differences in correlates of PrEP use for White participants and participants of color. To facilitate PrEP uptake, clinicians should provide PrEP education and screen all patients for PrEP candidacy. Additionally, public health messaging must reframe HIV "risk", highlight benefits of STI testing, and emphasize the importance of preventive healthcare for SMM.

Introduction And Background Delayed childbearing, a prominent trend in recent decades, reflects substantial shifts in societal norms, economic structures, and individual aspirations. Traditionally, childbearing occurred within a relatively narrow age range, driven primarily by cultural norms and biological considerations. However, the 21st century has witnessed a noticeable increase in the average age at which individuals embark on parenthood [1]. This phenomenon, commonly called delayed childbearing, has instigated a growing body of research to comprehend the multifaceted factors contributing to this trend [2,3]. Delayed childbearing, as a concept, emerged as a response to the evolving landscape of women's lives and aspirations. It refers to the trend of individuals, particularly women, choosing to postpone childbearing until later in life, often beyond their 20s and into their 30s or even 40s. This phenomenon challenges traditional notions of when parenthood should occur and is characterized by various factors influencing individuals' decisions [2]. As the world modernizes and societies become more diverse and interconnected, delayed childbearing is no longer limited to specific demographics or regions. It is a global trend that reflects the complex interplay between personal choices, cultural values, technological advancements, and economic considerations [3]. The decision to delay childbearing is a complex interplay of personal choices, sociocultural influences, technological advancements, and health-related considerations. As this trend progresses, it becomes imperative to delve deeper into the underlying factors driving this societal shift. Researchers, policymakers, and healthcare professionals can gain valuable insights into the broader implications for individuals, families, and societies by comprehensively examining the reasons behind delayed childbearing [4][5][6]. Understanding the factors influencing delayed childbearing is not solely an academic pursuit; it holds significant societal ramifications. These consequences extend from the dynamics of the labor market to the structure of family units, from formulating public policies to providing reproductive healthcare services. Furthermore, insights into delayed childbearing illuminate the evolving nature of gender roles, family dynamics, and individual aspirations, contributing to a more nuanced understanding of contemporary social dynamics [7]. This review article explores factors contributing to the growing phenomenon of delayed childbearing. In doing so, it will structure the discussion into distinct sections: a concise background, an elucidation of delayed childbearing, its global distribution, factors influencing delayed childbearing, long-term consequences, the novelty of this review, the rationale behind conducting this review, and, finally, the aims and objectives of this study. Notably, we will clarify the primary and secondary objectives in the closing part of this introduction to provide a more focused outline for the reader. By scrutinizing these factors from multiple perspectives, encompassing socioeconomic, cultural, technological, psychological, and policy dimensions, this article offers a comprehensive overview of the intricate web of influences shaping individuals' decisions regarding when to start a family. Our review encompasses a wide array of factors contributing to delayed childbearing, including the surge in women's educational and career aspirations, societal expectations surrounding motherhood, advancements in assisted reproductive technologies, evolving relationship dynamics, and the multifaceted health and medical considerations that influence fertility and pregnancy outcomes in later life. We aim to foster a deeper understanding of the motivations and challenges associated with delayed childbearing through an in-depth analysis of these factors. Additionally, we intend to highlight the policy implications stemming from this trend and suggest directions for future research to enrich our comprehension of this complex phenomenon further. --- Review Historical context --- Exploring Historical Trends in Childbearing Age Throughout history, childbearing age has been influenced by a combination of biological, cultural, and socioeconomic factors. In agrarian societies, where survival and productivity were closely tied to reproduction, individuals tended to marry and bear children at a younger age. However, a gradual childbearing age shift began as industrialization and urbanization reshaped societies [8]. The late 19th and early 20th centuries witnessed a transition as access to education and employment opportunities for women expanded. This shift, coupled with improvements in healthcare and contraception, contributed to a shift in societal norms surrounding childbearing. Women began postponing childbirth to pursue education, careers, and personal aspirations [9,10]. --- Societal Norms and Expectations in Different Eras Societal norms and expectations regarding childbearing have evolved across eras and cultures. In traditional societies, early marriage and parenthood were often emphasized, as they were essential for economic stability and the continuation of family lineage. These cultural expectations reinforced the notion that childbearing was a central role for women [11,12]. However, in the latter half of the 20th century, the feminist movement and changing perceptions of gender roles began challenging these norms. Women's increased participation in the workforce and the desire for greater autonomy shifted the balance between family and personal ambitions. These changes in societal values gradually opened the door to the concept of delayed childbearing [13][14][15]. --- Factors influencing delayed childbearing Economic stability and career aspirations: In the wake of evolving societal dynamics, the interplay between economic stability and career aspirations has emerged as a fundamental driver of delayed childbearing [14]. The increasing participation of women in higher education and the workforce has given rise to a paradigm shift in personal goals. With career aspirations becoming paramount, individuals postpone parenthood to establish a firm financial security foundation and attain their professional ambitions [15]. This strategic delay allows individuals to amass the resources necessary to provide for their families comfortably while avoiding the potential conflict between child-rearing demands and career advancement. The desire for economic stability intertwines with the pursuit of personal accomplishments, presenting a compelling incentive for deferring the initiation of parenthood [16]. Education and opportunities for women: The expanding landscape of educational opportunities for women has catalyzed a transformative narrative in the context of delayed childbearing. With increased access to higher education and a plethora of professional career pathways, women are empowered to make informed choices about the timing of parenthood. Higher levels of educational attainment equip women with the skills, knowledge, and self-confidence necessary to explore their academic and career potential [16] fully. Consequently, the decision to delay childbearing becomes calculated, enabling women to harness their educational achievements and chart well-informed trajectories for their professional growth. This educational empowerment encourages women to defer parenthood until they can optimally balance their aspirations and parenting responsibilities [17]. Impact of financial considerations: The decision to postpone childbearing is intrinsically entwined with financial considerations, significantly impacting reproductive choices. The escalating costs associated with raising children, from prenatal care to education, underscore the importance of financial preparedness. As the financial burden of parenthood becomes more pronounced, individuals often choose to delay having children until they feel sufficiently equipped to provide a stable and nurturing environment [17]. The imperative of a stable income and adequate savings in the face of these mounting expenses is a primary motivation to postpone parenthood. This pragmatic approach aligns with the broader societal trends of establishing a solid financial foundation before embarking on the transformative journey of raising a family [18]. Changing cultural perspectives on motherhood: Cultural perspectives on motherhood have evolved throughout history, reflecting the broader shifts in societal values and gender roles. Traditionally, motherhood was regarded as a central and defining aspect of a woman's identity, often overshadowing other potential roles or aspirations. However, contemporary viewpoints recognize that women's identities are multifaceted and extend beyond their roles as mothers. This paradigm shift acknowledges that women possess diverse talents, ambitions, and passions beyond parenting. This transformation in cultural perspectives empowers women to consider personal goals, career aspirations, and individual experiences before embarking on the journey of parenthood. Recognising the complexity of women's identities allows for greater agency in deciding when and if to become mothers [19]. Influence of family values and societal pressures: Family values and societal pressures significantly influence individuals' decisions regarding childbearing. In cultures where early marriage and parenthood are traditionally prioritized, there can be a considerable clash between these norms and an individual's aspirations. This tension between familial or societal expectations and an individual's desires can lead to decisions to postpone parenthood. Individuals may feel torn between fulfilling the expectations of their families and communities and pursuing personal goals such as education, career advancement, or personal growth. Negotiating this balance can be challenging, as individuals strive to honor their cultural heritage while embracing the evolving possibilities of delayed childbearing [20]. Role of media and pop culture: Media and pop culture have played a transformative role in shaping societal perceptions and norms, including childbearing. The portrayal of successful women in various roles and pursuits beyond motherhood has reshaped traditional ideas of what constitutes a fulfilling life. Media representations of women who prioritize careers, travel, creative endeavors, and personal growth have highlighted the potential richness of life experiences that can be pursued before parenthood. Moreover, the visibility of celebrities and public figures who delay childbearing for reasons such as education, career development, or personal fulfillment has normalized the concept. These public narratives help challenge the notion that parenthood should be women's primary and immediate goal, encouraging others to explore their potential before embracing the responsibilities of raising a family [21]. --- Technological advancements Assisted reproductive technologies : The landscape of ARTs has undergone transformative changes driven by innovations such as in vitro fertilization and egg freezing. These advancements have emerged as powerful tools, granting individuals unprecedented control over their reproductive timelines. ARTs can sometimes inadvertently contribute to a delay in childbearing. For instance, IVF combines eggs and sperm outside the body, enabling fertilization before transferring embryos to the uterus. While IVF has revolutionized family planning by helping couples overcome fertility challenges, it can also assure individuals that they can conceive at a later age. This confidence may lead some people to delay starting a family, believing they can always turn to IVF when ready [22]. Similarly, egg freezing, a pivotal aspect of ARTs, empowers women by allowing them to preserve their eggs at a younger age when fertility potential is higher. These preserved eggs can be thawed, fertilized, and implanted, offering women a means to extend their childbearing years and pursue life goals before embracing parenthood [22]. However, this option can inadvertently encourage women to postpone childbirth while focusing on career advancement or other life pursuits, assuming their fertility remains preserved. In practical terms, consider a scenario where a woman in her late twenties decides to freeze her eggs to ensure her fertility. With this safety net in place, she may delay starting a family and instead invest more time in her career or personal development, knowing she can rely on her preserved eggs later. While this is a valuable choice, it exemplifies how ARTs, like egg freezing, can influence the timing of childbearing decisions. Therefore, it's essential to recognize that while ARTs provide valuable reproductive options and flexibility, they can also contribute to delayed childbearing by offering individuals a sense of security and control over their reproductive choices [23]. Fertility preservation techniques: Fertility preservation techniques, exemplified by oocyte cryopreservation or egg freezing, have ushered in a new era of reproductive possibilities. This technique involves extracting a woman's eggs, freezing them, and storing them until a later time. This innovation is particularly significant for individuals facing medical treatments that may compromise their fertility, such as chemotherapy, or those navigating career trajectories that necessitate delaying childbearing. Egg freezing provides a biological safeguard and an empowerment tool, allowing women to align their reproductive decisions with their life circumstances. By offering the potential to utilize preserved eggs at a future point, fertility preservation techniques offer an avenue to mitigate the impact of time on fertility and support more informed family planning choices [23]. Impact on extending the biological clock: The synergy of technological advancements and medical interventions has notably extended the biological clock for women. While it is well-established that fertility diminishes with age due to a decline in egg quantity and quality, the advent of ARTs has offered a measure of counterbalance. By leveraging the advances in reproductive science, these technologies have allowed individuals to conceive beyond what nature's timeline might otherwise allow. This extension of the biological clock is flexible, as fertility naturally declines with age. However, technological strides have created an avenue for individuals to pursue parenthood later in life, blurring the lines of traditional agerelated constraints and offering new horizons for those who wish to optimize their reproductive potential [24]. --- Relationship dynamics Delayed marriages and implications for childbearing: In the landscape of modern relationships, the age at which individuals decide to enter into marriage has significantly shifted, subsequently influencing the timeline of parenthood. As societal norms adapt to changing opportunities and priorities, many individuals emphasize personal growth, educational pursuits, and the establishment of their careers before considering marriage. This trend, known as delayed marriage, inherently impacts the decision to start a family. As couples delay marriage to achieve personal and professional milestones, the subsequent delay in childbearing follows suit. This phenomenon holds implications for family planning, as the biological window for childbearing narrows with advancing age [25]. Evolution of partnership dynamics: The landscape of partnership dynamics has transformed remarkably, reflecting the diverse ways individuals form relationships and build families. Traditional models of marriage have expanded to include cohabitation, non-traditional partnerships, and diverse family structures. These shifts in partnership dynamics are pivotal in influencing decisions surrounding childbearing. In nontraditional relationships, considerations about parenthood often intertwine with personal aspirations, financial stability, and relationship dynamics. The evolution of partnership models underscores the need for tailored approaches to family planning and highlights the intricacies that individuals and couples navigate in deciding when to become parents [26,27]. Co-parenting in non-traditional family structures: Co-parenting has gained traction within non-traditional partnership arrangements, reshaping family planning strategies. Individuals in relationships that do not conform to the conventional norms of marriage may opt for co-parenting arrangements, where partners may not share a romantic relationship but collaborate to raise children. In these unique family structures, delayed childbearing takes on a distinct dimension. The flexibility of co-parenting arrangements allows individuals to carefully consider when and how they choose to embark on parenthood, aligning the decision with their personal and relational circumstances. This can lead to a deliberate delay in childbearing as individuals navigate the complexities of co-parenting dynamics [28,29]. --- Health and medical considerations Age-related fertility decline and associated risks: Biological factors profoundly affect the decision to delay childbearing. As women age, their reproductive capacity diminishes due to decreased quantity and quality of eggs. This decline in fertility is a natural consequence of the aging process, and women who opt to delay childbearing may face challenges when trying to conceive. The biological clock ticks faster as women move into their late 30s and 40s, making it more difficult to achieve pregnancy and increasing the likelihood of infertility. Moreover, the risks associated with pregnancy complications rise significantly with advancing maternal age. Older mothers are at an elevated risk of miscarriages, ectopic pregnancies, and conditions like placenta previa. Additionally, the risk of chromosomal abnormalities in the developing fetus, such as Down syndrome, substantially increases as maternal age rises [30,31]. Maternal and fetal health in older pregnancies: Delayed childbearing brings about medical considerations for mothers and their unborn children. Advanced maternal age, often associated with delayed childbearing, can have a pronounced impact on maternal health. Pregnant women in their late 30s and beyond are at an increased risk of developing gestational diabetes, which can lead to complications for both mother and child. Preeclampsia, characterized by high blood pressure and organ damage, is more common among older pregnant women and poses significant health risks. Notably, the mother's age also plays a role in fetal health. Older maternal age is correlated with a higher incidence of chromosomal abnormalities and congenital disabilities in newborns. Maternal and fetal health complexities in older pregnancies highlight the importance of informed decision-making and comprehensive prenatal care [32]. Advantages and challenges of medical interventions: Advancements in medical technology have provided hope for those who choose delayed childbearing. Prenatal screening and genetic testing offer valuable insights into the health of the developing fetus, helping parents make informed decisions. These interventions can identify chromosomal abnormalities and genetic disorders early in pregnancy, allowing individuals to prepare for potential challenges or consider alternatives if necessary. However, the advantages of medical interventions come hand in hand with ethical, emotional, and financial dilemmas. For instance, the decision to undergo prenatal testing may raise ethical concerns about selective abortion based on the results. Couples considering delayed childbearing must navigate the emotional complexities of these choices, and the costs associated with these medical interventions can pose a financial burden. Striking a balance between the benefits and challenges of medical interventions is a crucial aspect of the decision-making process for those contemplating delayed childbearing [33,34]. --- Psychological and emotional factors --- Ambivalence and Decision-Making Mixed feelings about parenthood: Delayed childbearing often ushers in a complex and profound ambivalence among individuals. While the aspiration to embrace parenthood is prevalent, this sentiment can be intertwined with a tapestry of conflicting emotions. Becoming a parent is laden with uncertainties and complexities, which can evoke various feelings. On the one hand, the desire to experience the joys of nurturing a family is present; on the other hand, concerns about how parenting might disrupt existing personal goals and lifestyles emerge. This emotional ambivalence introduces layers of intricacy into the decision-making process. Individuals grapple with weighing the fulfillment of their ambitions against the profound commitment and responsibilities of raising children. This ambivalence often postpones childbearing until a more opportune and balanced time in their lives [35]. Balancing personal goals with parenting desires: As individuals navigate the trajectory of delayed childbearing, the dynamic equilibrium between personal aspirations and the desire for parenthood takes center stage. Pursuing higher education, career advancement, and exploring the world through travel are among the diverse personal goals individuals seek to accomplish before embarking on parenthood. However, these ambitions can sometimes be at odds with the pull of parenthood. The endeavor to strike a harmonious balance between these seemingly divergent paths becomes an intricate negotiation. The decision of when to transition into parenthood requires carefully considering how individual aspirations coexist with the commitment to raising a family. This intricate negotiation exemplifies the delicate interplay between embracing one's individuality and fulfilling the role of a parent, ultimately shaping the timeline of delayed childbearing [36]. --- Psychological Well-Being Impact of societal expectations on mental health: Societal expectations significantly influence individuals' mental well-being regarding delayed childbearing. Prevailing norms and attitudes regarding the "ideal" age for parenthood can create pervasive pressure and anxiety for those who postpone having children. The fear of being judged by peers, family, or society and the perception of deviating from established norms can lead to feelings of inadequacy and self-doubt. This phenomenon can contribute to adverse mental health outcomes, including heightened stress levels, lowered self-esteem, and increased vulnerability to conditions like anxiety and depression. Recognizing and addressing these societal pressures is vital in fostering positive mental health for individuals opting for delayed childbearing. Challenging the stigma surrounding unconventional timelines and promoting an inclusive dialogue can alleviate the psychological toll of societal expectations [37]. Coping strategies for emotional challenges: Opting for delayed childbearing can introduce individuals to a unique set of emotional challenges. The pressure to conform to societal norms while simultaneously striving to fulfill personal goals can create an internal conflict that affects emotional well-being. Individuals often adopt coping strategies that provide emotional support and resilience to navigate these complexities. Seeking solace in close relationships with friends and family, sharing concerns, and seeking advice can provide comfort. Additionally, engaging with mental health professionals who specialize in addressing the emotional intricacies of life decisions can offer valuable tools and strategies to manage emotional stressors effectively. These coping mechanisms are central to empowering individuals to navigate the emotional terrain of delayed childbearing while maintaining their mental equilibrium [38]. Parenting self-efficacy and identity: Delayed childbearing can significantly impact how individuals perceive themselves as potential parents. With the accumulation of life experiences, skills, and self-confidence, individuals often develop a heightened sense of preparedness to embrace the challenges and responsibilities of parenthood. The journey toward delayed parenthood, marked by personal growth and self-discovery, can instill a mature and self-assured approach to parenting. As individuals chart their path to parenthood, their parenting identity evolves in distinctive ways. The confidence gained through life experiences and the capacity to navigate complex life decisions can foster a sense of competency and self-efficacy in anticipating the responsibilities of raising a family. This more informed and intentional perspective can lead to enriched parenting experiences, enabling individuals to approach parenthood with a deeper understanding of themselves and their caregiver roles [39]. --- Policy implications --- Work-Life Balance Policies Parental leave and flexible work arrangements: Work-life balance policies, encompassing parental and flexible work arrangements, emerge as pivotal components in facilitating delayed childbearing for individuals seeking to balance their careers with family aspirations. Paid parental leave is a cornerstone of these policies, offering new parents the crucial opportunity to devote time to caring for their newborns without jeopardizing their professional trajectories. This form of leave fosters stronger parent-child bonds and acknowledges the significance of family life within the context of a professional career [40]. Flexible work arrangements constitute an additional facet of these policies that have gained prominence. These arrangements, which encompass remote work options, adjusted schedules, and part-time positions, present an adaptable approach that caters to the evolving needs of working parents. By accommodating the challenges of juggling family responsibilities and career demands, flexible work arrangements pave the way for individuals to transition into parenthood more easily [41]. --- Supportive Workplace Environments A crucial facet of enabling delayed childbearing lies in creating supportive workplace environments that recognize the importance of family life. Organizations that invest in their employees' well-being and family needs exhibit a forward-thinking approach to modern workforce dynamics. By providing amenities like onsite childcare facilities and dedicated lactation rooms, employers foster an environment where the practical needs of working parents are met, alleviating the pressures associated with raising a family while managing professional responsibilities [42]. Furthermore, workplaces that offer parenting education resources contribute to employees' holistic wellbeing. Employers bolster the confidence and capabilities of individuals navigating the complexities of parenthood by facilitating access to guidance on parenting skills, child development, and work-life integration [43]. --- Influence on Career Trajectories The profound impact of work-life balance policies on career trajectories cannot be overstated, particularly in the context of delayed childbearing. For women, in particular, these policies play a vital role in counteracting the "motherhood penalty," a phenomenon where career advancement is hampered by the perception of diminished commitment due to parenting responsibilities. By implementing supportive policies that accommodate parenting leaves, flexible arrangements, and opportunities for continued professional development, organizations create an environment where individuals, regardless of gender, can pursue their career goals while fulfilling their aspirations for parenthood [44]. --- Access to reproductive healthcare --- Availability of Fertility Treatments and Information Ensuring widespread access to comprehensive reproductive healthcare services is a cornerstone of supporting individuals who opt to delay childbearing. This entails providing a spectrum of fertility treatments and counseling that cater to the specific needs and aspirations of those considering delayed parenthood. In a rapidly advancing field, individuals contemplating the postponement of parenthood can avail themselves of various options, including fertility preservation techniques and medical interventions. Fertility preservation, such as egg freezing, empowers individuals to safeguard their reproductive potential until they are ready to start a family. Access to accurate and informative resources about reproductive health also equips individuals with the knowledge necessary to make informed decisions aligned with their life goals [45]. --- Ethical Considerations in Assisted Reproduction The rapid evolution of ARTs has introduced a host of ethical considerations that policymakers must grapple with. Innovations like egg freezing, surrogacy, and genetic testing have redefined family planning possibilities and raised complex ethical questions. Balancing the principles of individual autonomy, reproductive rights, and societal well-being is paramount. Policymakers are tasked with creating regulations that safeguard the rights of all parties involved while navigating issues related to consent, potential exploitation, and the well-being of both donors and recipients. Addressing these ethical concerns is crucial for ensuring these technologies' responsible and equitable use [46]. --- Health Insurance Coverage for Reproductive Services The role of health insurance in shaping family planning decisions must be considered. Comprehensive health insurance coverage for reproductive services, particularly fertility treatments, enables individuals to make informed choices about delayed childbearing. With the high costs associated with fertility treatments and ARTs, insurance coverage mitigates financial barriers that might otherwise deter individuals from pursuing their desired timelines for parenthood. By providing access to these services, insurance coverage ensures that individuals have the necessary resources to navigate the complexities of fertility preservation and reproductive interventions, offering them a range of possibilities regardless of their socioeconomic status. This coverage empowers individuals to take control of their reproductive journey and aligns with the broader goal of promoting equitable access to healthcare services [47]. --- Family support and social infrastructure Importance of familial and community networks: Familial and community support networks are pivotal pillars in facilitating the journey of delayed childbearing. In an increasingly diverse and dynamic society, these networks provide invaluable emotional, practical, and childcare assistance to individuals and couples navigating the complexities of parenthood at an older age. The presence of supportive family members, friends, and neighbors can mitigate the challenges that come with delayed childbearing. This support extends beyond mere assistance, encompassing a sense of belonging and shared experiences that foster resilience and mental well-being. Familial solid and community networks offer guidance, share wisdom, and provide a safety net that enables individuals to approach parenting with greater confidence, even in the face of potential hurdles [48]. Childcare facilities and support systems: Providing well-developed childcare facilities and comprehensive support systems is fundamental for individuals who opt for delayed childbearing. As more individuals choose to balance their family aspirations with career pursuits and personal ambitions, the availability of accessible and affordable childcare options becomes paramount. These facilities offer children a safe and nurturing environment and allow parents to engage in other pursuits, including education and professional growth. Access to reliable childcare helps alleviate the strain on parents, allowing them to maintain a work-life balance conducive to their personal and familial well-being. These support systems empower individuals to actively participate in parental roles while nurturing their aspirations and goals [49]. Government policies to promote family planning: Government policies aimed at promoting family planning and supporting the concept of delayed childbearing wield significant influence over the decisions of individuals and couples. Policies designed to incentivize delayed childbearing can take various forms, such as tax benefits, subsidies for fertility treatments, and comprehensive healthcare coverage. By acknowledging and supporting the diverse reproductive timelines of individuals, governments can contribute to creating an environment where personal and family aspirations align harmoniously. Public awareness campaigns about reproductive health and family planning options further empower individuals with information, allowing them to make informed choices about when and how to embark on parenthood. These policies reflect the evolving needs of modern society and lay the foundation for a future where delayed childbearing is embraced as a valid and respected choice [50]. --- Future directions and research gaps Identifying areas for further research: As we explore the intricate complexities inherent to delayed childbearing, it becomes evident that several domains demand further scholarly inquiry and investigation. These uncharted research territories present opportunities to unveil hidden facets of this phenomenon, enriching our understanding and shaping informed approaches [51]. Cultural and regional variations: Exploring cultural norms, values, and regional disparities concerning delayed childbearing has the potential to unveil a global tapestry of influences. As cultures vary greatly in their expectations, beliefs, and societal structures, understanding how these factors intersect with the decision to delay parenthood can illuminate the worldwide scope of this trend. This examination may reveal the intricate ways in which local and global forces interplay, leading individuals to navigate the choice of parenthood within the context of their unique cultural landscapes [52]. Impact on fathers: While the focus of this review primarily centers on maternal aspects of delayed childbearing, an equally important avenue for investigation lies in understanding the experiences, decisions, and well-being of fathers. Exploring how delayed parenthood affects fathers can shed light on their evolving roles and responsibilities within the family unit. This inquiry may uncover how fathers navigate their aspirations, career trajectories, and personal development in conjunction with the decision to delay childbearing, contributing to a more holistic understanding of familial dynamics [53]. Intersectionality: The rich tapestry of human experiences is woven with the threads of various identities, including race, ethnicity, socioeconomic status, and sexual orientation. Investigating the intersections of these identities with delayed childbearing can unravel unique challenges and opportunities faced by diverse populations. The emerging disparities and commonalities can provide insights into how societal norms, expectations, and access to resources shape decisions to postpone parenthood among individuals with varying backgrounds. This exploration can deepen our comprehension of how systemic factors intersect with personal choices, fostering a more inclusive and nuanced understanding of delayed childbearing [54]. --- Long-term consequences of delayed childbearing Health outcomes for parents and children: Understanding the health outcomes of parents who opt for delayed childbearing offers valuable insights into this trend's physical and emotional dimensions. As individuals postpone parenthood, their age at conception rises, which can influence maternal and paternal health. Research in this area explores the potential links between advanced parental age and health outcomes. It investigates the incidence of pregnancy-related complications, such as gestational diabetes, hypertension, and cesarean deliveries among older mothers. Additionally, it examines the psychological well-being of older parents, considering factors such as stress, emotional resilience, and the experience of raising children later in life. Equally significant is the examination of long-term effects on the offspring. Studies delve into the potential associations between delayed childbearing and children's health, developmental milestones, and psychological well-being, offering valuable insights into the holistic implications of this trend for both generations [55]. Social and economic implications: Delayed childbearing transcends the individual and family realms, encompassing broader social and economic dimensions. Investigating the social and economic implications provides a comprehensive view of how this trend interconnects with larger societal structures. For instance, the impact on intergenerational dynamics considers how delayed parenthood might affect relationships between parents and their adult children. Exploring the caregiving responsibilities that older parents might face as they age further illuminates family dynamics. From an economic standpoint, delayed childbearing can influence workforce participation and retirement patterns. Individuals who become parents later might intersect with their peak career years, potentially impacting productivity and retirement planning. By examining these implications, we gain insights into the potential ripple effects of delayed childbearing on various facets of society, from family structures to labor markets and social support systems [56]. --- Anticipating societal shifts and evolving trends Technological advances and reproductive choices: The relentless progression of reproductive technologies stands to redefine the contours of delayed childbearing. As these technologies continue to evolve, they can revolutionize family planning options. With advanced fertility preservation techniques, individuals can extend their biological windows, fostering greater autonomy over when and how they choose to embark on parenthood. As these technologies become more accessible and refined, they will likely play a pivotal role in reshaping the landscape of delayed childbearing, offering new avenues, and expanding the range of reproductive choices available [57]. Policy adaptation: The dynamic interplay between delayed childbearing and policy frameworks is critical. As societal shifts transform family dynamics, policies that accommodate these changes must adapt. Continuous monitoring and evolution of policies to support delayed childbearing are paramount. This includes offering comprehensive reproductive healthcare and implementing work-life balance initiatives that facilitate the dual roles of career and parenthood. The responsiveness of policies to the evolving needs of individuals and families will be instrumental in promoting overall family well-being and societal cohesion [58]. Changing definitions of parenthood: Parenthood is undergoing a profound transformation. Delayed childbearing intersects with this evolution, offering fresh perspectives on traditional notions of family and parenting roles. As individuals prioritize personal goals and aspirations, they reshape the narrative around parenthood, highlighting the diverse pathways leading to family formation. Future research can explore how delayed childbearing contributes to this evolving definition of parenthood, embracing a spectrum of choices and fostering a more inclusive understanding of familial relationships [59]. --- Conclusions In conclusion, this comprehensive review has shed light on the intricate web of factors influencing delayed childbearing, offering a clear picture of the multifaceted decision-making process. By synthesizing historical context, societal dynamics, psychological intricacies, and policy implications, we have underscored the multidimensional nature of this phenomenon. Our findings emphasize that delayed childbearing reflects evolving aspirations, changing expectations, and expanding opportunities in the modern world. Recognizing the long-term consequences and major factors in this trend is crucial. Proactive measures are essential to navigate the challenges it poses and harness its potential benefits. This includes the development of supportive policies, understanding the emotional nuances involved, and fostering open dialogues among individuals and within society as a whole. By doing so, we can create an environment that not only embraces the diverse pathways to parenthood but also thrives in the presence of delayed childbearing. In the everevolving landscape of parenthood, research, understanding, and empathy will be the cornerstones of a society that not only adapts to the complexities of delayed childbearing but also flourishes in its presence. --- Additional Information --- Disclosures --- 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.

This review article delves into the intricate landscape of delayed childbearing, shedding light on the factors influencing individuals' decisions to postpone parenthood. In a world undergoing rapid social, economic, and technological transformations, the concept of when and why to become a parent has evolved significantly. We explore historical trends, societal norms, psychological dynamics, policy implications, and prospects surrounding delayed childbearing. This review underscores the diverse influences shaping this trend, from economic considerations and changing cultural perspectives to advancements in reproductive technologies and the complexities of work-life balance. By examining the emotional dimensions and longterm consequences, we comprehensively understand the implications for individuals, families, and societies. As we conclude, we emphasize the importance of addressing challenges and embracing opportunities to create a supportive environment for those navigating the complex decisions tied to delayed childbearing.

Introduction The process of adoption and advancement of e-government led to the birthing of the Open Government Data movement that has been embraced by both developed and developing countries [1]. OGD is government or public data that is available, usable, reusable and accessible at the least cost possible; such data should be both technically and legally available [2]. The concept and practice of OGD were introduced as a means of avoiding secrecy in government by making government data technically and legally available to citizens [3]. OGD can be regarded as "machine-readable data which is discoverable, available, and downloadable through dedicated internet portals without cost to potential data users" [4]. The OGD ecosystem's main stakeholders are Data Publishers and Data Users , who have differing roles and views of data [5]. These stakeholders are required to collaborate despite their differences in roles and views of OGD to enable effective implementation [6]. The expected outcome of implementing OGD is to create and generate public value [7]. Governments who want to formally engage in the OGD movement, join by signing as partners of the Open Government Partnership Initiative and by completing biennial Action Plans. Ghana joined the OGPI in 2011 and has developed and completed multiple Action Plans since 2011, however the publication and use of OGD remained limited. This has been attributed to inadequate intermediaries and challenges of data quality [8]. Data quality is concerned with validity, confidentiality, privacy concerns, liability, completeness, metadata, technical and semantic interoperability [9]. Data quality in developing economies is a challenge [10] that influences both the publication and use of government data. Although OGD research has received a lot of attention in Information Systems research [1,2], there is still a paucity of contextualized research on how socially constructed behavioral patterns and practices influence the institutionalization of OGD publication and use in sub-Saharan Africa. Such factors have been recognized as important determinants of the successful institutionalization of IS [11]. While data ownership is regarded as important in IS research [12], this is lacking in OGD research. While the factors that influenced OGD in Ghana may have included data quality, there existed underlying socially created and recreated patterns and practices that were influencing the publication and use of OGD and its institutionalization. For this reason, a case study was performed that drew on theories such as Giddens' structuration theory in order to meet this study's research objective, namely to understand how social processes have influenced the institutionalization of OGD publication and use in Ghana. The paper is structured as follows. First, key concepts are introduced and the case setting explained. Subsequently, the research method and theoretical underpinnings are provided. This is followed by an analysis and discussion of the findings. Finally, the conclusion of the study reflects on the findings and provides recommendations and suggestions for future research. --- Background: Key Concepts Open Government Data is linked to open data as well as open government. An amalgamation of these two terms gives rise to the concept of OGD [13]. Open data refers to the free, unrestricted access, use and reuse of data [14] while the open government is an initiative by governments to make their data available on data web portals to promote transparency, accountability and to increase collaboration with stakeholders [15]. However, OGD should not only be associated with the availability of data on government web portals but also with the provision of data that has reusable capabilities [16]. Actors in the OGD ecosystem include public administrators, bloggers, NGOs, academic researchers, data journalists, international organizations, donors and beneficiar-ies [5]. Actors in this study are individuals who fit into the description of Data Publishers and Data Users. The OGD ecosystem actors' understanding of OGD is shaped by contextual factors such as interest and power [5]. Institutionalization refers to the process of routinizing cultural practices, rules and norms [17]. Institutionalization attempts to explain how institutional rules, cultural practices and norms become accepted or rejected in a social system or structure [18]. Institutionalizing OGD practices and policies in public institutions has become a challenge due to the inability of government to merge openness into known rules, norms and cultural practices [19]. The significance of understanding the role of power in IS cannot be underestimated. Power is often used by actors as a way of influencing each other [20]. The process of information dissemination and control among actors in organizations unearths issues of how power is distributed within the organization [21]; how actors acquire this power determines how they disperse it. The innovation and routinization of IS by actors in an organization, unravels different notions of power; hence the need to combine different theories for suitable interpretation and conceptualization [20]. Ownership has been difficult to define due to its complexity [22]. Understanding the concept of ownership is regarded to be as important as the acquisition of technical skills, education, finance and infrastructure in the era of openness [23]. An organization's perception of data ownership influences critical decision-making such as IS outsourcing, centralization or decentralization [22]. Networks are created through the social interaction between actors in an organization as well as between organizations [24]. Actors play different roles and occupy different authoritative positions in the networks [24]. Such roles and positions create an atmosphere of perceived trust between members of the network. Networks vary in size and are dependent on the actors in the social connection and their interaction. Organizational success is dependent on social networks [25]. In the context of knowledge management, networks are created to transfer tacit knowledge and develop new knowledge, which organizations depend on for future endeavors and transfer of skills [26]. In the knowledge network, organizations support a repository of both explicit and tacit knowledge which is distributed among actors [27]. Research in OGD has thus far paid limited attention to network aspects. This study considers the role of network creation in the successful use of OGD by Data Users. Networks can enable Data Users access to funding support, international exposure, determine data quality and access to data. The primary theoretical lens used in this study to highlight the social processes influencing the institutionalization of OGD is Structuration Theory [28]. The relevance of ST to the study lies in its ability to surface underlying social factors that influence people's behavior and practices over time [29]. Castells' [30] and Honoré's [31] theoretical explanations to power in networks and data ownership were used to provide additional theoretical explanations to the empirical findings. The combination of theories is recognized as an acceptable practice in IS research [32]. --- Case Study Context Ghana is a country in West Africa with an estimated population of 28 million people. The country is associated with democracy since 1992. Ghana has an elected president, a parliament, an independent judiciary, electoral commission and different public sector institutions. The public sector institutions are in charge of implementing government initiatives and fostering the relationship between government and citizens. Ghana became a signatory to the 'Open Government Partnership Initiative' in September 2011. This was a way of strengthening prevailing open government frameworks incorporated in the practice of democracy [33]. The OGPI seeks to attain these goals through government commitments. The completion of the first Action Plan set the motion for the development and implementation of subsequent Action Plans. Lessons learnt, gaps identified, and fissures recognized from the previous Action Plans informed the activities of subsequent ones. Recent developments of the OGD initiative in Ghana can be traced to several activities such as workshops that focused on creating data awareness and increase in stakeholder engagement with data. The workshops also aimed at training participants on how to use available government data to create mobile and web applications through hackathons and advocacy activities. For example, a hackathon challenge was held in April 2019, this led to creating a "waste to gold" platform with an aim of tackling waste management in Ghana [34]. The workshops also discussed how to promote collaboration between different institutions, identification of data needs, promoting data use and improving data quality [35]. The Ghana OGD ecosystem has two government approved publishers. The two publishers are the Ghana Statistical Service and the National Information Technology Agency . The Ghana Statistical Service is regarded as the 'Government Statistician'. After Ghana's signatory to the OGPI, the responsibility of GSS has extended beyond just a collection and disseminating of statistical data. GSS is mandated to ensure that government data is open, accessible and in user-friendly formats. GSS is now also charged with the responsibility of developing, collecting, disseminating and reporting government data and further assessing it based on the indicators of the Sustainable Development Goals . The National Information Technology Agency was established in 2008 under the Ministry of Communication. The agency was mandated to act as a backbone for e-government in Ghana and the implementation of government ICT policies. After Ghana became a signatory to the OGPI, the responsibilities of NITA were extended to include developing the open data web portal in 2011. NITA was expected to coordinate and publish data from the various government institutions . This was required to remove data redundancy and create a unified platform that makes government data available and accessible and at the least cost possible. DUs include Non-Governmental Organizations , data enthusiasts, data analysts and data journalists, who create mobile-based technology solutions and educate citizens with the help of data. Their demand for and use of government data is critical because they can create and trigger major impacts. Some key NGO's are Mobile Web Ghana, Esoko, Famerline, as well as data enthusiasts, data scientists and data journalists. These NGOs have been actively involved and collaborated with government and international organizations due to the OGD movement. For example, Mobile Web Ghana has been actively involved with organizing the Ghana Open Data portal upload challenge and hackathon in 2019 and mapping for the Open Cities Accra Project [34]. The Open Cities Accra Project uses OpenStreetMap, field data collection and remote mapping to make flood-prone areas in Accra more resilient to flooding. Mobile Web Ghana has been involved in a workshop on domestic violence, child labor and Data Management and Publication Training for Government Agencies and Ministries. Likewise, Esoko partnered with government to collect and publish data on about 10,000 farmers in 10 districts. This data was linked with a social interventionist program called Livelihood Empowerment Against Poverty [36]. --- Research Methodology The research strategy for this study entailed a single descriptive case study [37], namely the case of OGD in Ghana. The study was executed in an interpretive fashion. Data were initially collected from the two main OGD publishers in Ghana and several OGD users . The initial data sample was changed due to emerging themes, for instance, the researcher added other government institutions who were not regarded as OGD publishers. The sample population of OGD users were mostly based in the Greater Accra, Ashanti and Central regions. These three regions are urban cities and well populated. The Greater Accra region, for instance, is the capital city and also the seat of government, the region houses all the government institutions. Semi-structured interviews were conducted with the assistance of an interview guide. Table 1 provides a summary of the interview participants and data sources. Giddens' structuration theory was used as the main theoretical underpinning for this study. However, during the analysis of the empirical findings additional explanations were required. The findings indicated that data ownership, power and network creation were socially constructed and thus had been created and recreated over time. Data ownership, power and network creation were already existing structures that kept routinized activities the same, which is line with the Giddens' structuration theory [28]. However, Ghana's signatory to the OGD movement triggered changes which were not in-tune with these routinized existing structures. Thus, these existing structures instead of aiding the OGD movement rather led to the un-institutionalization of OGD. This affected both the publication and use of OGD in Ghana. Thus, in the discussion of the findings, the study blended structuration theory [28] with Castells' categorization of networks [30] and Honoré's categorization of data ownership [31]. Giddens' structuration theory is concerned with the reciprocity between human agency and social structure [28]. Structuration theory highlights the duality of interaction between social practices and human actors. The interaction between human actors and social behavioral patterns are produced and reproduced over time [29]. Social structure, in terms of rules and resources, can either facilitate or constrain social activity [29]. There is, therefore, a recursive relationship between structure and actions [28]. Structuration theory is used by researchers to understand social occurrences and the reproduction of behavior and practices across space and time [29]. The repetitive nature of these behavior and practices overtime become institutionalized as part of the social system. Structuration theory can be summarized by the dimensions of the duality of structure namely structures of signification, domination and legitimation, linking respectively to interactions of communication, power and sanction [28]. Table 2 lists the theoretical constructs of structuration theory that are applied in this study. --- Table 2. Concepts of Structuration Theory applied in this study --- Construct --- Propositions Particularization Agency Also referred to as human, social actors, individuals or people. Giddens' assumed and recognized the knowledgeability of actors within the social system. Agency also determines acceptable and unacceptable behavior within a structure plus its accompanying rewards or sanction. Actions are replicated recursively. The ecosystem has multiple actors with different roles and varying meanings attached to OGD activities [5]. --- Structure The structure consists of rules and resources. The existence of structures is dependent on the intertwined relationship between structures and agency. Structures enable the recursive production and reproduction of social systems [28]. Different rules and resources are used by actors with the aim of either publishing or using OGD. There national, international and organizational rules that have been produced and reproduced over time --- Domination The structure of domination is evident in every social system and noticed via the unevenness of allocative and authoritative resources [28]. Social actors perceived data as a resource which needed to be controlled. Observations during the group interviews showed the existence of actors' ability to control others based on positions of authority and symbolic capital. --- Communication Individuals within a structure communicate based on commonly acceptable schemes. Individuals are conscious schemes as they part of Actors created networks purposely to get access to government data, exposure, funding and determine the quality of datasets. the structure of signification . --- Legitimation Norms are part of the legitimation structure which allows sanctions to be evolved if not adhered to. Actors within the social system are aware of such norms as well as the sanctions. Organizational actors were aware of the norms that guide data publication and use as well as the sanctions. But suctions in this context was symbolic. Castells [30] categorizes four types of networks namely: 1. "Networking Power: the power of the actors and organizations included in the networks that constitute the core of the global network society over human collectives and individual who are not included in these global networks. 2. Network Power: the power resulting from the standards required to coordinate social interaction in the networks. The exercise of power is dependent on rules of inclusion and not by network exclusion. 3. Networked Power: the power of social actors over other social actors in the network. The forms and process of networked power are specific to each network. 4. Network Making Power: the power to program-specific networks according to the interest and values of the programmers, and the power to switch different networks following the strategic alliance between dominant actors of various networks". These types of power in networks exist in social systems where actors perceive they have power over other actors due to the personalization of the control of data. However, the type of network is determined and defined by the particular network actors are involved in within the social system. Honoré's [31] categorization of ownership includes: "The right to use whatever is owned; the right to control the use of whatever is owned; and the right to remain in control of what is owned, without interference from others". Honoré's [31] grouping identifies ownership from the perspective of property that requires identification and personalization by individuals. Honoré's work is relevant to this study, since it was found that actors' perception of data led to individual personalization of data which constrained data sharing and publication. --- Findings --- Thematic Analysis A thematic analysis was performed [38]. "Thematic analysis is a method for identifying, analyzing, and reporting patterns within data" [38] p 79. The thematic analysis in a study already starts when transcribing the data and ends with the completion of the study. Table 3 lists the themes and how often they were referenced. Data Ownership. Ghana's signatory to the OGPI has unearthed some contextual issues which were related to who owns the data. DPs and PSs were conscious of the concerns surrounding data ownership. The issue was frequently mentioned in interviews. Data owners were contextually defined to be the actual government institutions in charge of collecting, processing and storing citizen's data. PSs perceived that institutional resources were used to accumulate and store citizen's data. This gave institutions a sense of ownership over data. Also, the existence of unintegrated and manual processes in the public sector was explained as a contributing factor. Both DUs and DPs explained the importance of giving recognition to the sources of data during use and publication. To DUs this acts as a form of protection. OGD publishers and users identified public institutions and online portals as critical primary sources of data. The contextual explanation of ownership resulted in complexity with the inception of the OGD movement. OGD required signatory governments to develop a web portal aimed at publishing government data. However, critical attention has not given to data ownership. Consequently, there was empirical evidence of various arguments about who owned data within the OGD ecosystem. These arguments on data ownership were unearthed due to the inadequate nature of the published data. While some of the interview respondents believed that government data belonged to "government' others believed it belonged to the citizens. However, in a report on the impact and status of Open Data in Africa, the authors explained that "data belonged to the citizens" [39]. While some interview participants explained that data on the government-mandated web portal was regarded as government data, others asserted that, though the data was available on government web portals, the data is about the people, hence it belongs to them. As emphasized by some of the interview respondents: "The misconception is, who does the data belong to? This is a critical discussion that we need to have but has not been seen as important. It is like it is difficult for the very people whose data has been accumulated to get access to it" . DUs recognized the value of understanding data ownership within the local OGD ecosystem. Data was understood as a powerful tool of which the owners or publishers should be recognized. Understanding data ownership enabled DUs to disassociate themselves from issues that arose after publishing stories. Identifying and recognizing data owners was essential as it provides a type of credibility in situations where sensitive stories had to be told. Recognizing data owners within the OGD ecosystem by DUs enabled them to identify which DPs' data to use. The interview participants explained that data was obtained and accessed from different sources. These sources ranged from national to international websites. While some of the DUs had a pre-determined mindset on their sources of data, others could be categorized as "freelancers" who used government data from any source; to such DU the most important aspect is the availability of the data. An interview participant stated the following: "I sometimes get it from data.gov.gh, which is down at this moment. In the last two years, if I need government data, that is where I got to first. --- I also check the website of DP1 and the websites of other government institutions. I check there to see if I can find the data I want." . Actors' perception of Data Ownership led to data control. Controlling data led to institutional personalization of data and the desire to control the data both within and outside the institutional boundaries. Institutional personalization followed from the use of resources such as organizational time, money and skills used in collecting, process and storing data; there was a desire to exert some kind of influence. In addition, the subtle autonomy and integrated system led to the development of the concept of institutional personalization. Some of the participants elucidated that time and money invested in data gathering, processing, storage and disseminating data influenced the attitudes towards control. Institutional personalization of data also resulted in an attempt to control data use by potential and actual users. Some OGD publishers explained that it was essential to control data both within the institution and outside the institutions for ethical reasons including data ownership. Institutional personalization of government data further resulted in the desire to continuously monitor or control data use. It influenced both OGD publication and use. --- "If I give you my data, I have the right to know what you are going to use it for and be acknowledged. I need to feel it is still mine. But the moment I give it out I lose that feeling […]. ". Network Creation. Networks were intentionally and unintentionally created by Data Users. Networks were social connections created between DUs, international donor organizations, national technology enthusiasts as well as government institutions. These networks were created for different purposes. The purposes included financial and technical resources as well as data access. These networks provided a type of social connectedness, which were established as a means to express domination with regards to access to data and as a source of competitive advantage. Competitive advantage existed among DUs within the OGD ecosystem. This social connectedness was established either based on friendship, social engineering or professional standards. Creating social connections within the OGD ecosystem was deemed as a requirement because it provided the DU both social and technological exposure. Most of the DUs used available data either to tell stories or create technological mobile applications that were intended to trigger social conversation and create an impact. However, most of the DUs were in existence before the introduction of the OGD movement in Ghana. The OGD movement and created networks led to an increase in the incorporation of government data into their routinized activities and structures. "We are supported by different organizations, it is needed for exposure. When you go to our website you will see them […] they are mostly international. You must be connected since your connections give you an advantage over other organizations in this thing that we are doing. Knowing the right people in this space is necessary, not just nationally but outside Ghana" . Created networks were used as a means of interaction between data publishers and users. This interaction was required to either enable quick access to data or provide additional understanding of the available data. The networks created by DUs were: 1) National Networks, this was sub-categorized into network created between DUs, PSs and DPs and Networks created among DUs; 2) International Networks. --- Power in Networks. The networks created by DUs transcended beyond the local environment to include international networks or connections. The networks, both local and international, were established to ensure that government data was technologically and legally opened and to reuse data to create value for citizens. Reasons for creating international networks also included funding and exposure. These networks led to the creation and creation of power-related structures that were founded on access and use of government data plus other resources. The type of power emanated from the ability of the DU to control or influence the activities of DP and government at large. The power within the networks was used to trigger responses from both local and international OGD communities. The changes and responses that have occurred within the OGD ecosystem would not have been possible without the actions of the socially created power in networks. For example, by using and analyzing different government data from parliament, Odekro was able to establish trends that critiqued parliament attendance by parliamentarians from 2013 to 2016. The report also exposed the performance of parliamentarians, this was debated as a key determinate of elections. The report of Odekro was one example of power within networks that critiqued government representatives and made citizens aware through the use of government data. "[…] providing citizens, communities, media and civil society with the necessary data with which to hold parliament and MPs accountable […]. This report was critical because the lack of quorum brings to a halt government business and may delay or even rush the consideration or passage of crucial bills[…]" [40]. --- Application of Theory to Findings In this section, the theories introduced in section 4.2 are applied deductively to the empirical findings. Agency and Structure. The OGD ecosystem consists of multiple actors with multiple roles and responsibilities who engage in social practices through social interactions. The interactions within social practices are shaped by commonly understood interpretive schemes, sanctions and communication [29]. The actors could be broadly categorized into Data Publishers, Public Sector Institutions and Data Users. OGD actors also include public administrators, bloggers, NGOs, academic researchers, data journalists and their beneficiaries [5]. Within each of these broad categories existed social actors with varying roles and responsibilities [29]. These actors existed and interacted in various ways within the OGD ecosystem. Such interactions lead to the creation and recreation of both meaning and untended effects of Data Ownership, Power and Network Creation. The interaction between actors was shaped by rules and resources [28]. For example, actors' perceptions of data ownership can be derived from the identification of data as a resource which is guided by different socially constructed legitimation criteria in a network. Data Users interact in a given network to determine data quality before use via a set of acceptable rules and available resources. --- Structures of Signification. Actors within the OGD ecosystems have produced and reproduced subjective meanings associated with Data Ownership, Power and Network Creation over time. For example, Public Sector Institutions have both organizational and individually personalized and subjective view of organizational data which resulted in the desire to control it. Such created meanings influenced the actors' interaction within the social structure or OGD ecosystem and thereby continuously affecting the lack of institutionalization of OGD publication and use in Ghana. Interaction between actors in a social system has a recursive effect on how social systems are shaped over time [28]. While DPs and Public Sectors recognized their respective institutions as owners of data, Data Users also recognized government and citizens as owners of data. The concept of data ownership is linked with the perceptions and behavioral patterns of actors inside rather than outside organizations [41]. The perception of Data Ownership has contributed to the un-institutionalization of OGD publication in Ghana. This can be attributed to fear of losing control and relevance as 'Data Owners'. Using ST to understand the concept of ownership reveals the difference between equity and dialectic control existing between actors in a network as well as the influence of power in such institution [42]. From the perspective of ST social behaviors were to result in sanctions; but the inadequate publication of OGD in Ghana by DUs and PSs which can be linked with specific social behaviors was yet to attract such sanctions. By blending Giddens' ST with Honoré's [31] work, data ownership can be further explained. From the findings, DPs and PSs expressed their right over data in their possession due to the resources used in the accumulation of data. It was assumed that data belonged to such institutions which gave them institutional and individual right of the data. This influenced data control, use and quest to continuously remain in control of data in their possession. Honoré [31] explained that the perception of ownership gives actors three rights: usage rights, controlling usage rights and continuously remaining in control. From the findings, actors held the perception that data belonged to their respective organizations; in some cases individuals in authority, thus they had the right to use, control and continuously be in control of data use. Consequently, such actors reserved the right to determine when data can be released. --- Structures of Legitimation and Domination. The creation of networks by DUs became a type of legitimation criterion for data access, determination of data quality, funding and international exposure. Although network creation became a norm among DUs, there were no clear sanctions attached. ST explains the link between norms and the structure of legitimation plus its attributed sanctions [29]. Organizational actors' perspectives on data ownership and networks created two forms of power: collective and individual power. Collective power extended beyond the quest to control data; it also included access and use of organizational resources needed in the process of accumulating and storing data from a collective standpoint. Data or information distribution often brings out issues of control and power [20]. Organizational actors used their institutionalized power to affirm their positions and control over data, this led to the creation of bureaucratic data request structures that were legitimized overtime. Domination was exercised on external actors who needed data for social intervention programs or publication. Findings from this study explained that actors had different perceptions of power, relating to actors' affiliate organizations, roles and historical factors surrounding data ownership. Knowledge about the existing complexities with the theme of data ownership needs contextual explanations in this era of openness due to large amounts of data being produced and the multiple sources of data [23]. Data ownership and network creation revealed the concept of power, and its implications were expressed by the actors involved. Actors perceived data as an authoritative resource, this led to the creation of bureaucratic structures to control its accumulation, publication and use. Giddens explains power as a social construction, expressed via the use of resources [28]. The creation of networks showed the existence of power relations between OGD actors and international donor partners. Actors, through network creation, exercised power based on the number of international partners they had, funding and ability to access data. Power in created networks can be explained from the perspective of Castells [30]. Networks were not formed in a vacuum, as such there existed evidence of power within these created networks. Although Castells categorized the power in networks into four types, the findings from this study supported three: 1) From the findings, it was evident that DUs created and valued networks created on the global level. It was perceived as a form of international exposure and power, the number of such networks created including the type of international organization meant was used as a form of power over DUs. Castells categorized this type of network power as 'Networking Power'. 2) DUs also considered their ability to have access data which was not published through social interaction as an exercise of power. To Castells, such power was 'Network Power'. 3) DUs, DPs and PSs exercised the third categorization of Castells' grouping. To Castells 'Networked Power' exists when actors perceive to have power over other actors within a social system which is defined by a particular network. --- Conclusion This research set out to understand and explain the factors influencing OGD publication and use in Ghana. Despite Ghana's signatory to the OGPI in 2011, OGD publication and use is yet to be institutionalized by DPs, PSs and DUs. An interpretive case study was performed, drawing from document analysis and stakeholder interviews. The research demonstrated the need for both OGD practitioners and researchers to have adequate knowledge of the contextual backgrounds and ecosystems of countries prior, and during the implementation of OGD. An understanding of the different contextual backgrounds revealed different behavioral patterns and practices that were created and recreated by actors within the various structures. Although these behavioral patterns and practices have existed for many years, the possibility of transitioning must be considered as important. The empirical findings revealed that Data Ownership, Network Creation and Power were socially created and re-created over time by DPs, PSs and DUs. These social patterns and practices existed prior to the implementation of OGD and were recognized as critical factors that influenced the institutionalization of OGD publication and use in Ghana. Actors within the OGD ecosystem had personalized the interpretation and meanings associated with Data Ownership. Ownership of data was seen in three ways by actors: government, organizational and citizen's ownership. The perception of Data Ownership led actors to view data as an organizational and individual property whose access and use needed to be controlled continuously by the 'owners'. The issues surrounding Data Ownership were attributed to the inadequate transitioning from manual, un-or semi-integrated systems to fully automated systems. The implication of Data Ownership on OGD publication led to the creation and recreation of networks by DUs. These networks were used by DUs as a means to obtain data that should have been published as per the OGD signatory. The Networks had both national and international partners; national DUs relied on social connections to enable them to have access to data and determine data quality. International networks were used as a form of exposure and to secure funding. Within these networks existed a socially constructed perception and mutually communicated understanding of power relations that influenced OGD acquisition and use. Castell's explanation of Power in Networks provides an understanding of this part of the findings: 1) DUs created international networks for exposure and a way of expressing power over other DUs. Such networks were displayed on websites as a sign of prestige: this is referred to as 'Networking Power'. 2) Some DUs, thanks to their networks, had access to data that others did not possess; such data was used to trigger national debates and changes: 'Networked power'. 3) DUs also created networks that gave them the power to partner with government and organizes training workshops for civil right advocacy groups. Such networks were based on interactions and principles of social inclusion. In summary, this study uncovered that social patterns and practices around Data Ownership, Network Creation and Power have a major influence on the institutionalization of OGD. This study, therefore, recommends the inclusion of change management principles that take explicit cognizance of these practices to achieve OGD implementation in developing countries due to the disruptive nature of the phenomenon. Future research can focus on the application of change management principles in OGD implementation and quantitative analysis of these themes. Future research could also focus on reasons behind the use of symbolic sanctions.

The promise of Open Government Data (OGD) rests on the publication, availability, use and reuse of government data. This research focused on how social factors such as data ownership, network creation and power enabled or constrained the publication and use of OGD in Ghana, a developing country in West Africa. Ghana's government data was expected to be both legally and technically open. However, socially constructed behavioral patterns and practices such as power, data ownership and network creation played critical roles in influencing the institutionalization of OGD in Ghana. An interpretive descriptive case study analysis helps understand how social processes influenced the institutionalization of OGD publication and use in Ghana. Giddens' Structuration Theory was used as the main theoretical lens in this study because of its ability to investigate the dynamic interplay between social agency and social structures. Findings from the study indicated that power within Ghana's OGD ecosystem is associated with legitimatized practices and behaviors such as data ownership, culture and networks.

Background China in the recent years has the largest number of persons suffering from type 2 diabetes in the world. Type 2 diabetes patients are widely distributed in China, and the diabetes national prevalence is estimated at as of 9.32 % in 2014 [1]. Diabetes patients suffer from many complications that reduce the quality of their life [2,3]. Given that diabetes is a chronic disease and cannot be cured, improvement or maintenance of adequate quality of life is one of the most important public health challenges to both developing and developed nations [4,5]. Previous studies have documented that it's multi-faced and not multi-faced factors influence quality of life, including the stage of the disease [6], life style [7], access to medical services [8], and social economic status [9]. However, few studies have investigated the determinants of quality of life among Chinese diabetes patients [10,11]. Culture that has been considered an ecological-level factor influencing mental processes, human behaviors and health, may also influence quality of life [12]. Different from the western countries where individualist culture prevails, the Chinese culture is more collectivist. Individuals with collectivist cultures tend to seeking for social support and maintain social status in the social structure in which they live [13]. According to the Chinese culture, diabetes patients may buffer physical and mental suffering of diabetes by reconstructing their social networks and obtaining material and emotional social support from their social networks. Social capital has been regarded as an important invisible health resource [14]. With growing recognition of the social determinants of health and numerous empirical studies, largely conducted in Western societies, social capital is an increasingly important construct in healthcare [15][16][17]. Social capital is defined as "features of social organization, such as trust, norms and networks, which can improve the efficacy of society by facilitating coordinated actions" [18]. Aiming at minimizing survival risk, social capital is the produce of human socialization, and the result of conscious or unconscious human investment strategy [19]. Social capital produces effects on human health through the following approaches, diffusion of knowledge about health promotion, maintenance of healthy behavior, access to healthcare services and amenities, acquirement of emotional or material support, and maintenance of mutual respect in social networks [15]. Holtgrave et al. examined the association between social capital and the prevalence of obesity and diabetes at the state level in USA, and concluded that greater levels of social capital might be protective against obesity and diabetes [20]. Years later, the protective effects of social capital on glucose control of diabetes have been found, suggests that social capital may improve symptom of diabetes [21,22]. Social capital may be more important in the Chinese culture because the general nature of guanxi, which is similar to social capital, plays an important impact on social and economic relationships [23]. Chinese adult are more likely to establish relationship through the geographical and kinship relations to fulfill similar functions and collective benefits, especially old people [24]. However, no studies have been conducted in China to assess the association between social capital and quality of life among diabetes patients. The objectives of this study were to describe the relationship between social capital and the quality of life among type 2 diabetes patients in Anhui province, China. We hypothesized that social capital was positively associated with the quality of life in type 2 diabetes patients. --- Methods --- Ethics statement Ethical approval for the study was obtained from the Biomedical Ethics Committee, Anhui Medical University. --- Study population and data collection Based on the geographic distribution and economic level, we selected three cities in Anhui province: Fuyang-north, lower economic level; Hefei-central, high economic level; and Tongling-south, middle economic level [25]. We selected the respondents by simple random sampling in each city. The total sample size was evaluated by the formula: n = Z 2 α/2 / ε 2 P = 385, α = 0.05, ε = 0.10, Z α/2 = 1.96, P = 0.5. Inclusion criteria included individuals who were diagnosed with type 2 diabetes, aged 18 or older, living at home, and did not have cognitive impairment and disability. A convenience sample of 446 adults with type 2 diabetes was recruited for participation in this study. Cross sectional surveys were conducted in these cities between August and October, 2014. All of the eligible respondents were identified from the chronic diseases database at the three local Centers for Disease Control and Prevention . With assistance from CDC staff, respondents were interviewed face-to-face by trained investigators from the Anhui Medical University after they expressed a verbal understanding of the purposes and procedures of the study and signed consent forms. The data collection took about 30 min. Each respondent received about a dollar's worth of gift as compensation for their time after the interview. --- Assessment of quality of life The Short-Form Health Survey is one of the most widely used instruments for evaluating people's quality of life [26]. The Chinese-translated version of the SF-36 questionnaire has previously been used and verified as a scale with high reliability and validity for the measurement of Quality of life among older Chinese adults with diabetes [27,28]. In this study, Quality of life of respondents was evaluated using this Chinese version. SF-36 contains 8 dimensions, including physical function, role physical, bodily pain, general health, vitality, social functioning, role emotion, and mental health. These 8 dimensions can be simplified into physical component summary and mental component summary , reflecting physical health and mental health respectively. PCS and MCS scores were assessed and calculated by T-score transformations [28]. After that, each respondent's PCS and MCS scores were dichotomized by the cutoff point of the first quartile of PCS and MCS scores: scores lower than first quartile scores meant poor quality of life [29]. --- Social capital measurement Social capital has been measured via individual cognition and behavior in the health field. It can be divided into cognitive and structural social capital which may have different effects on health outcomes, even though the controversial about its measurement remains [30][31][32]. Structural social capital includes extent and intensity of associational links or activity, and cognitive social capital covers interpersonal trust, reciprocity, social support and cohesion [30]. This study adopts a perspective of cognition and structure on social capital. According to the World Bank's Social Capital Assessment Tool and the related literature, six dimensions of social capital were considered in this study: social participation, social networks, social support, trust, reciprocity and cohesion [33,34]. We selected some commonly used items corresponding to the six dimensions and adapted them to the Chinese context. The social capital questionnaire was reported in Additional file 1 and had been attached to the last. First, six dimensions were summed by the scores of corresponding items. Second, cognitive social capital and structural social capital were measured by producing a component score of six dimensions using factor analysis. Finally, each respondent's cognitive and structural social capital scores were dichotomized by the cutoff point of the mean component: high cognitive and structural social capital , low cognitive and structural social capital . --- Socio-economic status and risk factors SES contained information of residency , gender, age, ethnicity, marital status, education level, main occupation and monthly income. Risk factors contained current smoking, participation in a moderateintensity physical activity , two-week prevalence of any diseases, and comorbidity of other chronic diseases [10,35,36]. --- Data analysis Descriptive statistics was performed on the sample, and the results were expressed as mean ± standard deviation or percentage. Cronbach's alpha values were calculated and used to evaluate the reliability of social capital scale in reliability analysis. Scores of social capital were calculated in factor analysis. Crude odds ratios and 95 % confidence intervals were calculated in order to analyze associations between Quality of life and social capital by binary logistic regression . The adjusted ORs measuring the association were estimated by controlling for SES variables and both SES and risk factors . The statistical analysis was performed using the SPSS statistical package , and p ≤ 0.05 was taken as the statistically significant level. --- Results A total of 436 respondents completed the questionnaire, with a respondent rate of 97.8 %. The overall Cronbach's alpha coefficient for social capital was 0.87, and for each dimension Cronbach's alpha coefficient was 0.79 , 0.62 , 0.75 , 0.91 , 0.63 and 0.87 . The Cronbach's alpha coefficient was 0.88 for cognitive social capital and 0.79 for structural social capital. Table 1 present descriptive information about the study sample, the respondents had a mean age of 67.1 ± 10.2 years , most of whom were female . The majority of respondents were Han ethnicity and married . About two thirds had low income . A large proportion of respondents participated in the moderate-intensity physical activity per week, did not smoke , and were not ill over the past two weeks . A little more than half of respondents had other chronic diseases , including cardiovascular diseases, cancer, chronic respiratory diseases. The mean scores of PCS and MCS were 67.4 and 64.9, respectively. Based on the cutoff points for PCS and MCS , 106 participants were in poor PCS and 109 in poor MCS. The proportions of respondents who had low cognitive and structural social capital were 47 % and 64.4 % , respectively. The results of logistic regression are showed in Table 2. Cognitive social capital was significantly associated with PCS and MCS. In model 1, diabetes patients with higher cognitive social capital had higher odds of high PCS and MCS . After adjusting for SES and risk factors, cognitive social capital was still significantly associated with PCS and MCS in model 2 and 3. The ORs for high structural social capital compared to high PCS and MCS level were not significant in all three models. --- Discussion This study was to investigate the association between social capital and quality of life among type 2 diabetes patients in Anhui province, China. Findings of this study document that the levels of both quality of life and social capital were relatively low among diabetes patients, but strong positive associations between quality of life and social capital. Given that diabetes patients suffer longterm physical and emotional complications of type 2 diabetes, our results may provide important information regarding improvement of quality of life via the enhancement of social capital. To our knowledge, it is the first study in China to investigate such associations in type 2 diabetes patients. Influenced by the social environment, awareness and treatment among Chinese diabetes patients is less than the western countries [37]. There are deficiencies in treatment and management of Chinese type 2 diabetes patients, including low ratio of community health service staff to patients, insufficient service and public health funding, and limited access to medical services [38]. These deficiencies may reduce quality of life among Chinese type 2 diabetes patients, and lead to a higher rate of mortality [1]. We selected items for six dimensions of social capital according to the related literature. Scores of social participation were on the low side, indicating that behavior of social participation may be certain differences between mainland China and western countries. Formal organizations defined in western countries are rare in China. Culture of association is not popular in mainland China, especially those non-government organizations. We considered that individual level social capital is only accumulated by individual's social networks and interpersonal norms in the majority of Chinese residents. Previous studies showed that social capital was positively associated with quality of life in the elderly [39][40][41], adults [42], long-term social assistance [43], patients with fibromyalgia [44], multiple sclerosis patients [45], women [46,47], and AIDS patients [33]. In this cross-sectional study, we found some consistent evidence to support the hypothesis that higher cognitive social capital was associated with higher PCS and MCS, the two domains of quality of life, after adjustment for SES and risk factors. Cognitive social capital indicates ability that individual can use to acquire social resources from family, community, medical services, and society. Patients with high cognitive social capital may actively seek for information, material, and emotional support networks, comply with social norms and peer control, trust and work closely with others in their daily activities, all of which could lead to receive adequate medical services and psychological support to buffer sufferings caused by diabetes [13]. The crude analysis indicated the association between quality of life and structural social capital not significant. In this study, structural social capital was mainly composed of social participation. According to our study, the majority of respondents were rarely participating in formal organizations, such as politic parties, sports associations, religious and professional originations. The low level of participation in such organized activities may lead to low or non-association between this type of social capital and quality of life. --- Limitation This study was subject to several limitations. First, the study population in this survey was a convenient sample, with low representativeness that may deviate from overall Chinese population. Second, because of the nature of the cross-sectional study, the relationship is just a pure association, and need more information to support the possible causal relationship. Third, because social capital used in this study was measured at the individual level, the impact of ecological level and the entire social capital or six different dimensions on Quality of life were not considered. Finally, other risk factors such as blood sugar control and diet habit were not included in the study, which may undermine the main findings. --- Conclusions This study suggests that cognitive social capital may have an important protective role in improving the Quality of life of type 2 diabetes patients in Anhui province, China. This initial finding suggests that the social capital theory may provide a new idea to solve the shortage problem of physical resources in diabetes prevention and control, especially in Low and Middle Income countries . --- Additional file Additional file 1: Social Capital Assessment Tool. --- Competing interests The authors declare that they have no competing interests. ---

Background: To investigate the association between social capital and quality of life among type 2 diabetes patients in Anhui province, China. Methods: In a cross-sectional study, 436 adults with type 2 diabetes were interviewed. The two domains of Quality of life, physical component summary (PCS) and mental component summary (MCS), were measured using the Short-Form Health Survey (SF-36). A modified instrument scale was used to measure cognitive and structural social capital. Multiple logistic regression models were used to assess the associations between social capital and quality of life, adjusting for social economic status and risk factors for health. Results: 24.3 % of participants (106) were in poor PCS and 25.0 % (109) in poor MCS. The proportions of participants who had low cognitive and structural social capital were 47.0 % (205) and 64.4 % (281), respectively. Results of logistic regression models showed that cognitive social capital was positively associated with PCS (OR = 1.84; 95 % CI: 1.12, 3.02) and MCS (OR = 1.65; 95 % CI: 1.03, 2.66). However, the associations between structural social capital and PCS (OR = 0.80, 95 % CI: 0.48, 1.34) and MCS (OR = 0.62; 95 % CI: 0.38, 1.01) were not statistically significant. Conclusions: It is the first study in China to investigate associations between quality of life and social capital in type 2 diabetes. Findings document that cognitive social capital is associated with the quality of life of type 2 diabetes patients. Our study suggests that the social capital theory may provide a new approach to increase physical resources in diabetes prevention and control, especially in Low and Middle Income countries (LMICs).

Introduction Since the emergence of the COVID-19 pandemic in December 2019, about 7 million deaths have been recorded worldwide [1]. Although vaccination plays a crucial role in global Public Health and is the main driver for the eradication of several, but not all, infectious diseases, a substantial proportion of people worldwide are not convinced regarding the necessity and safety of vaccines and remain hesitant [2,3]. Vaccine hesitancy, the phenomenon of refusing or delaying vaccination when necessary and despite the availability of safe and regulated vaccines, is a major obstacle to preventing the spread of infectious diseases [4]. According to the WHO, about 5 billion individuals received the primary COVID-19 vaccination regime globally [5]. However, this has not been adequate, and the phenomenon of vaccine hesitancy has resulted in thousands of preventable deaths during the three-year duration of the pandemic [6,7]. Concentrating in the European Union region, about 45% of adults did not receive the full COVID-19 vaccination regime, while only 15% received the complete regime [8]. In the two countries where the current study focuses, Greece and Cyprus, about 35% of the population remained unvaccinated during the pandemic, skipping the required booster doses [8]. Vaccine hesitancy has previously been linked to concerns regarding safety, a presumable high risk of adverse effects, and concerns about low vaccine efficacy [9]. Additionally, mistrust in official healthcare information [10] and general political mistrust [11] have been linked to higher vaccine hesitancy. Additionally, extensive misinformation and disinformation, particularly through certain social media platforms, termed 'infodemic' in a pandemic context by the WHO [12], appear to play a crucial role in fueling vaccine hesitancy. Furthermore, vaccine hesitancy has also been detected among healthcare workers [13], who are themselves responsible for promoting and administering vaccines [14], further complicating the issue of reduced vaccine uptake. COVID-19 vaccine hesitancy has previously been linked to various demographic and socioeconomic factors. Specifically, socioeconomic factors, such as lower income and residency in deprived or rural areas, have been linked to lower vaccination uptake in a recent systematic review [15]. Similar findings have been identified sporadically in individual studies, with people of lower educational attainment [16], lower employment status [17], and residing in rural areas [18] being more vaccine-hesitant. This association between socioeconomic factors and vaccine hesitancy has not been consistent in all populations investigated [19]. Differences in demographic and epidemiological characteristics, economic and infrastructure circumstances, health system structure, and legal and political aspects in different countries were found to influence the process of universal and fair vaccination. This gives rise to differential influences of socioeconomic circumstances between countries [15]. The role of socioeconomic factors in vaccine hesitancy and refusal also appears to be influenced by cultural circumstances [20][21][22] unique to each population. Focusing on the Mediterranean region, lower educational attainment, a worsening economic situation, and rural living were associated with higher vaccine hesitancy and refusal among Italian adults [23][24][25]. Although the phenomenon of vaccine hesitancy has been previously investigated in the Eastern Mediterranean region [26][27][28][29], socioeconomic determinants are usually overlooked. A comprehensive literature review on vaccine hesitancy concluded that the phenomenon was more pronounced in Eastern Europe and the Middle East [30]. In contrast, evidence of socioeconomic inequalities in vaccine acceptance in these regions is scarce [15]. Given the above, a research gap exists given the scarcity of evidence on socioeconomic determinants of vaccine hesitancy and uptake in different countries and regions around the world, including the Eastern Mediterranean region. This study aims to contribute to this direction by reporting findings from a population-based survey in two countries from this region, Greece and Cyprus, revealing clear socioeconomic inequalities in COVID-19 vaccine hesitancy and coverage. --- Materials and Methods --- Study Setting and Participant Recruitment We conducted a cross-sectional epidemiological observational study in two highincome Eastern Mediterranean European countries, Greece and Cyprus, from early March 2022 to mid-April 2023. The study population consisted of 576 participants over the age of 18, who lived in Greece or Cyprus in the last 3 months, could read in the Greek language, and gave their signed informed consent to participate in the study. Due to the aggravated epidemiological situation due to the COVID-19 pandemic restrictive measures in the countries where the survey was conducted, the sample was selected by sampling predetermined strata . Specifically, a sample of an equivalent percentage to the corresponding proportion of each stratum according to the last census of the population of Greece and Cyprus [31,32] was taken for gender, age, and geographical region . The sampling units for each stratum were selected non-randomly, in person, or online. The sampling approach followed was identical in the two countries, and no country-specific differences or obstacles were noted, presumably due to the very similar societal structures and cultures of the two countries. Probabilistic sampling is the ideal approach in any epidemiological survey. However, it has not been possible to apply such an approach due to the adverse pandemicrelated societal circumstances in the two included countries during the data collection stage, amid stringent containment measures and total universal lockdown for all citizens. Following guidelines from the literature regarding achieving roughly representative samples in epidemiological studies, we chose proportional quota sampling as an acceptable alternative to random sampling approaches, anticipated to reduce selection bias [33][34][35]. Such obstacles to sampling have been faced by numerous population-based studies conducted during the pandemic, with some relying on non-ideal approaches such as convenience sampling [36][37][38][39][40]. On the contrary, our sampling approach of choice, proportional quota sampling, has been proposed as an alternative method for obtaining representative population samples during the pandemic, using both in-person and online participant recruitment cost-effectively without compromising study validity if properly conducted [41][42][43]. The target population proportions for the main strata were the following based on the latest census of each country [31,32]: Gender ; age-group ; geographical region . Participants were recruited from all major regions of Greece and Cyprus , based on the European NUTS classification [44], giving rise to a nationwide sample in both countries. A systematic approach was followed to ensure that selection bias was minimized as much as possible, recruiting participants from different locations and among different socioeconomic strata in order to achieve representativeness. For example, urban, semi-urban, and rural regions were chosen, while in Greece, our sampling was conducted on both the mainland and the islands. Similar stratified and targeted sampling approaches have been followed in similar nationwide studies during the pandemic [45]. In order to determine the minimum sample size required to evaluate our main associations of interest, namely those between socioeconomic factors and vaccine hesitancy, we performed power analysis using the package pwr from the R software environment. For this purpose, we assumed a four-category independent variable , a small-to-moderate association with vaccine hesitancy as a numeric outcome variable, a significance level of 0.05, and a power of 0.90. This analysis estimated a minimum of 360 participants for determining mean differences of small-to-moderate size in COVID-19 vaccine hesitancy between categories of the major socioeconomic factors analyzed . Our sample size of 576 participants exceeds the minimum requirement, indicating that our cross-sectional study was adequately powered to detect the main associations of interest. The final sample size of 576 participants was reached after the termination of data collection in April 2023, as a result of the containment of the spread of COVID-19, to a large extent, in the two countries of interest. --- Questionnaire and Data Collection The questionnaire used was developed for the purposes of the present study. It included questions on sociodemographic factors, COVID-19 vaccine uptake, and other questions relevant to perceptions and intentions toward vaccination . In order to pilot test the questionnaire, we recruited an initial sample of 69 participants, who provided feedback regarding intelligibility, including comprehension and cultural appropriateness of questions, as well as length and overall layout of the questionnaire. The pilot testing did not reveal any issues I any of the questions included. The first questionnaire section assessed data on demographic and socioeconomic characteristics such as gender, age, country of origin and residence, area of residence , marital status, educational attainment, and monthly income. The second section assessed COVID-19 vaccination uptake and vaccine hesitancy. A COVID-19-specific vaccine hesitancy scale was derived as a composite score based on the following: number of doses received; short-term intention to get vaccinated; long-term intention to get vaccinated; voluntary or forced vaccination; and feeling of satisfaction after getting vaccinated. Further details on the assessment of COVID-19 vaccine hesitancy can be found in the subsequent sub-section. In order to evaluate the face validity of key sections of the questionnaire, namely socioeconomic characteristics and vaccination-related information, we also conducted a series of focus groups among this initial sample. Further details about face validity can be found in subsequent sub-sections. Following the pilot testing of the questionnaire, in which no issues were identified, and after confirmation of the face validity of key assessments, the questionnaire was administered in both hard copy and electronic form . The questionnaire took approximately 15 min to complete. Prospective participants were identified online or in person through targeted sampling. This process continued until the percentages of gender, age, and geographical area within the sample approached the corresponding percentages in each general population . Both the online and in-person participant recruitment was conducted with the help of postgraduate research assistants who visited places where there was a large transit of people . Potential participants were approached by research assistants and were either provided with the print version of the questionnaire or, if the participant requested it, the link to the online version. The print version was usually filled out on the spot, while the online version was filled out by participants on their own time. Responses in the electronic version of the questionnaire were automatically collected in the form of a dataset via the Microsoft Forms platform. Participant recruitment took place from both urban, semi-urban, and rural areas, as well as from more affluent and less affluent areas within major cities. In order to estimate the response rate, research assistants were requested to record the number of completed questionnaires and the total number of participants approached. Since data collection was conducted amid the pandemic's strict restrictive measures and in order to reduce the high burden of data collection, it was conducted only for the first 100 participants approached. Among those, 52 agreed to participate and completed the questionnaire, giving a response rate of 52%. Given that non-response did not alter during the duration of the study, we suggest that this estimation accurately reflects the overall response rate throughout data collection. --- Assessment of Main Exposures and Outcomes The study questionnaire included an assessment of key socioeconomic indicators, which were treated as the main exposures of interest. Questions on these indicators were phrased and recorded based on the national census of each of the two countries of interest . Area of residence was derived from self-reported city/town/village of residence, which were categorized based on local demographic criteria as: urban ; semi-urban ; rural . Educational attainment was assessed as the highest qualification acquired based on pre-defined categories . For analysis purposes, these categories were re-grouped into the following four-category variable: Up to High School; College Degree; Undergraduate University Degree; and Postgraduate University Degree. Participant income was assessed as the typical monthly income, based on pre-defined categories . For analysis purposes, these categories were re-grouped into the following four-category variable: <500 €/month; 500-1500 €/month; 1501-2500 €/month; >2500 €/month. COVID-19 vaccine hesitancy and uptake were the outcome variables of interest. The following approach was used for deriving the COVID-19 vaccine hesitancy scale: participants missing 1 dose , participants missing 2 doses , and participants missing 3 doses . Additionally, 1-5 points were added based on the answers to the following 5-item Likert scale questions: "If you were not vaccinated against COVID-19, do you intend to do so?" "If you were vaccinated against COVID-19, was this with your initial intention or were you forced to do so?" "If you were vaccinated against COVID-19, how do you feel about this action?" "If your country adopts long-term vaccination , based on guidelines from international organizations , for total long-term control of COVID-19, do you intend to follow them?" The derived COVID-19 vaccine hesitancy scale ranged from a minimum of three hesitancy points for participants receiving all three vaccine doses and answering very positively to all questions pertaining to initial intention, perceived satisfaction, and long-term COVID-19 vaccination. On the other hand, it reached a maximum of 25 hesitancy points in participants receiving no doses who answered that they would definitely not get vaccinated either in the near future or in the long term, in case this was required. This variable was used as the main outcome of interest in our analyses. In order to investigate the face validity of our COVID-19 vaccine hesitancy scale, we compared the qualitative data collected from the focus groups to the self-reported information gathered through the questionnaire. In particular, the discussion during the focus groups revealed a clear theme characterized by mistrust in authorities and obvious misinformation, with this group of individuals expressing major concerns regarding a presumable exaggeration of the seriousness of the COVID-19 pandemic, with some even denying the existence of the virus. This group of individuals also expressed extreme levels of vaccine hesitancy, claiming that they would definitely not get vaccinated against the specific virus. We labeled these individuals 'pandemic and vaccination deniers,' a term very commonly used by the public in both countries of interest. In particular, the face validity of the quantitatively assessed COVID-19 vaccine hesitancy scale was evaluated by comparing the mean score among those identified as 'pandemic and vaccination deniers' based on the focus group interviews to those not identified as such . This difference in the mean COVID-19 vaccine hesitancy score was highly statistically significant . A binary variable, termed 'incomplete COVID-19 vaccine uptake,' defined as <3 selfreported COVID-19 vaccine doses throughout the COVID-19 pandemic. This variable was additionally used as a secondary outcome in our analyses. --- Ethical Issues All research participation procedures were approved by the Cyprus National Bioethics Committee . Participation in the survey was anonymous, voluntary, and without coercion. Moreover, all participants were aware of the aims and scope of the study, the time of completion of the questionnaire, and how the data would be managed, stored, analyzed, and evaluated. They were also made aware of their right to withdraw their participation from the study at any time. At the same time, the platform through which the version of the questionnaire was administered did not record any identifier of the participant except for the date and time of submission. The data collected was anonymous, and to ensure the rights and confidentiality of the participants, the electronic files were kept password-protected, with only the principal investigator having access to the data. --- Data Analysis Descriptive statistics were calculated for a series of sociodemographic characteristics of interest by country of residence . The chi-squared test was used to investigate the independence between each set of categorical variables , providing a p-value. For inferential statistics, regression models were run with the categorical socioeconomic factors of interest included in separate models as independent variables and COVID-19 vaccine hesitancy and vaccination uptake as the dependent variables, respectively. When vaccine hesitancy was used as the outcome of interest, multiple linear regression was the analysis of choice. When incomplete COVID-19 vaccination uptake was used as the outcome of interest, multiple logistic regression was the choice analysis. The influence of possible confounding factors on the above associations was investigated by including these as covariates in the aforementioned regression models, providing estimates adjusted for these factors. In order to investigate whether the association between socioeconomic factors and vaccine hesitancy differed by country or by major demographic factors, we ran our linear regression models described above, stratifying by country of residence , by gender , and by age group . We also included an interaction term between each socioeconomic factor of interest and the three aforementioned potential effect modifiers . A significant pvalue corresponding to each interaction term indicates a statistically significant interaction . Since these were secondary analyses, relevant results are presented in Supplementary Material . To check the statistical significance of the above-described statistical tests, a p-value at the 5% significance level and a 95% Confidence Interval are reported. All statistical analyses and data processing were performed using the statistical software R, version 4.3.0. --- Results Table 1 below displays the demographic and socioeconomic characteristics of study participants for the whole sample and by country. Our study sample approaches the basic demographic structure of each population in terms of age, with a slight underrepresentation of individuals aged over 60 years and geographical region for Greece . The sample deviates from the demographic structure of the source population in terms of geographical region for Cyprus and gender in both countries . Given the over-representation of women and the under-representation of older individuals , we cannot infer whether our findings definitely apply to specific sub-groups of older people and men, who might not be adequately represented in our sample. The majority of participants were married and resided primarily in urban areas . The majority of participants had a university education , with the proportion being higher in Cyprus than in Greece . The biggest between-country difference in socioeconomic factors was observed for income . In Cyprus, a much higher proportion of participants reported a monthly income of over €2500 compared to Greece . This discrepancy can be attributed to the different socioeconomic circumstances in the two countries, with Greece having been through a major economic crisis during the past decade. In fact, the GDP per capita in the two countries [46] and the average full-time adjusted salary per employee [47] justify the income differences between the two countries observed in our sample. Among the whole sample, 101 participants received no vaccine doses , 79 participants received 1-2 doses , and 362 participants received 3-4 doses . Small differences in vaccination coverage between countries did not reach statistical significance. The proportions observed are consistent with the population vaccination coverage reported in the two countries during data collection [8]. This indicates that the study sample accurately represents the vaccination uptake in the two populations under investigation. Table 2 above presents findings from a linear regression model with socioeconomic factors as exposures and COVID-19 vaccine hesitancy as a numeric scale outcome. Starting with an area of residence, it appears that those residing in rural areas do not substantially differ from those residing in urban areas in terms of vaccine hesitancy. However, those residing in semi-urban areas show much higher hesitancy . This association was apparent both in Greece and Cyprus, although in Cyprus, those residing in rural areas also had increased vaccine hesitancy . None of the tested interactions by country reached statistical significance . When stratified by gender and age group, the increased vaccine hesitancy observed among those residing in semi-urban areas was particularly apparent among men and individuals younger than 60 years . In terms of educational attainment, there is a clear, statistically significant, inverse association with vaccine hesitancy, indicating increasing hesitancy levels with each step up the educational attainment ladder from a College degree and above. This inverse social gradient was apparent in both Greece and Cyprus and slightly more pronounced among men than women . Interestingly, educational attainment showed no association with vaccine hesitancy among older individuals . Monthly income revealed a similar inverse social gradient, with a statistically significant decreasing trend in vaccine hesitancy with increasing income levels above €1500. Although apparent in both countries, this inverse association was more pronounced in Greece than in Cyprus and among men than among women . As for educational attainment, income does not appear to be associated with vaccine hesitancy among older individuals . Table 3 above presents findings from a logistic regression analysis with socioeconomic factors as exposures and COVID-19 vaccine uptake as a binary outcome. Odds ratios represent the likelihood of incomplete vaccination uptake by categories of each socioeconomic factor. These findings corroborate those from Table 2, indicating a higher likelihood of incomplete vaccination uptake among those from semi-urban areas and a decreasing trend of incomplete uptake with increasing educational attainment and monthly income. Differences in the extremes of educational attainment and income reach statistical significance. --- Discussion This is the first study to report clear socioeconomic inequalities in vaccine hesitancy and COVID-19 vaccination status in Greece and Cyprus. We reveal a clear inverse social gradient by educational attainment and income on vaccine hesitancy and vaccination coverage, characterized by decreasing hesitancy and higher coverage with increasing socioeconomic position. We also report a novel finding, revealing that in the two populations under investigation, people residing in semi-urban areas have increased vaccine hesitancy and lower uptake of COVID-19 vaccines. These findings are generally consistent in both countries under investigation, are slightly more pronounced among men than women, and are not apparent among older individuals. --- Comparison to Similar Studies in Other Populations Our results are consistent with previous evidence suggesting socioeconomic differences in vaccination uptake and hesitancy during the COVID-19 pandemic [15]. Since such socioeconomic stratification of vaccine hesitancy and refusal is context-specific and differs from region to region [15], we confirm that the type of socioeconomic gradient observed in the two Eastern Mediterranean populations under investigation is characteristic of what was observed in high-income European countries and North America. The aforemen-tioned social stratification is characterized by higher vaccine hesitancy and lower uptake of available vaccines among individuals from lower socioeconomic strata and those residing in non-urban areas. For example, a nationally linked data study in England revealed that vaccination uptake was lower in more deprived areas and among lower socioeconomic groups [48]. Similarly, inequalities in vaccination uptake and intent were observed among various sociodemographic groups in the Canadian Community Health Survey 2021, particularly concerning low educational attainment and region of residence [49], while in a large-scale US survey, vaccine hesitancy was strongly determined by employment status [17]. Interestingly, the majority of evidence on socioeconomic inequalities in COVID-19 vaccine hesitancy and uptake follows the above pattern [15]. However, a study of another Mediterranean population, namely Italy, revealed slightly different patterns. In this case, they observed a U-shaped association between educational attainment and COVID-19 vaccine uptake, with decreasing uptake in the highest educational attainment group [50]. Along the same lines, the specific study identified the lowest vaccine uptake in lowdensity rural areas rather than intermediate-density areas, as identified in the present study. These regional differences probably stem from different characteristics in the analyzed populations. For example, intermediate-density areas in the aforementioned Italian study include large towns and city suburbs, which are probably sociodemographically different from the semi-urban regions in Greece and Cyprus included in the present study, which have a more rural character. --- Potential Explanations for the Inverse Social Gradient in Vaccine Hesitancy and Uptake Inverse social gradients in adverse health-related outcomes have been repeatedly documented in the literature with different health outcomes [51] as well as health-related behaviors [52]. Different models have been proposed for explaining these social inequalities, including the materialist model, suggesting that inequalities are partly explained via differences in material conditions, amenities, and a deprived built environment [53]; the cultural-behavioral model, suggesting that inequalities are partly explained via differences in health-related behavior and attitudes [53]; the access to healthcare model, suggesting that inequalities are partly explained via differences in access, utilization, and uptake of health services, characterized by the phenomenon of the 'inverse care law' [54]; and the psychosocial model, suggesting that inequalities are partly explained by differences in exposure to psychosocial stressors in the working environment and everyday life, involving aspects such as job strain, effort-reward imbalance, and social isolation [55]. Our study only presents initial findings on the phenomenon of socioeconomic inequality in vaccine hesitancy and uptake. It does not include any results on explaining this effect via mediation analysis. However, it could be speculated that all of the abovesuggested models could contribute to the observed phenomenon. For example, vaccination coverage has been found to be lower in more deprived areas [56] . In contrast, vaccine-hesitant individuals have been found to follow specific health behaviors [57] . Similarly, compromised access to health care [58] and an adverse psychosocial environment [59] have also been linked to higher vaccine hesitancy and refusal. From the aforementioned models, it could be speculated that the most likely scenario is that individuals from lower socioeconomic strata in the two analyzed populations may have a cultural-behavioral profile associated with lower health literacy [60], rendering them more prone to misinformation and institutional mistrust, which have been linked to vaccine hesitancy in the specific populations [61,62]. Material deprivation and reduced access to health care are anticipated to play a minor role in socioeconomic inequalities in this case since the vaccination programs put in place in both countries have been well-organized, systematic, widespread, and free of charge [27,63]. Interestingly, stratified analysis revealed that socioeconomic inequalities in vaccine hesitancy differ slightly in the two countries under investigation, with educational attainment appearing as a more important indicator in Cyprus and monthly income appearing more important in Greece. This might be expected given the differing distribution of income in the two populations, with a large proportion of the Cypriot sample belonging to the high-income category . Based on this finding, income might be a more important socioeconomic indicator in relatively less affluent populations. Socioeconomic differences in vaccine hesitancy were also more apparent among men than women and among younger individuals. The latter finding could be explained by the fact that older individuals were under imminent threat from the virus. Therefore, it would be logical and expected that the elderly from all socioeconomic strata would be keen to get vaccinated, hence the lack of any observed association. Regarding the area of residence, our results agree with the existing literature that individuals residing in non-urban settings are more prone to vaccine hesitancy [25]. However, we show that this phenomenon is more pronounced in semi-urban than rural areas, particularly in Greece. This could be country-specific and might be explained by the fact that in the two populations under investigation, rural areas are usually populated by older people who are generally less vaccine-hesitant [64]. It could be speculated that a combination of relatively younger ages and low educational attainment, which characterize semi-urban regions in the analyzed populations, might be driving the higher vaccine hesitancy observed in residents of such regions compared to residents of both urban and rural regions. As in the case of educational attainment and income, the increased vaccine hesitancy in semi-urban areas is more pronounced among men and younger individuals. --- Strengths, Limitations, and Novely of the Present Study The strengths of our study include the recruitment of nationwide samples from two populations. In these populations, the phenomenon of vaccine hesitancy and refusal has only been investigated either regionally or based on convenience samples, making recruiting participants accessible to researchers conducting these studies [26,61,62,65]. These previous studies are usually not in the appropriate depth [66] or do not systematically investigate social inequality [28,67]. In contrast, our approach involves the recruitment of participants from different settings, regions, and socioeconomic strata. This provides nationwide samples from both countries of interest, allowing a comprehensive and systematic evaluation of socioeconomic determinants. To our knowledge, this is the first study systematically investigating and revealing clear evidence of socioeconomic inequalities in vaccine hesitancy and uptake in the two analyzed populations and the wider Eastern Mediterranean region. Our study also has some limitations, which should be clarified. Firstly, although we followed a stringent sampling methodology applying proportional quota sampling based on recommended procedures [33][34][35], an approach that actually advocated for the recruitment of roughly representative samples during the pandemic [41], we did not manage to reach all required population strata. Hence, men and older individuals are slightly under-represented in our sample, compromising the generalizability of our findings to some extent. However, we note that the phenomenon of over-representation of women and under-representation of the elderly in population-based surveys is common in the literature. Regarding the former, this has been suggested to stem from gender-specific sociocultural dynamics, such as general altruistic considerations [68], making women more willing to participate in research studies than men [45]. Regarding the under-representation of the elderly, this has also been observed consistently and probably stems from several factors. Older people tend to experience a higher frequency of adverse health conditions, disability, and cognitive decline, which might limit their survey participation. Additionally, a significant proportion of this population group is institutionalized and not accessible for recruitment in population-based surveys [69]. Given this, any stratified results by age group presented in the Supplementary Material should be interpreted with caution. due to the relatively low sample size of individuals ≥ 60 years of age . An additional limitation of our sampling approach is that we were not able to ascertain the response rate throughout the study due to stringent containment measures involving universal lockdowns in both countries under investigation. Therefore, our study's response rate is estimated based on an initial sample of 100 participants. We argue that, given the adverse conditions and the restrictive measures during the COVID-19 pandemic involving strict universal lockdowns in both countries, non-ideal sampling should be expected to some degree. It should be noted that even in long-term national health surveys such as the Health Survey for England, sampling and data collection were adapted during the COVID-19 pandemic to account for imposed restrictions and the sensitivities of potential participants [70]. --- Public Health Implications Our findings revealing lower vaccination uptake among population groups with specific socioeconomic characteristics could inform local Public Health programs aiming to achieve the required vaccination coverage in epidemic contexts or regarding routine vaccinations. This could be achieved by tackling more effectively the alarming phenomenon of vaccine hesitancy, identified as a major threat to global health [4], by enabling the targeting of population groups who are particularly vaccine-hesitant, rendering such approaches more tailored and effective. Based on the current findings, such approaches might be particularly effective among men and even more so among younger people than older people. Based on the literature, low educational attainment is an almost universal determinant of higher vaccine hesitancy [15]. Therefore, increasing educational attainment and focusing health education resources on less educated groups will probably have a positive effect on vaccination uptake among specific social strata and the population as a whole. Such approaches would simultaneously reduce social inequalities in health, a major global priority according to the WHO [71]. Furthermore, although not directly deriving from the current findings, it should be noted that vaccinators themselves should be educated further in an attempt to reduce the phenomenon of vaccine hesitancy observed among this important population group [13,14]. Not addressing this crucial matter will pose major obstacles for any vaccination campaign among the general public, possibly fueling further health inequalities. Finally, it should also be noted that the current findings present a specific snapshot during the COVID-19 pandemic and might not reflect overall attitudes towards vaccinations among the study groups of individuals in a non-pandemic context. In fact, views and attitudes on vaccination in particular population groups studied may change in the postpandemic era. Given this, there is a need for further studies in order to consistently inform Public Health policy relevant to tackling vaccine hesitancy and improving vaccination uptake. In conclusion, for the first time, we report clear socioeconomic inequalities in vaccine hesitancy and refusal in two Eastern Mediterranean populations. We suggest that Public Health approaches aiming to tackle vaccine hesitancy at the population level should not ignore the socioeconomic patterning of the phenomenon, making any attempts more effective and socially fair, helping manage epidemics more efficiently, and reducing health inequalities. --- Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/vaccines11081301/s1, Table S1 --- Data Availability Statement: The data presented in this study are available on request from the corresponding author. The data are not publicly available due to restrictions related to data confidentiality and the prohibition of public sharing, as imposed by the local bioethics committee, which granted approval to the current study. --- Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. ---

Despite the rigorous investigation of the phenomenon of vaccine hesitancy and refusal during the COVID-19 pandemic, the socioeconomic determinants of this phenomenon remain poorly investigated on a global scale. Following proportional quota sampling, we conducted a populationbased cross-sectional study. We recruited participants on-site and online from different settings, regions, and socioeconomic strata in two Eastern Mediterranean populations, Greece and Cyprus. Our approach provided a nationwide sample (n = 576) approaching the adult population structure of the two countries, with a slight underrepresentation of men and older people. Our results indicate clear socioeconomic differences in vaccine hesitancy and vaccination coverage, consistent with wider social inequalities in health. In particular, we reveal a clear socioeconomic gradient characterized by lower vaccine hesitancy and higher vaccination coverage, with increasing educational attainment and income. Additionally, participants residing in semi-urban areas show higher vaccine hesitancy and have lower vaccination coverage than those residing in urban and rural areas. Our results could inform Public Health approaches aiming to tackle the alarming phenomenon of vaccine hesitancy by enabling the targeting of population groups who are particularly vaccine-hesitant, rendering such approaches more targeted and effective while at the same time reducing inequalities in the control and prevention of infectious diseases.

Introduction As the process of urbanization and globalization has intensified, an increasing number of people are affected by migration. Many of them are left behind by migrants in sending areas to circumvent the costs and uncertainties associated with migration. Migration is often conceptualized as a household strategy for risk diversification and income maximization. Hence, the economic benefits of out-migration through remittances are usually considered the most crucial channel of the migration effects. This perspective, however, overlooks potential psychosocial costs of household migration, as migration often entails family disruption and prolonged separation. The previous literature on family disruption, mostly focused on marital dissolution in Western societies, provides the basis for understanding how emigration may influence the psychosocial health of the family members left behind. This line of research documents --- Background The stress process, social support, and health Stress is conceptualized as a socially embedded process with important consequences for mental and physical well-being . The sources of stress consist of discrete life events or chronic life strains that entail a considerable amount of change or are perceived as disruptive or undesirable. In developing countries, for instance, poor living conditions and potential sociopolitical tensions are commonly thought to contribute to acute or chronic stress . Stress may show a variety of manifestations, including psychological disorder, cardiovascular illness, and weakened immune system . These detrimental health consequences occur as a result of a disequilibrium of organism functioning and the physiological stress reactivity. The stressful experience, however, does not render all people equally vulnerable because various social resources that people mobilize in response to stress, in particular social support, can intervene at different points in the process and alter the consequences of stress through management of stressful circumstances or modification of the perceptions of problems . Social support represents resources provided by others in the social structure. These resources can be emotional or instrumental that help meet an individual's functional needs. It is often found to protect individuals from the adverse influences of stressful circumstances . Social support also has a direct and positive effect on health because it produces positive experiences and a sense of stability and self-worth . This is mainly reflected in the emotional support and companionship made available by the intimate members of the social structure. Informational and physical resources also can be crucial to health if trusted members of the network promote healthier behaviors, help with daily needs, or provide necessary material resources. Previous studies have noted the relations between stress, social support, and family roles. Women and men are expected to differ in the sources of stress. Relatively speaking, workrelated stress would likely have a significant influence on the well-being of men, whereas experiences within the family domain would be primary for women's psychosocial wellbeing . Gender variations in the health consequences of stress and social support are also documented. Gove and Tudor and Kessler et al. demonstrate the ramifications of gender roles for mental health and attribute higher levels of emotional vulnerability observed among women to the experiences of gendered socialization. This is also related to differing norms of acceptable emotional expression for males and females. As a result, men likely exhibit externalized expressions of distress when facing stress, whereas women tend to show internalized or emotional reactions to stress such as depression. Furthermore, gender roles account for the variation in the use of social support . Women tend to be less likely to suffer from the adverse effects of stress than men because seeking help is congruent with the traditional feminine gender role, and thus, women often mobilize social support more effectively than men to buffer the impacts of stress. --- Family status and health An extensive literature has linked the stress and social support paradigms with family processes, especially marital dissolution. There is unambiguous evidence of an adverse effect of divorce, such as a higher risk of mortality, mental health disorders, stress-related chronic illnesses, and unhealthy behaviors . The key claim is that divorce is a stressful process involving a loss of social resources. Such an experience generates emotional strains and reduces social integration and reciprocal responsibilities. Economic strain is also thought to partially account for this detrimental impact. Economic deprivation, instability, and in extreme cases, poverty, are often associated with divorce, especially among women . Other forms of family disruption have also been investigated with respect to health. The death of a spouse is often perceived as a severe life stressor , leading to the loss of psychological, social, and economic resources, and subsequently bearing deleterious health implications . Incarceration, during which individuals are removed from their families, is often linked to adverse health outcomes because it diminishes families' economic resources as well as destabilizes and stigmatizes them . Military deployment constitutes a stressful experience due to the frequent separation of family members, possible financial concerns, and substantial uncertainties . --- Migration as a form of family disruption The New Economics of Labor Migration links migration and the families left behind . This theory contends that migration decisions are made collectively by families to diversify risks and maximize household economic welfare, particularly in less developed societies with inadequate credit systems and little institutionalized provision for insurance against crop failure or loss of productivity in old age. Thus, families send some of their members out to work in wage labor while others tend the fields, generating surplus capital from the income of the migrants. Family ties often remain strong in the face of migration and a large proportion of migrants' income is devoted to remittances . These transfers not only occur between husbands and wives but also intergenerationally between adult children and older parents . The remittances not only increase household consumption and improve living standards, but can also direct more resources to health-related investments and improve health utilization . Beyond financial remittances, emigration also brings about social transfers of knowledge, attitudes, and practices . Previous research usually perceives monetary remittances to be the most crucial channel of the emigration effects. What has been largely neglected is the important psychosocial dynamics that cannot be directly attributed to the economic transfers . The out-migration of family members disrupts family life and transforms social structures within the family. This change diminishes reciprocal social support, subsequently increasing the family members' emotional burden and resulting in added stressors to compensate for the reduced labor supply needed at home . The detrimental impact of labor shortage is especially strong in areas dominated by agricultural activities. These families have to handle additional agricultural tasks and take charge of the daily operation of the family business. They must also fulfill added family obligations such as child care and family maintenance. In these respects, out-migration gives rise to stressful circumstances and likely has adverse effects on physical and mental health. These impacts tend to extend to the entire family, rather than restricting to particular family members . This comes as an unintended consequence of the migration decision-making process, and the stress level of people left behind is exacerbated when migrants encounter especially stressful circumstances. Meanwhile, it should not be neglected that out-migration could have some beneficial effects on emotional health, by reconstructing local gender relations and increasing left-behind women's autonomy and decision-making power . This change may create emotional gains and mitigate the detrimental psychosocial consequences of emigration. Earlier empirical research has placed much emphasis on the health of migrants and depicts migration as a stressful process for migrants . Only recently have researchers begun to examine the health impact of migration for families left behind. Most existing work has focused on children . There is limited but growing evidence concerning the health impact on the adult population. Kuhn shows a strong positive impact of out-migration on physical functioning and survival in rural Bangladesh. Roy and Nangia find a negative impact of migration on sexually transmitted diseases in rural India among women, which is speculated to be a result of reduced support when their husbands are away. Gibson et al. demonstrate that those left behind are more likely to report poor general health in New Zealand. --- The present study The present research examines the psychosocial consequences of out-migration for adults, focusing on the larger-scale internal out-migration in the study setting. For the reasons discussed above, I expect out-migration to be a stressful experience among people left behind and thus to have an adverse effect on psychosocial health. These health effects are likely to exist for both spouses and parents left behind. I further assess the differential effects by gender and the moderating role of social support. First, due to the differences in gender roles, people likely respond to stress and distress with sex-typical outcomes. Women left behind may be more likely than men to exhibit internalized affective reactions such as depression. In addition, in many developing societies, men assume greater obligations to maintain their families, which may put a greater level of stress on them. Also, women may use social support more effectively because seeking help is more congruent with the traditional feminine gender role. Consequently, men may be more susceptible to stress induced by out-migration. Given the stress-buffering role of social support, the level of disruption due to out-migration depends on the amount of social support one receives. In many migrant-sending areas, kinship linkages, usually reflected in co-residence of extended family members, may influence an individual's adaptation to migration. Extended families often fill the roles normally assigned to the absent migrants and offer emotional and instrumental support. Therefore, emigration may have a weaker effect on the psychosocial health when support from extended family is present. Also, the influence of out-migration may depend on the length of emigrant's absence. From a cumulative effects perspective, the longer the absence, the greater the distress experienced by family members left behind. Over time, however, households may develop better coping strategies as migration becomes ingrained in local life. The psychosocial impact of out-migration may thus increase over the early stages of migration and diminish over the long term. --- The study context Indonesia is the fourth most populous nation and has the largest Muslim population in the world. Over the past few decades, Indonesia has experienced rapid economic change. The country has also experienced concomitant improvements in access to health care and in common measures of health status, such as life expectancy and infant mortality rate . More recently, the country is undergoing an epidemiologic transition, with hypertension, diabetes, and mental disorders emerging as important health concerns . Indonesian cultures have consistently been documented as collectivistic . Many Indonesians, especially those in rural areas, live in extended families that adopt the strong tradition of exchange and provision of goods and assistance. Children also have a strong sense of obligation toward their parents. Although over time there has been a shift from extended to nuclear family living arrangements, the cultural emphasis on collectivism has continued . It is also worth noting that the kinship system in Indonesia is predominately bilateral; this social structure supports a relatively high status for women . Urbanization has occurred at a rapid pace in Indonesia, exceeding that of many other developing countries . The 2000 census reports that one in ten Indonesians can be classified as migrants . The internal migrant stream is largely characterized by economically motivated migration. Indonesia also is one of the world's major sources of international migrant workers into Southeast Asian countries and the Middle East, sending a total of around 2.5 million immigrants . But the strong economic growth makes domestic migration destinations equally attractive and the internal migrant stream is considerably larger than that of international migration . Migrants often maintain strong ties with their families and return periodically to their home areas. As in many other settings, Indonesian migrants remit a considerable proportion of their income to their families . In the late 1990s, 25% of rural households in the country relied on labor migration and remittances . --- Data and methods --- Data The data used come from the Indonesia Family Life Survey , a national representative longitudinal survey of individuals, households, and communities. The first round of data was collected in 1993 using multi-stage probability sampling and included interviews with 7,224 households and 22,347 individuals. In 1997, IFLS2 was conducted to reinterview all IFLS1 households and respondents . Considerable efforts were made to minimize attrition. Following the practice of IFLS2, IFLS3 and IFLS4, which were conducted in 2000 and 2007, successfully interviewed over 90% of the households in previous waves . The high follow-up rate substantially reduces data concerns that can arise from selective attrition. The IFLS collected a broad array of demographic, socioeconomic, and health information. Much of the information was repeated across waves. Importantly, the IFLS contains a detailed household roster, migration histories, and a wide range of health indicators. The household roster includes information on whether a member who lived in the household in previous waves had moved out, and, if so, the main reason for their migration. In the migration histories, information on place of residence at age 12, each trip longer than six months since age 12, and the date and purpose of the trips were gathered. Starting in IFLS2, measurements of blood pressure were included. IFLS3 and 4 further included questions measuring depressive symptoms. For these reasons, this study uses IFLS2-IFLS4. --- Variables As measures of psychosocial health, I used hypertension and depressive symptoms . Three readings of blood pressure were taken to improve accuracy and the average was used in the analysis. I created a dichotomous indicator using the WHO cutoffs: a systolic blood pressure of at least 140 or a diastolic blood pressure of at least 90. The IFLS3 and 4 included a short screening survey for psychological distress. One question was designed to measure depressive symptoms: "Have you experienced sadness in the last four weeks?" This measure has been shown to identify people with depressive affect reasonably well in Indonesia and in other settings . I constructed a dichotomous variable of depressive symptoms, coded 1 if the respondent had often or sometimes experienced the symptom. A better measure of depressive symptoms, a short version of the Center for Epidemiologic Studies Depression Scale , was added in IFLS4. CES-D is a widely used across many societies . I first constructed a continuous scale by adding all 10 inventories . This measure ranges from 0 to 30, with higher scores representing more depression. I then applied the standard cutoff of 16 to create a dichotomous indicator . The present study focuses on the dichotomized health measures because they are based on clinically meaningful thresholds. It takes a conservative view that only values in the abnormal range of the health outcomes should be treated as possible cases of disorder. I also carried out sensitivity analysis examining the continuous health measures. The major predictor, household out-migration status, was constructed from the rosters and migration histories. The rosters asked whether each individual recorded since the first wave lived in the household and, if not, the date of and main reason for moving out. A member is considered a labor migrant if he or she was absent from the original household and was reported to have moved out for work-related reasons. A household was defined as an emigrant household if it sent out at least one labor migrant. To separate the immediate and longer-term effects of out-migration, a discrete measure of length of out-migration was constructed to distinguish nonmigrant households, households with recent labor migrants , households with medium-term migrants , and those with longer-term migrants , based on the distribution of length of out-migration. Other covariates include standard demographic and socioeconomic characteristics of the individual and household: age , gender, years of schooling, current work status, logged per capita annual household income, marital status, family structure, household size, and female head of household. The analytic sample is rural adults who never moved across the selected waves. The strategy is to contrast eligible rural adults living in emigrant households with those in nonmigrant households. By definition, I eliminated people who ever moved during the study period . This is to avoid confounding the effect of out-migration with that of a person's own migration experience. I used age 65, the life expectancy in the late 1990s, as the upper age limit to avoid bias due to differential mortality at older ages. I focused on rural households, which make up over 50% of the data and represent over 80% of the emigrant households. --- Methods The effect of migration may be subject to unobserved heterogeneity-unmeasured factors influencing migration decisions that may also affect the health status of household members. This study exploited longitudinal data to obtain fixed-effect estimates. The FE approach compares the same individuals over successive waves, which effectively adjusts for time-invariant heterogeneity at various levels. Although there may be time-varying unobserved factors at play, this assumption may not be seriously violated because many important unobserved factors are either past events or are highly heritable . In each year: where p it is the probability that an individual i experiences a health problem at year t; Mig it is the measure of household out-migration status; X it is a column vector of the explanatory variables; μ t is an intercept; and α i represents time-invariant individual unobserved heterogeneity. The basic idea is to cancel out α i using conditional maximum likelihood , by conditioning the likelihood function on the total number of events observed for each person . The interpretation of the FE models is similar to logistic regressions. A caveat to the FE approach is that time-invariant factors cannot be explicitly modeled . Also, when the outcome is dichotomous, the method deletes individuals without outcome variations over time. While this can lead to reduced statistical power, it is not a major concern because there is generally sufficient variation in the sample. To increase the sample size for FE modeling, I employed all three waves to retain respondents with at least two observations, which captures any change over the 10year period. This procedure preserved close to 50% of the cases. I also included interaction terms between province of residence and survey year to account for macroeconomic shocks and province-level contextual effects. The analysis of depression resorted to the lagged dependent variable method, because the measure based on CES-D is available only in 2007. This approach models depressive symptoms in 2007 as a function of household out-migration status and other covariates while controlling for prior depressive symptoms in 2000. Similar to the FE method, LDV helps address possible selection bias to the extent preexisting differences are captured in the lagged measure. Because CES-D was not measured in 2000, I included a lag measure of depressive symptoms measured by the single-item question asking about sadness. I also conducted sensitivity analysis using a FE model with sadness as the outcome. In all analysis described above, the Huber-White robust estimator was used to correct the standard errors for clustering of individuals within households . Finally, it should be acknowledged that the approaches described above may not adequately adjust for other aspects of potential bias, such as selection of return migration and of the situation where all household members moved . I conducted sensitivity analysis to evaluate the robustness of the results to return migration and discussed the implications of the other aspect of selection for the results. --- Results --- Descriptive statistics Table 1 presents descriptive statistics for IFLS2-IFLS4. In 1997, 16% of rural households had sent someone out for work. This fraction increased almost 8% between 1997 and 2000 and over 10% between 2000 and 2007. The measure does a reasonably good job of capturing households with emigrants, as the estimate is close to those of other studies . When breaking down the overall change by different types of transitions, the majority of the transitions are accounted for by moving from nonmigrant to emigrant households. Additional analysis suggests that many of the left behind were parents and spouses of migrants. With respect to health, in 1997, 27% of rural adults had high blood pressure. It increased to 29% by 2000 and 34% by 2007, partly owing to the aging of the population. Of note is the substantial increase in the fraction of people who experienced some depressive symptoms, reaching around 30% by 2007. The fraction with severe depressive symptoms was close to 10% in 2007. In the analytic sample, the quantity of missing information is relatively small. The final sample is based on complete cases, after deleting 7.5% of cases with any missing data. The individual attrition between 1997 and 2000 was about 18% and between 2000 and 2007 was 29%. Since a large fraction of the people lost to follow-up were migrants and thus would not qualify for inclusion , the concerns due to selective attrition are largely reduced. --- Household out-migration and health Table 2 shows that household out-migration is a crucial determinant of psychosocial health compared to most other explanatory variables in the models. Adults in migrant households seem to be more vulnerable to high blood pressure and depressive symptoms. Put in statistical terms, the odds of experiencing hypertension increase by about 37% and the odds of having depressive symptoms almost double for members of households with emigrants as compared to those in nonmigrant households. The analysis with sadness as the outcome yields similar findings, which suggests that the results are relatively robust to different measures of depressive symptoms. Since a binary change in health indicates a substantial state transition, these results can be taken as evidence of strong impacts of out-migration. With respect to hypertension, the hypothesis is that increased physical and emotional burden is manifested in heightened stress, which in turn reveals in cardiovascular disorders such as hypertension. To strengthen this claim, other important risk factors for hypertension, smoking and overweight , were included. The large and significant impact of out-migration after adjusting for overweight and smoking suggests that factors other than unhealthy lifestyles, most likely stress, tend to play a crucial role in the disease process. This study further assesses the possibility that improved economics due to remittances may be associated with changes in consumption patterns, leading to overweight that subsequently contributes to high blood pressure. This possibility is not supported by the data, as household out-migration is not significantly related to overweight . Several additional analyses were conducted to evaluate the robustness of the results to several potential sources of bias. First, the out-migration measure may have underestimated circular migration and recent return migrants. I incorporated information from the migration histories to create a broader measure that includes circular and recent return migrants if the person reported having moved back to the current household within six months of the interview and having previously taken at least one other trip for work-related reasons. The analyses with and without circular and recent return migrants give highly consistent results and differ by only a small number of households . I also performed analysis that excludes individuals in households with return migrants across waves because migrants may return for health problems of other family members. This procedure does not seem to change the main story . Both analyses suggest that return migration is not likely to seriously bias the results. In addition, because several independent variables included in the models may be endogenous to out-migration, I retained only basic demographic variables that are unlikely to be affected by migration. The results are highly consistent with models including the full set of control variables in Table 2. Furthermore, I studied the continuous measures of blood pressure and CES-D as the dependent variables. This analysis is estimated for the full sample and leads to qualitatively similar results . This is especially true with respect to depressive symptoms that out-migration is associated with heightened psychological distress. When I examined diastolic and systolic blood pressure separately, the results appear to be largely driven by diastolic blood pressure, which is particularly sensitive to the levels of tension and physical activity and which is considered to be of more concern than a high systolic blood pressure that is often momentary . This result suggests that the detrimental health repercussions of family disruption due to outmigration can be quite serious, as they are especially evident in diastolic blood pressure. Lastly, for the models of hypertension and CES-D in Table 2, I performed corresponding linear FE and LDV regression models, which is expected to yield more efficient estimates . This analysis also provides an estimate of marginal effects , which is more straightforward to interpret and can be understood as the probability of the health deficit occurring. Whereas the magnitude of the coefficients is considerably smaller, the results tell a similar story: people left behind tend to experience a 3% and 5% increase, respectively, in the probability of developing hypertension and depression. --- Variations by relationship, social support, and gender I specifically examine the effects by major categories of relationships between migrants and people left behind, namely spouses left behind and parents left behind by adult children . The results show that parents left behind are especially likely to develop hypertension and depressive symptoms when a primary household labor and caregiver migrated for work. This is also likely to be attributable to the wear and tear on the body with age, which tends to increase the responsiveness to life stressors. By contrast, for spouses left behind, the health costs exist largely with respect to depressive symptoms, and to a lesser extent hypertension. This may be because many spouses left behind are women and that the different gender roles suggest varying vulnerabilities for men and women. This speculation seemed to be supported when I stratified the analysis by gender. But these relation-specific analyses should be interpreted with caution because the procedures used to derive the relationships between migrants and left-behind household members may introduce considerable error and lead to smaller sample sizes. With respect to gender differences, women have a higher risk of experiencing depressive symptoms when separating from spouses than men. This may be explained by gender norm expectations that are linked to the exhibition of mental health problems. The other possibility-that migration transforms gender relations and leads to improved psychological well-being for women-is not supported by the data, as the relationship between out-migration and the decision-making power of women is negligible. This is not surprising in Indonesia because a bilateral rather than patrilineal kinship system has predominated and women in general hold relatively high status. As for stress-induced physical health, men seem to be more vulnerable to hypertension. This is partly because men tend to assume greater work obligations and the well-being of men is more closely tied to work-related stress. Men thus tend to face greater pressure when experiencing household labor shortages. In addition, women are more likely to mobilize social support from kin and co-villagers than men. The data lend some support to these conjectures: men left behind tend to work eight hours more per week than women left behind . Also, in emigrant households, women were more likely to live with extended kin than men . Regarding the potential buffering role of social support, I find that the negative consequences of out-migration for hypertension become smaller in extended families, but the difference lacks statistical significance. The effect on depressive symptoms is reduced by over half for individuals in extended family arrangements. These findings provide some evidence that additional social support from extended families helps compensate for the disruption associated with out-migration, especially in the form of emotional support. With respect to the length of out-migration, results show that the detrimental psychosocial effects are not immediate but tend to accumulate over time. Within the first year of out-migration, there is little impact on physical health. The effect on emotional health is more evident, suggesting that the distress of separation tends to be more immediately felt than stress induced by labor shortage. But this effect is still relatively small. With extended periods of absence , we see a greater impact of out-migration as prolonged stressors lead to chronic strains. Over a longer period of time, the speculation is that households may develop better coping strategies and remittances may come in to compensate for some of the deleterious effects. However, the data provide insufficient evidence. The pace of recovery seems to be faster for psychological health than for physical health. This may in part be because emotional health is generally more responsive to change than physical health. Also, over time as people age and start to experience diminishing physiological functioning, they are less likely to recover from abnormal blood pressure. More work is needed to definitively pin down the longer-term effect of out-migration. --- Conclusions and discussion This research links several strands of research on family, the stress process, and social support to examine the psychosocial consequences of household out-migration. It adds to the understanding of the role of migration for individual well-being from the perspective of sending areas. The findings suggest that the influence of out-migration should be understood as a result of not only a socioeconomic process associated with remittances, but also a psychosocial process due to family disruption and reduced social support. In the Indonesian setting, adults left behind by internal migrants tend to be more prone to stress-related health conditions and psychological distress. These findings are consistent when parents and spouses left behind are examined separately. The results also demonstrate the buffering role of social support from extended families and the moderating role of gender. These findings add a demonstrative case to the literature on family dynamics and well-being by documenting out-migration as a form of family disruption, both similar and different to other stressful family transitions. Examining the case of migration is crucial, as in less developed settings the majority of split families arises from labor migration rather than from marital dissolution. In addition, whereas most studies on the health consequences of the stress process concentrate on Western societies and the immigrant population, this study offers important insights in a resource-constrained setting and in the context of internal migration. The use of high quality panel data permits various sensitivity analyses to reduce potential biases due to endogeneity, data quality, and measurement. Although consistent impacts of out-migration are shown across models, these analyses cannot completely rule out all possible sources of bias, in particular time-varying factors associated with both household out-migration and depression. In addition, several other limitations warrant discussion. First, it would be informative to explicitly examine the moderating role of remittances as well as the frequency of contact between migrants and families. This is because remittances may help mitigate the excessive distress on families by reducing the demand for subsistence labor, and may even generate nonpecuniary psychological gains due to improved economic status and standards of living. Improved resources from remittances may also enhance the physical health status of people left behind, which has been shown to lead to better psychological well-being. The analysis of depressive symptoms by length of emigration lends some modest support to these speculations. It shows that over time , the psychological costs of emigration are reduced and become marginally significant. Ideally, the analysis would compare migrant households with and without remittances, but the data do not supply such information. The study also cannot adequately capture migrant households where everyone has moved as they may be different from households where some members are left behind. In addition, the data are limited with respect to direct measures of stress and social support. Some findings regarding the stress process thus have to rely on relatively indirect inference. Despite these limitations, this study has yielded some useful observations about the health consequences of out-migration. These observations warrant particular attention in an age of global migration and of surging psychological and cardiovascular illness. The solution is not to impose stringent mobility restrictions, but to devise effective programs that can address the psychosocial costs of out-migration. Facilitating regular contacts between migrants and families left behind, such as those that lower the cost of communication and transportation services, is promising. This strategy may be especially effective because it helps reduce the distress encountered by both migrants and their families in the origin, as earlier research demonstrates strong mental health costs for migrants themselves . The story told here is part of a broader process of how migration has reshaped individual well-being. The Indonesian case provides lessons for other settings, not only other examples of internal migration but instances of international migration. For internal migrants, movement between work locales and home villages is relatively simple. The consequences for families left behind by international migrants, especially undocumented immigrants, may be substantially stronger, given the greater difficulties in visiting home and longer episodes of separation. Comparative work examining how the effect of emigration is contextualized within the larger socioecultural sphere would greatly advance our current understanding on the association between migration and health. --- Results from Sensitivity Analysis, IFLS 1997-2007

An extensive literature demonstrates various negative health consequences of family disruption in Western societies, which is largely due to marital dissolution. In developing settings, family disruption commonly arises in the context of labor out-migration. However, studies on household emigration often focus on the economic benefits from remittances, overlooking emigration as a source of stress and loss of social support. This research examines the psychosocial consequences of internal out-migration using longitudinal survey data collected in Indonesia between 1993 and 2007. Results demonstrate considerable psychosocial costs of out-migration, with adults left behind by migrants more susceptible to stress-related health impairments such as hypertension and to psychological distress such as depressive symptoms. These findings largely hold when specific relations are investigated, including spouses left behind and parents left behind by adult children. This study also finds some support for the stress-buffering role of social support from extended families and the differential psychosocial processes for men and women.

Introduction In social computing, systems aim at facilitating communication and cooperation among users who are either at the same location or at different locations. Social media summarises concepts and systems that aim at an active participation of users during an interaction, easy exchange of information, and sophisticated self-presentation [11]. Developing concepts for those systems is a challenging task and has been researched for more than two decades [12] [7]. They often have a strong influence on the structure and flow of the interaction in the group, as Schmidt [18, p. vii] explains: 'the development of computing technologies have from the very beginning been tightly interwoven with the development of cooperative work'. And he [18, p. vii] continues: 'our understanding of the coordinative practices, for which these coordination technologies are being developed, is quite deficient, leaving systems designers and software engineers to base their system designs on rudimentary technologies. The result is that these vitally important systems, though technically sound, typically are experienced as cumbersome, inefficient, rigid, crude'. Patterns have a long and successful tradition for drafting, for documenting, and for reusing the underlying concepts. Very prominently, Christopher Alexander has suggested and provided design patterns in architecture [2]. He introduced a pattern language to describe solutions that were repeatedly applied to reoccurring challenges in the design of buildings. In software engineering software design patterns have been successfully used for documenting and reusing knowledge and provided a 'way of supporting object-oriented design' [20, p. 422]. With respect to social computing, design patterns can document the knowledge and experience with developing cooperative technology. All these different types of design patterns provide valuable input for cooperative systems and social media. However, there are also limiting factors: software design patterns primarily help structuring software, and cooperative design patterns are primarily based on the analysis of existing cooperative systems or on some specific ethnographic studies. Therefore, the gap is that the complex task of making both types of patterns compatible is in the hands of software designers and developers. In this paper we build on the history of patterns and present overarching design patterns for social computing systems. For this purpose we leverage on the works of Erving Goffman who studied social interaction among humans and their use of their technical environment for several decades and derived a framework for social interaction. He uses a metaphor of a performance where everybody is an actor that present her-or himself and acts with others. In the next section we provide a background of patterns. We then introduce the framework of social interaction of Erving Goffman. We discuss how this framework informs the design of cooperative systems and we derive design patterns for cooperative systems that are modelled in a unified modelling language format for software designers and developers. Finally, we summarise our contribution. --- Background of Design Pattern Christopher Alexander et al. [3] were the first to systematically distil patterns from reoccurring solutions to reoccurring problems. In the domain of architecture they identified a language of connected patterns for designing buildings. In this section we introduce patterns related to the design of software in general as well as for cooperative systems and social media in particular that build of Alexander et al. --- Software Design Patterns Software designers and developers widely use software design patterns. Gamma, Helm, Johnson and Vlissides [5] suggested the most notable pattern language for object-oriented software development. They characterise a pattern as a composition of a problem that during the development frequently occurs, a principal solution to the problem, and consequences from applying the solution. Their pattern language includes 23 patterns for classes and objects in three categories: creational, structural, and behavioural patterns [21]. Cooperative systems and social media use network-based and distributed software architectures in the background. POSA2 offers a rather technical pattern language addressing the challenges of distributed software architectures especially in the context of object-oriented middleware such as CORBA, COM+, or Jini [17]. It has four categories representing the main challenges of object-oriented middleware: 'Service Access and Configuration', 'Event Handling', 'Concurrency', and 'Synchronisation'. The description of patterns is extensive and contains precise design implication for the named middleware along verbose source code examples. While software design patterns are substantial for sustainable software development, they still leave the burden of the complexity of social interactions to software designers and developers of cooperative systems. --- Design Patterns for Cooperative Systems and Social Media Design patterns for cooperative systems and social media typically focus on human behaviour and interaction. We describe patterns that support designers and developers of cooperative systems and social media. A pattern language for computer-mediated interaction condenses features and properties of existing cooperative systems [19]. It has three categories: 'community support', 'group support', and 'base technology'. The variety of patterns reaches from simple ones to complex ones . The description of patterns is very detailed and considers caveats as well as implications for security. Specific patterns for privacy and sharing provide solutions to problems concerning the quality of use of cooperative systems [4]. They result from field studies, notes, and design sketches that were translated into three patterns: The 'workspace with privacy gradient', the 'combination of personal and shared devices', and the 'drop connector'. Descriptive patterns have been suggested to allows a better facilitation of the communication in interdisciplinary design teams during the development process [13,14]. They are comprehensive and express 'generally recurrent phenomena' extracted from ethnographic studies at workplaces. The resulting descriptive pattern language consists of six patterns: 'multiple representations of information', 'artefact as an audit trail', 'use of a public artefact', 'accounting for an unseen artefact', 'working with interruptions', and 'forms of co-located teamwork'. Their patterns are extracted from fieldwork results using two types of properties: 'spatially-oriented features' and 'work-oriented features'. Their patterns can be extended with a 'vignette', which describes real examples as special use cases and provide further design implications. Despite the fact that patterns for cooperative systems and social media provide detailed insights into practices and requirements of users working together, they mostly lack the dynamic notion of such systems, where users can take advantage of a throughout personalisation of their environment. --- Goffman's Framework of Social Interaction We introduce the background and major concepts of Goffman's framework of social interaction that are relevant for designers and developers of cooperative systems. --- Fig. 1. Major Concepts in Goffman's framework of social interaction Goffman [6] studied social interaction among humans for several decades and developed a conceptual framework of social interaction among humans in face-to-face situations. It is based on his own observations, on observations of other researchers, and on informal sources. In the following, we describe Goffman's framework in three categories: participants, regions, and performance . --- --- Regions Regions are spatial arrangements used for performances and include specific media for communication as well as boundaries for perception. Goffman names three types of regions: stage, backstage, and outside. A stage provides a setting for the actual performance and is embroidered with decorative properties . It supports performers in fostering a situation. Both the performers, as well as the audience can access the stage, having different perspectives. The backstage is a region that performers can access to prepare and evaluate their performance. Also team members suspend backstage. The audience cannot access the backstage. The outside region describes the third type that is neither stage nor backstage. Although it will be excluded from a performance, performers will prepare and use a dedicated front for the outside . --- Performance For Goffman a performance means social interaction as a finite cycle of expressions to define a situation and of responses as feedback of validity. A performance takes place in a region of type stage. For a performance, each performer prepares a set of fronts, which represents her towards the audience. A front unites material and immaterial parts. Sign equipment is a front's material part and denotes to all properties required to give a convincing performance. The personal front is a front's immaterial part and denotes to certain types of behaviour of a performer . It combines 'appearance' and 'manner' . Characters make the appearance of performers on the stage. A character-as a figure-is composed of a 'front', which is specifically adapted to the audience and performance. In a performance team, the team as whole has a united front and each member has a character with an associated front to invoke during staging. During a performance a character plays routines to convey acceptable and to conceal inacceptable expressions. In a performance team multiple characters will follow this behaviour. Expressions are information that is communicated by a character using 'signvehicles' . There are wanted expressions that are acceptable and foster a situation as a valid projection of reality, and unwanted expressions that are inacceptable and inappropriate for a given performance in front of a particular audience. In order to manifest a performance that is coherent, a performer strives to communicate expressions consistently through their characters towards an audience. Thus a performer's character endeavours to conceal unwanted expressions. Responses are all kinds of feedback. An audience continuously verifies the performance according to the defined situation and the overall reality as well as to the front of the character. It responds the result to the performer. In order to manifest a valid performance, performer and audience agree on three principal constructs that prevent a false or doubtful projection of reality based on contradictive expressions or discrediting actions: The 'Working Consensus' is an agreement on the definition of the situation and describes a temporal value system among all participants. The 'Reciprocity' means that performers guise their characters to act according to the situation and that the audience responds to performance according to the situation . The 'Interactional Modus Vivendi' describes that an individual in the audience only responds to expressions that are important for the individual; the individual in the audience remains silent in things that are only important to others. Goffman describes additional participants. For instance, the team support, which is one of the following: colleagues that constitute the weak audience, training specialists that build up a desirable performance, service specialists that maintain a performance, confidants that listen to a performer's sins, or renegades that preserve a idealistic moral stand that a performer or team failed to keep. Goffman also defines outsiders as being neither performers nor audience having little or no knowledge of the performance. They can access the type outside region. --- Informing the Design of Social Computing In this section we transform Goffman's framework into design patterns. We used three steps. We first identified key statements of Goffman's framework concerning structural aspects and dynamic aspects . In a second step, we augmented these aspects with literature reviews and lessons learned from conceptualising and developing cooperative systems-especially concerning the transition from physically co-present humans to virtually co-present humans . In a third step, we iteratively derived four design patterns for cooperative systems and modelled them in the unified modelling language . --- Structure of Social Computing The structure of social computing systems refers to entities and their relations as essential ingredients. Our UML class diagrams emphasise entities involved, their compositions, and their dependencies. We use interfaces for modelling general entity behaviour that can be applied to a variety of instances. We use abstract classes for modelling entities that share implementations, and we use standard classes for modelling specific entities. The first structural pattern we introduce is the Social Entity Pattern . It describes the general setting of people involved in an interaction and their roles. The interface SocialEntities refers to humans that are explicitly included in an interaction. A social entity has general knowledge of the world and specific knowledge of particular domains. It relies on Routines as 'pre-established pattern of action […] which may be presented or played…' [6, p. 16]. It conveys information it likes to share with others, and conceals information it likes to hide from others. There are four classes implementing the interface SocialEntity: ActiveIndividuals, ActiveTeams, PassiveIndividuals, and PassiveTeams. --- Fig. 2. Social Entity Pattern as UML class diagram ActiveIndividuals refer to Goffman's performers and are instances of classes with a repertoire of Faces. They anticipate the behaviours of others and select as well as fit their faces towards them. An ActiveTeam consists of at least two ActiveIndividuals: which refers to 'any set of individuals who cooperate in staging a single routine' and '…an emergent team impression arises which can conveniently be treated as a fact in its own right…' [6, p. 79]. Teams have an overall goal. As noted above members of a team can have in individual activity or a shared activity. Since the delegation of an ActiveTeam's members can vary from team to team, it is the responsibility of extended classes to implement that behaviour. An ActiveIndividual and ActiveTeam can rely on their Support . PassiveIndividuals as an abstract class implements the interface social entity with the ability to observe an action. Further implementations of such passive individuals are the PassiveTeam, which refers to the audience that participates in the interaction. About the relationship of active individuals and passive individuals Goffman states: '…the part one individual plays is tailored to the parts played by the others present, and yet these others also constitute the audience' [6, p. xi]. Parallel to the team above a PassiveTeam is an aggregation of PassiveIndividuals; Goffman writes: 'There will be a team of persons whose activity … in conjunction with available props will constitute the scene from which the performed character's self will emerge, and another team, the audience.' [6, p. 253]. In the pattern a Face class lays out the foundation for a distinct configuration of an active individual or team as a prototype to be applied in an interaction. Our notion of a face refers to Goffman's front; it is the 'part of the individual's performance which regularly functions in a general and fixed fashion to define the situation for those who observe the performance' [6, p. 22]. An ActiveIndividual can have multiple faces as a repository of communication methods and properties towards passive individuals. Since, in cooperative systems simultaneous interactions are likely, it is important to note that an active individual may have multiple active faces at a time . A Character is a specific configuration of a face. When instantiated in an interaction, an ActiveIndividual selects and transforms a face into a Character containing information and dissemination methods: 'When a participant conveys something during interaction, we expect him to communicate only through the lips of the character he has chosen to project' [6, p. 176]. In our pattern the interface Artefact refers to work-related and leisure-related objects . In contrast, Goffman narrows the performance down to interacting individuals or teams; for Goffman external objects contribute to the overall expression of a situation as a setting: 'there is the setting, involving furniture, decor, physical layout, and other background items which supply the scenery and stage props for the spate of human action played out before, within, or upon it.' [6, p. 22]. However, in social computing often an Artefact is an essential part of an interaction. It relates to virtual or physical objects that can be created, edited, and deleted in the course of an interaction. In a routine, a composition of artefacts that can be involved; in social computing systems this is typically represented as collaborative editing or sharing. The Interaction refers to Goffman's performance. It is a composition of characters of one or more active and one or more passive individuals. It has three phases: in the preparation an active individual sets her role; in the execution a character acts towards passive individuals or a passive team; and in the finalisation an active individual collects responses from its interaction and uses the outcome for further refinements of its faces. A history as set of interactions is important in social computing systems for verifying information and deducing information . The second structural pattern is the Region Pattern . It maps Goffman's regions into a combination of Visibility and Locality that can be applied in Interactions. Goffman writes: 'A region may be defined as any place that is bounded to some degree by barriers to perception.' [6, p. 106]. As described above, Goffman distinguishes the regions stage, backstage, and outside. However, in our opinion, social computing systems require a more flexible representation that should allow for and contribute to in-between regions. --- Fig. 3. Region Pattern as UML class diagram The interface Visibility represents filters for types of information and dissemination methods to be applied to interaction with social entities. While active individuals and active teams can access a huge amount of information, passive individuals and passive teams can only access designated information. The interface Locality also refers to filters, but they provide methods as a boundary of real locations or virtual locations . Combining Visibility and Locality provides means for sophisticated configurations than the region types proposed by Goffman could cover. The combination reflects individual sharing preferences that also apply during a system's automatic inference of information . For instance, an interaction can span real and virtual locations at once while communication is still filtered. The filtering can be achieved by matching properties of CommunicationEntities towards the properties of passive entities involved into the interaction. Subsequently, we introduce the interfaces DirectSocialCommunication and MediatedSocialCommunication along their patterns. --- Dynamics of Social Computing The dynamics of social computing systems refers to the general communication behaviour of humans within the system. The two patterns focus on the interaction between an ActiveIndividual and a PassiveTeam as the execution of an interaction. We show two patterns as use cases in UML sequence diagrams. Each diagram shows the entities involved in the execution, and sequences of synchronous and asynchronous calls used in it . --- Fig. 4. Direct Social Communication Pattern as UML sequence diagram The first dynamic pattern is the Direct Social Interaction Pattern . It starts with the path an ActiveIndividual executed to setup its Face and Character and activates a PassiveIndividual-summarised as anticipate-call in the diagram. After that, an ActiveIndividual instantiates a Character object for direct social interaction. According to Goffman, faces are selected and adapted, rather than created; he writes: 'different routines may employ the same front, it is to be noted that a given social front tends to become institutionalised in terms of the abstract stereotyped expectations to which it gives rise, and tends to take on a meaning and stability apart from the specific tasks which happen at the time' [6, p. 27]. This manner of stereotypical selection and adaptation allows PassiveIndividuals to recognise familiarity between Characters of different ActiveIndividuals and thus simplifies the validation process. The Character object creates the DirectSocialCommunication object for delivering information. In the loop of direct social communication, a Character calls its associated Face to obtain valid and appropriate information. It then delegates this information to the DirectSocialCommunication object for further distribution in an Interaction. Goffman describes direct social interaction as a communication of 'signactivity'-the transmission of expressions towards the audience relying on 'signvehicles'. He distinguishes two 'radically different' types of communication: the given and the given-off [6, p. 2]. In this pattern DirectSocialCommunication refers to the type 'given'. It stands for communication in a narrow sense as it consists of verbal or written symbols . All social entities involved in an interaction are familiar with the encoding and decoding these symbols. The process of delivering DirectSocialCommunication occurs frequently in a loop and simultaneously during an interaction, the resulting calls are asynchronous ones. As described previously, a PassiveIndividual receives the information and matches its consistency. A PassiveIndividual responds accordingly concerning the information's inner validity as well as regarding previously received ones . An ActiveIndividual can emphasise information during sending DirectSocialCommunication as it can adapt a Face using the responses received-Goffman speaks of governable aspects [6, p. 7]. --- Fig. 5. Mediated Social Interaction Pattern as UML sequence diagram The second dynamic pattern we introduce is the Mediated Social Interaction Pattern . It reflects the process of accessing an artefact and distributing occurring information of accessing it towards the passive individuals. In cooperative systems and social media applications this type of information is typically used for providing awareness information to the users [10]. MediatedSocialCommunication refers to communication in a broader sense and is related to Goffman's 'given-off'. It consists of a range of behaviours that can hardly controlled or manipulated-Goffman writes of ungovernable aspects and is of 'more theatrical and contextual kind, the non-verbal, presumably unintentional kind, whether this communication be purposely engineered or not.' [6, p. 4]. When accessing an Artefact at least one separate object MediatedSocialCommunication is created automatically. The PassiveIndividual receives information and matches it with previous objects of the type DirectSocialCommunication and MediatedSocialCommunication and responds accordingly. --- Discussion and Conclusions In this paper, we have argued that designers and developers of social computing systems face complex design decisions. To support them, we identified key concepts of Goffman's framework and derived structural and dynamic UML patterns. Our study of Goffman's framework and the derived patterns relate to some findings of previous work on patterns-corroborating these findings. Our patterns bridge between the artefact-specific patterns of Martin et al. [13,14] and the collaborationspecific patterns of [19]. The Social Entity Pattern represents the typical behaviour of users frequently switching their hats between the two roles of an active individual and a passive individual. The faces they rely on during their performances are diverse in terms of contained information, actions, and reactions. System should address this need for diversity by providing a repository of faces the users can chose from and evolve their characters upon. Yet, our pattern reaches beyond the existing ones, as it allows multiple, persistent, temporal and spatial active characters. The Region Pattern addresses the requirement of diverse spaces for preparing, sharing, and acting. Social computing systems should provide these spaces, as users need them for their performance. On the one hand users prepare the interaction using more 'technical standards' in a 'backstage' region where 'the suppressed facts make an appearance.' [6, p. 112]. On the other hand, users interact on 'stage' type regions using more 'expressive standards'. Providing stability of locality and visibility in this pattern is important for preventing users of unmeant disclosures that Goffman calls 'some major forms of performance disruption-unmeant gestures, inopportune intrusions, faux pas, and scenes.' and 'When these flusterings or symptoms of embarrassment become perceived, the reality that is supported by the performance is likely to be further jeopardised and weakened' [6, p. 212]. The Direct Social Interaction Pattern and Mediated Social Interaction Pattern cope with the performance itself and provide a model for Goffman's two communication types of 'given' and 'given-off'. Users require means of dramaturgical discipline-for instance, the anticipation of the passive individuals-to manage their impression validly. The patterns explicitly inform designers and software engineers of social computing systems to apply the Region Pattern in order to consider the hardly to governable type of communication . For future work the structural and the dynamic patterns should be applied in the design of social computing systems so their actual benefit for designers and developers in conceptualising and implementing can be measured in empirical studies. Furthermore, Goffman offers detailed descriptions of more social processes and best practices that may supply further patterns towards an extensive language of patterns for social computing systems.

Developing cooperative systems and social media requires taking complex decisions concerning the social interaction to be supported as well as the technical foundation. In this paper we build on the long and successful tradition of design patterns and the social framework of Erving Goffman. We present design patterns that address both challenges of social interaction and technical foundation-they provide input for software developers with respect to structuring software and to providing adequate support for the interaction of users with the environment and with each other.

Contextualising Social Housing --- Defining Social Housing? Social housing, although thought to be a common concept in countries around the world, is actually not available in all countries . The definition of social housing provided by the OECD explains that it is "Residential rental accommodation provided at sub-market prices and allocated according to specific rules" . Unfortunately, there are variables in the provision for social housing and who is able to have access to it. In fact, social housing provision is affected by different conditions and circumstances and varies from country to country . In Europe, the aftermath of World War II left people homeless and poor, which resulted in an increase of people on the poverty line . In light of this, there was an increase in government-based provisions, specifically in terms of social housing, which allowed for a noticeable shift in directing public funds . Before this shift, in a lot of countries, private companies, the Church and charitable institutions used to offer and manage social housing. The idea came about to tackle the issue of homelessness amongst the working poor. However, during the 1980s and 1990s, governments began to reduce their spending on public housing provisions, thus, public housing policies were no longer a priority. Instead, local authorities or private funders were left to deal with the issue . Vocal political groups at the time claimed that the sector was "inefficient, unresponsive, monopolistic and anachronistic" , as it did not aid the most vulnerable but rather the selected few. Published by IDEAS SPREAD Social housing at present is focused on the most vulnerable within society, however, this emphasis has caused concern since it has led to members being excluded, discriminated against and social housing spaces being labelled as a ghetto . Challenges and the risk of housing poverty are not being experienced just by the working class, however, those within the middle class have been prone to find themselves in difficulty as well and this is mainly due to the global economic crises and the transformations happening in society and its response to needs. As a result of this, the need for social housing has augmented, and as a consequence its waiting list . --- The Objectives of Social Housing "The main objective of social housing is to provide low-income households or vulnerable people with a decent, affordable and secure housing option" . These housing options, however, usually have the rent connected to one's income. On the other hand, tenure service providers and their eligibility, varies in each country . Nowadays, various other entities or private landlords, besides governments, are gradually offering more social housing services . Regardless, this shift in social housing provision, remains synched with national policies. Socio-economic positioning is used to influence one's eligibility, however, when vulnerabilities are taken into consideration, they are then at times re-evaluated regularly . The objective and aims of social housing is to ultimately provide housing which is secure and one that has a longerterm tenancy, ensuring stability, control over one's home and better rights . For children living and being raised in social housing, the intentional outcome is that it will help foster better access to education, wellbeing, health, jobs and the retaining of employment. However, this still remains debatable . --- The Legal Framework The United Nations observes that satisfactory housing provision is a right and the international law states that "it is the right to live somewhere in security, peace and dignity", being a fundamental human right . Strengthening and protecting housing rights is at the core of a number of international organisations . In terms of EU law, right to housing or housing assistance was recognised in 2009 when the EU Charter of Fundamental Rights was incorporated into EU law . Despite this legal recognition of housing, past EU social policy identified the relevance of the right to adequate housing provision and creative housing policy because of the movement in the single market . Member States of the EU are free to implement policies and interpretate the law as they see fit, however, the Treaty of Lisbon aids as a guide for the understanding of what is the right to housing and that adequate housing becomes the norm . Governments are not obliged to provide housing to all, regardless if the right to adequate housing is implemented in law . In 2020, it was estimated that globally 1.6 billion citizens were living in scant housing settings with the rate of homelessness rising drastically since 2010 . In Europe, the rise in homelessness, worsening affordability, and social and housing polarisation have been a cause of concern for public policy for the last decade . An estimated 3 million citizens in Europe, lack adequate housing . Regardless of the high numbers of people who are homeless, recognising the right to suitable accommodation has become pivotal . Living in inadequate housing puts poor people in a position of choice, having to choose to prioritise their health, what to eat, what kind of shelter to choose and any further requirements . Having a place people call home gives one a sense of self which then impacts one's wellbeing along with one's integration into society, in terms of being socio-economically stable . The notion of suitable housing automatically ensures the preservation of communal and cultural rights including other rights such as those which are political and those which are basic civil rights . --- Social Housing, its Impacts and How it Contributes to Social Mobility Social mobility is defined as "the movement in time of individuals, families, or other social units between positions of varying advantage in the system of social stratification of a society" . Social mobility helps with new opportunities specifically for those who come from underprivileged backgrounds. In fact, education and employment are seen as the ideal ways to alleviate poverty, but not just, as family income also plays its role . Social mobility contributes to a persons' wellbeing Published by IDEAS SPREAD who in turn can excel in society thus, society benefits from this too. The economy is improved, opportunities become more accessible and a person's life is improved allowing them to extend their horizons . A home helps individuals find some sort of wellbeing which correlates to their experience of home and what it means to them . As a home becomes an integral part of a person, they tend to form a bond with it . This effects the human psyche as it reinforces the notion that when people live in adequate housing this affects them positively and contributes to helping them with other factors in their life, be it social, cultural, economic and even psychological . However, there are those scholars who disagree with this perspective . Self-development is essential to human existence and a home provides physical attributes which contribute to this, such as space, safety and physical sufficiency . Regardless of whether one needs to be eligible to attain a place in social housing, being given the chance to live in such housing which is affordable, gives one the opportunity to be surrounded by people from the same or similar socio-economic background looking for a better life . Well-thought-out schemes contribute to having social housing spill over to social cohesion, as it gives people purpose again and the will to integrate with others . Individuals who are caught up without residence end up losing their self-worth and purpose, socially cohesive neighbourhoods are aids for people to show up for themselves, feel empowered and included . As a matter of fact, even those neighbourhoods that are slightly more disadvantaged than others, even with the little resources that they have tend to foster social integrity . Good planning and infrastructure in social housing will help with building a better community and allow for the residents to feel at home fostering a better community, an advancement in social mobility and sustainability in the long run . Social cohesion offers individuals stability and the sense of community which comes from social housing improves one's health which also directly affects social mobility . When a person is mentally sane and their overall wellbeing is improved, they are at a greater capacity of keeping a job, unlike others who are categorised as being in poor health . Poor health, many a times, has been found to exclude people from work, finding work or holding down a job which will continue to deter their health . The stability afforded to tenants who live in social housing has helped improve their health and as a result of this any health costs have been kept to a minimum . Depression and poor health are known to be more present amongst those who do not live in a stable home , whilst homelessness brings with it increased food insecurity, a shortened lifespan, a possible risk of premature death, worse health and widespread mental health challenges . Tenants could also experience a delay in accessing healthcare, increased levels of anxiety, relationship failures, loss of friendships, and access to other benefits . Therefore, when seeking stability and finding it, tenants' lives will be drastically improved as the positives then come to out-way the negatives. Youngsters experience home instability differently to adults as this impacts their educational prospects and achievements, contributes to depression, alcoholism, risk of teenage pregnancy and drug abuse; all this upon health issues . Murky and risky housing, which means that there are issues with the structure of the building and hazardous problems are present, has contributed to sickness such as tension and nervousness. Certain health issues like the incidence of dust mites have contributed to skin and other diseases, respiratory problems, and stress due to overcrowding. Unkept housing can cause physical problems due to accidents such as slips, and falls and accessibility issues for individuals who have a physical disability . The environment in which social housing is located in can have positive or negative associations. At times, negative associations such as crime levels in the area can cause social exclusion which contribute to stress and mental health conditions . Social and economic inequality have a direct effect on one's health and disrupts social integration .In actual fact , the area is associated with negative aspects like stigma, segregation, crime and social exclusion, then there is the possibility of long-term and long-lasting effects on tenants' well-being . Published by IDEAS SPREAD The category of tenants who live in social housing has changed over the years, seeing a shift specifically in Europe, going from those who were from vulnerable social groups directly contributing to the neighbourhoods' deprivation and segregation , to social housing being taken up more by pensioners and single-parent families. These families are mainly ethnic minorities and immigrants . The social economic demographic of these families is specific, as the majority of them tend to be low-income families, who's economic contribution is less than average . Due to this, they contribute to the area's socio-spatial segregation which can lead to social deprivation, namely liability , crime and community sabotage . Neighbourhood segregation as explained by Margery Turner can also be affected by ethnicity or colour, further decreasing opportunities for low-income families, therefore, their social mobility and access to better opportunities which can give them access to safer amenities is not made possible. Unfortunately, these neighbourhoods can be the breeding ground for concentrated poverty which is an area where socio-economic deprivation is found at an alarming rate in an area of high poverty . This specific kind of poverty accelerates the effects of poverty which continues to lower one's expectations and aspirations and it becomes increasingly difficult to leave the cycle of poverty . The Cambridge Dictionary defines being poor as "having little money and/or few possessions" which categorises someone as living on the poverty line or in the cycle of poverty. Therefore, when a person is living in poverty they are identified as being extremely poor . Lumen explains that the source of poverty is also in turn the consequences of poverty, hence why cycles of poverty are created. Social housing has a direct influence on what drives poverty, therefore, inhabitants of such housing find it difficult to break the cycle . Some, due to this believe that social housing can deter social mobility and segregation and not essentially a factor that creates these . Since social housing is a concentration of lived spaces, it is directly an effect of its environment which puts tenants at a disadvantage, essentially also contributing to the generational impediment of social mobility . Due to this, social problems will continue to be sustained . In fact, the allocation of tenants to housing is what effects the characterisation of the space, hence there could be high crime rates in the area because of the tenants or the tenants themselves could be the ones who work together to avert criminal behaviour and decrease criminal behaviour . The management of social housing helps the housing block to function properly in terms that there is mutual support and respect amongst the tenants who also come together to keep the area in order and manage any disruptions . Tenants are key players in keeping the social unit clean which will counteract problems such as vandalism and instill a sense of purpose . If overall support is given to tenants, not just socially but also through their landlord, then this will counteract any negative effects, such as dependency and inequalities, and foster participation in education and work . Housing affordability is a fundamental pillar in determining social mobility as this usually takes up a large share of nonrefundable income of tenants . 30% is the typical percentage for housing costs , and then the remaining income is distributed amongst other provisions which one may need. When tenants are over paying on housing, they then may divest themselves of other requirements which will lead to feelings of deprivation having an impact on their social status . Purchasing a residence or renting a habitat within the commercial rental market is costly and there have been situations where house prices have escalated radically within a few years, making it near impossible to own a house . Such housing becomes difficult to obtain especially for those low-income families, as their disposable income would not be enough to pay for it. The effects of such pressure to obtain this type of housing can cause health problems, emotional problems, mental health issues, increase in stress levels, disengagement at school for children to name just a few. On the other hand, when tenants have the opportunity to access affordable housing, they have a better chance of staying there longer, specific problems may start to reduce and the foundation of a good home will improve their social mobility as a family . One of the positive outcomes of access to good social housing is that for those people experiencing poverty there is a better chance of them slowly moving away from poverty as their income becomes more accessible for necessities which improve their self-sufficiency in society . Published by IDEAS SPREAD Employment raises one's standard of living and aids with helping one build themself up within a community , fostering social mobility. It helps adults especially, find their way out of poverty . 'Best & Shimili state: "Employment is recognised to be the single biggest factor in determining not only adults' life chances, but also those of children and grandchildren"' , aiding intergenerational change. Stable employment brings with it the provision of a secure income, however, not only, it allows one to preserve relationships and networks, focus on keeping a regular schedule, take on training opportunities, improved self-worth and the ability to assess possible future difficulty related to income . Furthermore, if someone is unemployed then this will impact them and their families keeping them in poverty allowing for mental strain to develop. This will also affect any children living within the family causing a disruption to their progress, schooling and future employment . Social housing provides stability and that stability alleviates one burden for those in positions of vulnerability and helps tenants find work allowing room for better success. It does not, of course guarantee employment, however, research has shown that the stability from housing has helped tenants look for and keep a job . Location plays a vital role, especially if social housing is located close to potential employment opportunities, for example, in city centres were rent is private, but the area has opportunities for low paid workers. However, many low paid workers due to their financial situation usually choose housing which is further away from work opportunities as they would be able to meet their renting needs. However, even though distance does not hinder one from working, the commute to work and not having as much time for leisure are factors affected by this distance . Fletcher et al , explains that individuals who live close to family and have a social network are not willing to move away for work reasons . Even though social housing and the stability that comes with it aids with helping individuals find jobs and maintain them, some have criticised it for encouraging tenants not to work and rely on social aids as many times tenants are usually unemployed . Hills explains that some tenants choose to remain unemployed due to the bureaucracy and difficulties they face with the tax and benefits system. Youths who are surrounded by adults who are jobless, are at times influenced by the resistance to work and this affects their future prospects of work . In fact, reports have concluded that it is not always about where one lives that deters one from working and remaining jobless, but it is usually the person's character and their unwillingness to work . Even though many tenants experience poor job quality and low pay apart from other issues, they end up being very heavily influenced by those around them and if they have a resistance for work then this may rub off on others. Jobseekers can also feel threatened to access the employment sector because of their lack of skills and experience. Like many consequences which are experienced by individuals living in social housing, joblessness can be intergenerational, since children or grandchildren of those who are unemployed could potentially be unemployed when they come to a working age . Children's futures are greatly affected by the area social housing is located in, as if they live in a poor area, then they will continue to experience high levels of deprivation and or segregation which affects their prospects . Amenities around the area such as safe playgrounds, open spaces, libraries and good schools also affect a child's social mobility and prospects at school . If the area experiences high rates of violence and problems such as addiction, children will find it challenging to interact and will become anti-social. The affects are different depending on age, for example, younger children can show challenging behaviour and truncated educational attainment whilst children who are older can show interest in violence, become emotionally detached and be at risk of forming or being groomed into gangs which will have problematic outcomes for their futures . Just like adults, children are affected deeply by their surroundings and their homes and it has been shown to affect their development and quality of life . Therefore, the quality of social housing and the overall environment will definitely determine the child's development and life chances . Researchers have found out that children do have better chances even assuming they habitat in civic housing, as it is the stability, the quality of the housing and the security that make the most difference Overall, there are benefits to social housing especially if the environment is a positive one as it can contribute to social mobility but itself alone does not cause much change . People are increasingly Published by IDEAS SPREAD impacted by harmful environments and for children this stunts their development . Residents within a housing unit are more likely to form supportive communities if the space they live in makes them feel like they belong . However, tenants with exposure to vulnerabilities or any social problems, having social housing will not solve these problems automatically but it will at least provide shelter . Taking an intersectional, multidisciplinary approach will allow tenants to benefit more from social housing and it will allow them to overcome other areas of difficulty in their lives such as education, and the opportunities to build new skills . --- Concluding Comments Social housing supports people who many times are living in poverty, and who may have varying vulnerabilities such as a disability, mental health issues, illnesses, and social issues. It is rental accommodation which can be afforded by people who struggle with making ends meet and those who lack a stable income. Social housing if looked after, can not only be called a home but a tool for those tenants who have diverse needs or vulnerabilities as it adds to a sense of self-assurance and finding stability. Social housing will become a foundation of support since tenure provides stability, security and peace of mind, and allow for tenants to have a better future. There are negative associations to it, as well as it is often blamed for contributing to tenants remaining dependant on social welfare, their aversion to find work, contribution to crime and addiction and fostering a breeding ground for social exclusion, it has been proven that social housing is not the main cause of these negative effects. Residualisation is the reason why these problems exist in an area since it is defined as an area where a large number of households are present showcasing great need of assistance due to lack of resources and opportunities. In order for social housing to be effective and contribute positively its tenants, it has to truly cater for people from all walks of life. To overcome residualisation and social exclusion, best practises must be set in place such as children's spaces and play areas, green spaces for all, support services which can facilitate job opportunities, training programmes, and guidance with finances, to name a few. When community services are provided and support to tenants is evidently clear, the social housing environment can foster social cohesion, creating safer spaces for tenants. Having a place where to come together and live positively with good amenities will allow tenants to build connections which will have a huge impact on their mental wellbeing which will help their social mobility and livability within a community. Being one of the most important areas in welfare, social housing cannot function in isolation. Policy makers need to ensure that other services are in place in order for tenants to thrive within their community. Will this be enough to get detached from the cycle of poverty. Overall, it is imperative that policy makers always keep the service users at the heart of their policy planning so that they can truly safeguard tenants' wellbeing by ensuring that social housing encourages tenants' social mobility and not hinder it.

The objective of the paper is to conceptualise how social housing impacts the life of tenants and how it affects their social mobility. Firstly, the paper looks at defining social housing to help us get a better understanding of the subject matter. It highlights that social housing is a residential rental space provided at specific rates to be afforded by those who are considered vulnerable in society. It is a tool to improve the lives of those in vulnerable positions due to poverty, disability and mental health issues to mention just a few. Children who live and grow up in stable social housing will have a better chance at accessing education, their wellbeing will be more positive, increased job prospects and a better outlook on opportunities. Secondly, the paper addressed the impacts social housing has and its effects on social mobility for tenants. It discusses how income plays a vital role in obtaining social housing, how the wellbeing of tenants is essential in fostering a better self, how individuals who live in social housing can come together and form communities to deter crime, how living in safe social housing can have a positive impact positively effect on mental health, how it encourages tenants to find work and how it helps children strive for and manage to break the cycle of poverty. However, notwithstanding the positive, social housing is not perfect and if not taken care of and is not funded properly then it could be a detriment to tenants' social mobility.Desk research was carried out to identify an in-depth understanding of the topic. Apart from this, key experts from the Faculty for Social Wellbeing shared their expertise through various informal discussions.

Background Human Immunodeficiency Virus/Acquired Immunodeficiency Syndrome has negatively impacted and disrupted families. Family members' productivity and functions have also been negatively affected, especially the caregivers of adolescents living with HIV/AIDS. 1 For all the efforts to control increasing numbers of people living with HIV, 38.4 million people globally, were living with HIV at the end of 2021. 2 Among these, 36.7 million are adults, 1.75 million are adolescents aged 10-19 years and of these, 1 million adolescents were estimated to be girls while 750,000 were said to be boys. Also by 2021, 1,750,000 adolescents were living with HIV, 150.000 adolescents were newly infected with HIV and 32,000 adolescents were dying of AIDS related causes. [3][4][5] From the above statistics, we can conclude that HIV/AIDS remains one of the most serious public health problems, especially among the adolescents. This is because adolescence is a critical period of development. Erikson's theory of psychosocial development explains adolescence as a transition from childhood to adulthood where adolescents begin to feel confused and insecure about themselves. They begin to establish a sense of self and they end up experimenting different roles like early marriage and behaviors like drinking alcohol and engage in early sex. Such changed behavior highly exposes them to HIV and also increases its transmission to those who are negative. It also leads to lower percentage of adolescents testing and poor adherence to treatment. Lower percentage of adolescents testing and adhering poorly to treatment has raised the number of adolescents living with HIV to 82% of the 2.1 million people living with HIV globally. This has had both direct and indirect effects on the population and caregivers of ALHIV have been severely affected. Depending on the intensity of this caregiving, such involvement in caring has been observed to negatively affect their domestic economy, health, physical and psychological wellbeing of caregivers. This has caused feelings of fear, grief, suicidal thoughts, financial burdens and household dysfunctions among these care givers. These dysfunctions are highest in low and middle income countries where they do not have enough income to take care of the family members. 6 In these countries, families and in particular parents or guardians are the main source of home-based care for adolescents living with HIV, but are often overwhelmed and burdened by caring for the adolescents living with HIV who are in this critical period at the same time living with HIV 7,8 and would want to be supported. In the past decade, the global community and individual countries have made progress in meeting the needs of children living with HIV. Support is often provided through targeted programs. Unfortunately, this support does not meet the needs of the population and challenges that arise during the adolescence period 7 have not been addressed. They have only targeted adults and children living with HIV and their families. It is upon this background that we explored the experiences of caregivers of adolescents living with HIV at Mbarara City Health Centre IV which is found in South Western Uganda. --- Methods --- Study Design This was a qualitative study rooted within a phenomenological design. It was employed to explore caregivers' experiences in regard to caring for adolescents living with HIV. The study was conducted at Mbarara city Health center IV. The qualitative approach was chosen because it was suitable for exploring and describing the unique experiences of caregivers, their challenges and the coping strategies that have been employed by caregivers of adolescents living with HIV. --- Study Setting This study was conducted from the antiretroviral therapy clinic at Mbarara City Health Centre IV in Mbarara City, South western Uganda. Mbarara City Health Centre is a large volume facility in the center of Mbarara City. They offer specialized services which include the adolescent clinic that is scheduled on a Tuesday every week. The Mbarara City Health Centre facility was chosen because it is one of the facilities with high numbers of adolescents living with HIV. They receive care from the ART clinic. It is located in Mbarara which is the second largest city in Uganda. --- Population The study's population involved care givers of ALWH at Mbarara City Health Centre IV, Mbarara City South Western Uganda. Participants were drawn from the Mbarara City Health Centre four adolescent HIV care clinic using the purposive sampling method. This is because the facility runs weekly ART clinic for adolescents. Also 15 care givers who had come with the ALHIV to the ART clinic were recruited. Potential participants were easily identified by the researcher since it was an adolescent clinic day. Adolescent peers helped in the identification of these participants. --- Inclusion Criteria All caregivers of who are providing direct care to adolescents living with HIV aged 10 to 19 years and consented to participate in the study. --- Exclusion Criteria Caregivers of ALHIV who were not able to use the English language and Runyankore-Rukiga which were the languages being used in the study. --- Sample Size Fifteen care givers were enrolled and they were determined by saturation of information. Saturation was a point where no more new information was being generated from the in-depth interviews. Data saturation occurs when there is repetition of ideas from participants and information becomes redundant. It is used to determine sample size in qualitative research. --- Instrument A semi-structured interview guide was used to collect the data. This was developed basing on the objectives of the study following the cited literature on this the subject. It comprised of two sections: part on the biographic data and the main questions which probed the respondents reaction to the diagnosis, their challenges and how they have tried to cope with the challenges. --- Data Collection Procedure The researchers piloted their semi-structured interview guide at Bwizibwera Health Centre Four. This helped them in adjusting and modifying the questions to suit the context and the objectives of the study. After obtaining the Research Ethics Committee approval and administrative clearance from Mbarara City Health Centre four, the researcher then explained to the study participants the study's topic, its objectives as well as the purpose of the study. They made it clear to the participants that participation was entirely voluntary. A written consent was obtained from all the selected participants of ALHIV . With the help of the adolescent peers, researchers used purposive sampling to recruit care givers of ALHIV who had come to the adolescent clinic day. Data was collected in July 2022 in a quiet place, precisely in private rooms which were allocated to the researchers within the hospital's premises. Researchers conducted 15 face to face in-depth interviews with caregivers. Their responses were recorded using an audio recorder and also in a note book. Participants that is both the care givers and the ALHIV adolescent were given two bottles of soda as refreshment during the study. Clients who attended the adolescent clinic and met the inclusion criteria were consented to voluntarily participate in our study. Data was collected using semi-structured interview guides. On average, each in-depth interview took about forty-five minutes. --- Data Analysis We adopted thematic analysis to analyze the data because it allows the researchers to fully reveal the meanings emerging from the data while conceptualizing narrative reports as per significant units. We first wanted to know the full meanings from each transcript by reading them thoroughly and carefully. We therefore noted down the different themes which were identified. The noted themes were compared to each transcript until we got the relevant themes equivalent to our research questions and objectives. We ensured that these themes provide the evidence from the data collected. Through the various discussions about the themes, we finally concluded on them and proceeded with identifying the compelling quotes from the transcripts which tallied with our working themes and research questions. Validity was ensured through continuous reading of the transcripts by all the authors Discussion of the codes which made us come to final changes which were incorporated into the findings. --- Ethical Consideration Ethical approval was obtained from Mbarara University Research Ethics Committee since the Research Ethics Committee at Bishop Stuart University was not yet accredited. This study complied in line with the Declaration of Helsinki ethical guidelines pertaining research involving human subject participants. We also sought permission from the city health officer and principle medical officer, Mbarara City Health Centre IV and the ART clinic in-charge who allowed us to interact with the caregivers of ALHIV. All the study participants provided consent in writing or through their thumbprint after being informed about the study. Participants were not penalized for refusal to participate in the study. In the informed consent form, Confidentiality was protected throughout data collection to ensure data quality. Participants were informed and authorized publication of the anonymous responses after meaningful analysis. --- Results --- Demographics of the Participants We interviewed a total of 15 caregivers and among them, were 11 females and 4 men in the age category of 30-35 years of age. Most of these caregivers were females and they were HIV seropositive. A considerable number of caregivers were peasants, who dropped out of school at primary level and they were Anglicans by religious denomination as shown in Table 1. The thematic analysis examined the challenges caregivers experienced while caring for the ALHIV and 6 major themes and subthemes emerged as illustrated in Table 2 below. --- Major Themes The responses retrieved from caregivers highlighted the challenges they experienced while caring for adolescents living with HIV and the following themes were generated. --- Theme 1: Diagnosis and Reaction to HIV Diagnosis At the time of diagnosis, some caregivers were not aware that their children were already living with HIV. This was found to be common among grandmothers who had no idea of what HIV was. Their reactions to the diagnosis of their children's status were scaring and it included suicidal thoughts, fear and grief. It was however a different story for the mothers who were HIV positive, it was not that surprising because this was expected by HIV positive mothers who knew there was a possibility of mother to child HIV transmission since some of them did not deliver from hospital. Since adolescents were diagnosed at a younger age and they were biologically related to the care givers; their reactions were divided into two subthemes. --- Subtheme 1: Knowledge on Adolescent's HIV Infection HIV positive mothers knew their HIV status and there was a possibility of mother to child transmission. So, when the results turned out positive, it was not surprising however, they were not happy as voiced in the quotations below; Another caregiver shared about taking the child to the hospital for an HIV test as early as possible so that she could be enrolled on ART if found positive. Since the couple was on ART, this caregiver knew the importance of early enrollment on ART. I tested my child when she was around three months but I didn't see positive results as a new thing because I knew the reason since my wife gave birth from home --- Subtheme 2: No Prior Knowledge on Child's HIV Infection Caregivers stated that they could not think about HIV because they believed their children were young and thus, they could not have involved themselves in sexual activities which is one of the surest ways of contracting HIV. It was after they got severely sick that they went to the hospital. After HIV tests were carried out, they found out that the children were HIV positive. The caregivers developed feelings of fear and hopelessness. This is because they were wondering how these young people were going to live with HIV for the rest of their lives. A number of questions thus, popped up on how and where they contracted the virus from, yet the caregivers themselves were HIV negative. But these questions could not be answered by either themselves or even the adolescents. Below are their reactions: I told the health workers to give him medications since I had nothing to do at the time. This is because the mother claimed that the son might have contracted the virus from the house helper and not her. So, I had to let the child begin ART since at the time he was too sick and was about to die of diarrhea and cough I am not the one who took her to the hospital. It is her aunt who took her there. When she reached there, they discovered that, she was positive. When they returned, the aunt asked me to enter the house. She didn't want anyone to listen to what we were discussing. She immediately showed me the results. I was numbed by what I saw. I ran short of words. However, I later on asked the girl to tell me how she thinks she got infected. She completely denied and refused to reveal the source of her infection. I was hurt and I never want to recall that story because it brings stress to me which I had put behind me Theme 2: Experiences on HIV Serostatus Disclosure Overall, after listening to the caregivers who were interviewed, we concluded that, times of disclosure have been a stressing moment for caregivers. Mostly, grandmothers who have to deal with their own responsibilities and advanced age, are always wondering how their grandchildren will be able to survive those situations. Most caregivers are hopelessness and some of them even contemplate committing suicide. A certain caregiver of a 12-year female adolescent was worried about the future of her daughter. She was worried that her daughter might never get married because of the discrimination and stigma she has witnessed herself. This continuously haunts her. I keep on wondering about what will happen to her when she grows up because I think she will be neglected by men in case she wants to get married. It has taken me a long time to overcome the above thoughts. However, the counselors have tried to help me out. --- Theme 3: Stigma and Discrimination There is minimal sensitization and awareness about stigma and discrimination related to HIV among the HIV negative people. It still affects some caregivers and their adolescents living with HIV. However, some caregivers stated that some of their relatives were supportive both financially and emotionally. --- Subtheme 1: Surrounding People Caregivers were found to be so sensitive to the abusive words from the neighbors and the surrounding society to their adolescents living with HIV. Seeing their adolescents being stigmatized and discriminated from others while playing with their neighboring children is something very hard for them to take in. At school, the adolescents reported to their caregivers that whenever they coughed during breakfast, they would be taken to separate rooms claiming that they had tuberculosis and that they were preventing them from spreading it to other children. Haaaa! Our neighbors abuse her and look at her like an AIDS person and they claim that she is about to die. They also claim she sold herself to acquire AIDS. This annoys me forcing me to exchange words with them. This actually affects her too and she has been stressed for a while though was comforted by the counselor… At first, they used to treat her badly, abusing her using all sorts of words. They assured her that she had AIDS, and that she is about to die. They also kept on telling her that her whole family is going to die. She asks me why they treat her like that. These words have affected her and there was a point when the stigma compelled her to leave school and run away from home. she was found in some place and I was telephoned to picked her up. She once asked me how she got infected with HIV AIDS yet others don't have it. I explained to her --- Subtheme 2: No Discrimination Some adolescents living with HIV and their caregivers have not experienced any form of discrimination. However, it is because they have not disclosed to anyone their HIV status. Relatives who knew about it treat them with care and love. Neighbors and friends do not know. Some relatives know about it and they don't discriminate her, we went at home and disclosed our status to them. So, they don't treat us in a different way and incase she doesn't have what to eat they always give us something to eat. Those who know about it take care of this child well and they have actually helped me in taking care of him well and they mostly give him comforting messages just in case he is lonely --- Theme 4: Care Disengagement This theme clarifies the consequences that have resulted from adolescents living with HIV. Adolescence being a period of changed behavior and wanting to discover, the adolescents tend to ask so many questions eg Why they are taking medications etc. They end up disengaging themselves from care which results into losing weight and adhering poorly to ART. This has been discussed under two sub-themes which is weight loss and poor ART adherence and retention in care. Caregivers were worried about their ALHIV losing and not gaining weight, yet they feed well. This could be because they were not taking ART as they are supposed to. An incidence of throwing away medications was voiced as reported below; At some point the viral load was high and he had lost weight and I wondered why until when the counselor probed and discovered that he had been throwing away the medications. Counseling was done until he resumed taking his medications and now, he is okay Subtheme 2: Poor ART Adherence and Retention in Care Some care givers have disclosed HIV status to the adolescents and given them the ART keep and be in control since caregivers believe that these adolescents are now mature. Adolescence being the age of role confusion and sometimes disrespectful, they intentionally refuse to take medications and their parents find it hard to convince them due to the questions they ask like why I am the only one taking drugs every day? What did I do wrong? As a result they disengage leading to poor ART adherence and retention in care She refuses to take medication and later falls sick. This disturbs me so much. Otherwise, she wouldn't have any problem Another care giver observed: She would have no problem but she has refused to take medication and it becomes worse when she is at school, I try to talk to her but still ignores me. She also has also developed a habit of not listening to me at all while taking to her, I don't know if it is this age or she is angry that I transmitted HIV to her --- Theme 5: Challenges with Care Giving As earlier on asserted in this study, adolescence is a vulnerable stage which is characterized by immature judgment and bad decision making. They ask questions which are very difficult to answer yet, if they are not given answers, they decide not to continue engaging themselves in care. Caregivers also have financial challenges especially during clinic days. They usually lack transport to the hospital and because of this they have to walk long distances to the hospital. And lastly, they also lack quality food yet they have families to take care of. --- Subtheme 1: Challenges with Adolescence Their age is very difficult to deal with. This has become a burden to the caregivers because they cannot listen to them in case, they want them to avoid a certain vice, for example engaging in early sex which could result into early pregnancies. They also ask their caregivers why they were born different from other children and why they take pills every day. Such questions stress and psychologically challenge the caregivers. Also, some caregivers have not told their adolescents their status because they claim it might be a terrible experience and so they are still preparing for the right time to tell them. It has not been easy, first of all they are young and cannot advise themselves to make the right decision. So, caregivers also have a burden of simultaneously handling their children's stupid age and their status at the same time. It becomes so stressful to the caregivers. More so their age is so hard to deal with. For example, the children ask why their parents did not make sure she was born negative like others. Yet some of them are not old enough for one to explain to them. This is a serious burden to the caregivers. Another care giver also added; Recently she began acting differently, saying that she was fed up with life and she even wrote on paper showing everyone how she was HIV positive. She could not allow anyone to talk to her. I am sure she didn't think twice before doing it. I can't tell you that I know what brought her to that situation. Most probably it might have been that adolescence age. Most of the interviewed caregivers have no formal education hence they are not employed. They depend on farming which cannot provide enough for their families. In addition to the burdens of family responsibilities, their adolescents need much supervision to make sure they are adhering well to ART. One care giver said: She is jobless. Thus, she has nothing to eat. The health workers told me that she has lost weight. I think it's because she lacks what to eat and because of this, she is stressed due to lack of a job A widowed caregiver lamented that she had lost her husband to AIDS. So, she lacks financial assistance and this financial incapacitation has made her caregiving situation worse. I give her the necessary care she needs. The only problem is that I am a single parent. This gives me hard time when providing care. The father died last year. That has affected her care. For example, when it's time to go to the hospital, it is hard for us to get transport to take us up to the hospital. I am the only one to provide for the children. So, she doesn't receive the right care that she is supposed to get, I even end up missing the real appointment days and come after it has passed which shouldn't be my wish and this affects my child too --- Subtheme 3: No Support from People The other group of caregivers stated that they had never received any support from anyone and that they would also wish to get support in order to help them reduce on their burden of responsibilities since they also have other siblings to take care of. They also added that they would want support in form of health education because these adolescents are in a critical stage which needs constant reminding about their status so that they can be retained in care. A care giver lamented; We used to hear that when you have a child like mine, you would be given support from the hospitals. For me I have never got anything like support for my child. I only suffer with her. I have never had a phone call telling me to go and pick for her stuff. I wouldn't mind any support given to me, be it school fees and what to eat so that she can be healthy so that people fail to realize that she is sick I have heard that different people get support mostly in education. But I have never received any form of the support --- Theme 6: Coping Strategies Caregivers also shared with us the strategies which they employ in order to cope with the challenges which they experience. They also stressed the importance of supporting caregivers of ALHIV. --- Subtheme 1: Support from People Caregivers revealed that what kept them motivated is the support which they got from relatives and other organizations from the hospital. For example, counseling although this has not been adequate. It is given once in the lifetime of the children. I only got support during the covid-19 pandemic. I was given 50000 shillings. I used it to buy chicken. This has helped me because my child eats an egg every day. We have also been filling forms to get support in vain Yes, I was given milk and transport the day I brought my child for an HIV test and results turned positive. It helped me so much. I think it also helped me in reducing the stress on the day of diagnosis --- Discussion This study was a qualitative exploration of the experiences, challenges and coping strategies employed by caregivers of adolescents living with HIV . The demographic data showed that there are many females who are primary caregivers compared to men. This is probably because we included only caregivers who were providing direct care to the ALHIV. In the African society care giving is perceived to be a responsibility of females and men can do it only if they are willing. It is believed that men are not supposed to engage themselves in caregiving since it is a role that is traditionally assumed to be for mothers or females in African settings. This has caused female caregivers to experience psychosocial disruption, hopelessness and fear and this could be the reason as to why adolescents were not gaining weight per age. The same findings correlate with the data that was reported in Uganda, Nigeria and Ghana by Uganda Aids Commission and Asuquo et al. [9][10][11][12]16 Experiences of Care Givers When Caring for the Adolescents Living with HIV Caregivers described this whole period from the time of diagnosis as a stressful and difficult moment. It is a period that is dominated and stirred by hopelessness, sadness, suicidal thoughts, shame and guilt. At the time of diagnosis, caregivers and their adolescents experienced worrying reactions towards the diagnosis, most especially caregivers who are not biological parents like grandmothers and uncles. The diagnosis usually occurred after adolescents had severely fallen sick. This prompted them to go to the hospital for health checkup. Most of them were found to have no idea about HIV. They thought it was some other kind of sickness not until they got severely sick, resulting into a permanent moment of painful emotions, stress and guilt to the caregivers. This is in line with a study done by Lazarus et al, where study findings also emphasize difficult moments for care takers after the diagnosis turning out to be positive. 17 The same experiences are consistent with findings in other studies. 14,18,19 In addition to the above experiences, caregivers kept wondering how they would explain this situation to these adolescents. They also got worried about their future and thought it would be difficult for them to get spouses due to their HIV status. --- Challenges with the Daily Care of Adolescent Living with HIV The different experiences emerged with a variety of challenges which affected both the caregivers and the adolescents as discussed from previous studies. 14,17 Adolescence stage which is defined by increased decision making, increased peer pressures and search for self is the most experienced challenge and it is burdensome for caregivers. This is because during this stage, most of the adolescents become disrespectful towards caregivers and fail to listen to important advice and end up with poor adherence to treatment. Disrespect itself to care givers can demotivate them and reduce on the care given to ALWHIV. The findings of a study done by Griffith et al, 36 show that if youth are not cared for, it can result into poor adherence and poor retention which means they need to be monitored and cared for to achieve good adherence and retention in care. Such changed behavior and immature decisions make caregivers wonder if it will be possible to keep these adolescents in school so that they can survive this vulnerable stage without messing their future. 37,38 They are also worried if they will manage to provide their education. Some care givers are still challenged on how and when HIV status disclosure to these adolescents should be done, they feel the society should not know therefore consider these adolescents too young to keep that secret. Fear of adolescents asking many questions for example how they acquired the disease contributes to failure of disclosure which keeps them living in fear. These findings are in agreement with the previous findings which reported age as a major factor for failure of disclosure. 39,40 HIV stigma and discrimination is also another challenge faced by people living with HIV. This study showed that it was from the neighboring society and schools. Our findings correspond with results from previous studies which showed that stigma still existed among PLHIV and majorly from the society and healthcare settings. [20][21][22] Caregivers also asserted that stigma and discrimination had slightly reduced as a result of counselors' interventions among those societies. This was because it had affected the adolescent's health and had caused a high viral load. This is because the adolescents were not adhering to ART. Our findings are comparable to other studies done by Gabriel et al and Katz et al. 23,24 According to Ashaba et al, continuous counseling, family and religious support can reduce stigma and discrimination. 25 Care givers are challenged by other factors including the adolescence stage which is characterized by immature decisions, early sexual relationships, being argumentative and disrespectful towards caregivers. Such factors have stopped adolescents living with HIV from engaging themselves HIV care. [26][27][28][29] Disengagement of these adolescents living with HIV from care has resulted into issues related to their health like weight loss, and poor ART adherence. 30 This has resulted into continuous weight loss of ALHIV at every hospital visit. Financial difficulties were also identified as one of the challenges faced by all caregivers. 16,32 This makes it difficult to access daily utilities like food, transport to carry them to ART clinics and school fees. Again our findings are similar to previous studies done in Limpopo province and India. 33,34 The caregivers lack stable incomes to provide for their needs since most of them are peasants. Studies done by Kang et al, indicate that poverty contributes a lot to the challenges faced by people living with HIV and is associated with poor adherence. 34,35 Because of the multiple challenges faced by caregivers, they have adopted coping strategies like support from the hospital through counseling by trained counselors about adherence to antiretroviral therapy and being retained in care. Caregivers have once in their life time received stuff like food and milk to help on the finances spent on food. Rencken et al, study findings highlighted the importance of support to adolescents living with HIV. 41 In addition Kose's study findings showed how support to the adolescents living with HIV with school based support improves adolescents living with HIV wellbeing and health. The findings also show that peer support from family members is also important and it can engage, access and sustain treatment which improves adolescent's health. Rumana's study findings showed how caregivers are burdened by depression, anxiety and stress and so they require motivation to keep doing their work effectively more so to informal caregivers. --- Conclusion Care givers experience a variety of challenges and families are the main source of home-based care for the adolescents living with HIV . Therefore, these families experience challenges related to family needs, psychological challenges due to the adolescence stage and so they should not be made the only shoulders to lean on but social network support should be strengthened for most caregivers as a coping strategy. Therefore, the following are the recommendations of the study findings: First and foremost, Ministry of Health should ensure free access of educational services for ALHIV to relieve them from financial difficulties which they are facing today. The services should include school fees waivers and school feeding schemes. This will reduce their present day unnecessary spending of the little income they have on education and feeding yet they have to take care of their health. Keeping them in school also makes them busy. This prevents them from engaging in premature relationships and bad groups. Ministry of Health and other non-government organizations which are interested in adolescents' health should also budget for these caregivers. At least let them receive support once in a year most especially those who cannot afford money to buy food. Furthermore, there is need to address the challenge of adolescence as a vulnerable stage by the care givers and health providers. They should engage ALHIV in activities like preparing their diet and picking their medication on the ART clinic day to keep their minds busy. Health providers should also provide educational talks. Health education services with fellow ALHIV could reduce stigma and cause more self acceptance. In these education talks, they receive counseling, peer advice and also get people they look up to as their role models. --- Strengths and Limitations We are confident about the findings of the study due to qualitative approach that is significant in exploring participants' experiences. In addition we managed to reach data saturation. The study was also conducted in the local language familiar to all the study participants and each interview was accorded enough time to capture all the views of the study participants. The study was limited by the fact that we captured only care givers experiences and not adolescents themselves. Therefore future studies should aim at capturing adolescent's experiences in regard to their HIV status to eliminate any bias created by this approach. --- Abbreviations HIV, human immuno-deficiency Virus; AIDS, acquired immunodeficiency syndrome; ALHIV, adolescents living with HIV; ART, Anti-Retroviral Therapy. --- Ethics Approval and Consent to Participate We obtained ethical approval from the Mbarara University Research Ethics Committee. Informed written consent to participate in the study was obtained from all participants upon enrolment. Permission was sought from the town clerk, principle health officer and the ART clinic in charge. All approaches were accomplished in accordance with the guidelines and regulations of the Declaration of Helsinki to uphold ethical standards and respect for the participants that guaranteed their safety and protected their health and rights. --- Consent for Publication Participants gave their consent for the data to be used for research purposes. They were also assured that any information about them will be anonymized. --- --- Disclosure The authors report no conflicts of interest in this work. --- HIV/AIDS -Research and Palliative Care --- Dovepress --- Publish your work in this journal HIV/AIDS -Research and Palliative Care is an international, peer-reviewed open-access journal focusing on advances in research in HIV, its clinical progression and management options including antiviral treatment, palliative care and public healthcare policies to control viral spread. The manuscript management system is completely online and includes a very quick and fair peer-review system, which is all easy to use. Visit http://www.dovepress.com/testimonials.php to read real quotes from published authors.

This study aims at exploring experiences of people caring for adolescents living with HIV, also known as caregivers. By 2021, 150,000 adolescents were living with HIV and 32,000 adolescents were dying of AIDS related causes. HIV/AIDS remains one of the most serious public health problems, especially among the adolescents. This has placed a heavy burden on many caregivers, yet they are essential in caring for ALHIV. However, focus of all interventions has excluded caregivers of ALHIV. Thus, this is the reason why this study is being conducted to find out caregivers' experience in caring for ALHIV. Participants and Methods: A phenomenological study was carried out. Purposive sampling was used to select a total of 15 caregivers to participate in the study. These participants were subjected to in-depth semi-structured interviews. Their responses were recorded, transcribed and translated for thematic analysis. Results: While analyzing the results, six themes emerged. They include: diagnosis and reaction to diagnosis, experiences on adolescent's HIV serostatus disclosure, stigma and discrimination, care disengagement, and lastly, challenges during care and coping strategies. Caregivers experienced feelings of fear, Guilt, suicidal thoughts after diagnosis. Stigma and discrimination of adolescents living with HIV which was common at school and from the neighbors and the adolescent stage were some of the challenges experienced by the caregivers and it makes it hard to retain ALHIV in care.Families are the main source of caregiving to the adolescents living with HIV (ALHIV). The study's findings indicate that caregivers in the families experience challenges related to family needs, and psychological challenges resulting from the adolescence stage. So, families should not be left to shoulder the burden of caring for ALHIV. As a way forward, social network and financial support should also be strengthened for most caregivers as a coping strategy.

INTRODUCTION Transgender is an umbrella term used to describe people whose assigned sex at birth is not fully aligned or congruent with their current gender identity or expression. 1 Prior research has documented a high prevalence of adverse health outcomes in some transgender communities, including mental health distress and suicidality, [2][3][4] substance use, 5 cigarette smoking, 6 and HIV and other sexually transmitted infections . 2,[7][8][9][10][11][12][13][14][15][16] Social stressors have also been shown to burden transgender subpopulations throughout development including experiences of verbal, physical, and/or sexual violence and victimization, 4,[17][18][19] and discrimination in employment and healthcare. 20,21 In the U.S., health surveillance systems do not routinely include standardized measures with which to identify transgender respondents to monitor the health of the transgender population. 1,22 As a consequence, a common methodological weakness of many studies in transgender health, especially those seeking to examine health disparities, is the lack of a cisgender comparison group. Therefore, clinic settings such as community-based health centers that offer services to transgender people, represent important locales and potentially rich sources of data with which to study transgender health, particularly in urban areas. 23 Although not representative samples, clinic samples offer the novel opportunity to not only understand more about transgender people engaged in healthcare, but also to compare transgender and cisgender patients in order to document and understand health differences in specific special patient populations. These data can inform delivery of healthcare services at clinic sites, including funding and resource allocation, cultural competency training efforts, and contextualize transgender patient care. 24 Data can also be linked to electronic medical record data to obtain provider-documented diagnoses, biomarker information, and health insurance claims documenting healthcare utilization to increase the rigor of information obtained via self-reported surveys. Despite the recognized potential of clinic-based data systems for transgender health, 25,26 there has been little methodological attention explicitly paid to study design issues and sampling of transgender patients in clinic settings for health research. The current study aims to fill this gap. Two research objectives are the focus of this analysis: Feasibility: can a cross-sectional clinic-based sample of patients be used to document health disparities by transgender identity within an urban health clinic? Efficiency: can the same study be designed more efficiently, using substantially less time and resources, and produce the same results? We formatively compare transgender and cisgender patients presenting for care at an urban community health center on key health and social stress indicators using two methods to investigate these research questions. First, we use a cross-sectional clinic-based sample . Second, we use a nested matchedpair subsample , race/ethnicity, educational attainment, and income). --- METHODS --- Study population The present study used de-identified data from The Community Health Center Core Data Project. The Core Data Project was a survey of patients engaged in medical care from 2001-2002 at Fenway Community Health Center in Boston, Massachusetts. 27 This project was initially conceived as a quality improvement initiative to inform FCHC's practice and improve patient care. During a 12-month period, all medical patients were invited at intake to voluntarily complete a one-time self-report, paper-pencil one-page anonymous questionnaire to help FCHC improve programming and clinical care. The survey was designed with input from FCHC providers to collect information about experiences and health concerns believed to be important for sexual and gender minorities who make up a large proportion of FCHC patients. The objective of the project was to improve clinical care and not to contribute to generalizable knowledge; thus the project was determined to be a quality improvement activity and not research. 28,29 The de-identified data set, as defined by the Privacy Rule, 30 did not directly or indirectly contain identifiable information. As such, it was determined to be research not involving human subjects and therefore not requiring Institutional Review Board approval. --- Study design Gender identity was assessed using a single item. Participants were asked to describe their gender with response options "male", "female", and "transgender". The inclusion of this question made this analysis possible. The full cohort sample was comprised of 2,653 patients who participated in the Core Data Project and fully completed the survey. The nested matched-pair subsample consisted of 155 patients drawn from the full cohort. In a nested case-control study, cases of a disease or condition that occur in a defined cohort are identified and, for each, a specified number of matched controls are selected from among those in the cohort who have not developed the disease. 31 For the purposes of this study, we used a hybrid approach with a matched-pair design and defined "cases" as transgender patients . Controls matched 4:1 were selected from cisgender patients to control for confounding and improve statistical efficiency. For each transgender case, two cisgender female controls and two cisgender male controls were selected, matched on age , race/ethnicity, educational attainment, and income. Matching was implemented to control for confounding and improve statistical efficiency. --- Measures The brief one-page survey was designed to minimize patient burden as much as possible given the survey was administered in the patient waiting area. Brief single-item screening questions were used for all measures. No psychometric information is available for the items, although many of the items were similar to other screening instruments commonly used in clinical settings . [32][33][34] Demographic matching-Demographics matched on were age, race/ethnicity, educational attainment, and income. Age was a continuous measure and calculated from year of birth to date of survey completion. Race was operationalized as white and any racial/ethnic minority , Latino/Hispanic, other race/ ethnicity). Education was coded as lower , moderate , or higher . Anticipating a non-linear association of income and health, 35 we first categorized income into quintiles to examine the socioeconomic gradient in health and used generalized additive models [36][37][38] to determine the most parsimonious coding specification of income. Based on this analysis, income was operationalized less than $20,000 versus greater than $20,000. Health-Five binary health indicators were operationalized: Lifetime suicidal ideation ; Suicide attempt ; Substance abuse history ; Lifetime smoking ; HIV serostatus . Social stressors-Four dichotomous stressors were asked and parameterized. Childhood abuse age<15 was queried , which included any experience of sexual, verbal and/or physical abuse). Intimate partner violence was assessed including victimization , a boyfriend/ girlfriend, or some other intimate partner"), and perpetration , a boyfriend/girlfriend, or some other intimate partner"). Violence in adulthood was assessed using three items for verbal attack, physical attack, and sexual harm. A binary variable of any verbal, physical, and/or sexual violence in adulthood was operationalized. Employment discrimination and healthcare discrimination were assessed. A dichotomous variable of any discrimination was coded. --- Statistical analyses SAS ® 9.2 was used to analyze data, where statistical significance was determined at the alpha 0.05 level. The general analytic strategy was to compare transgender patients to cisgender female and male patients . Transgender patients were also compared to all cisgender patients . For the full cohort analysis , bivariate statistics were conducted to identify differences in demographics, health, and social stressors by gender identity. Multivariable logistic regression models were then fit regressing health and social stressors on transgender identity and controlled for demographic characteristics . For the nested sample analysis, bivariate analyses were not estimated given the sample was demographically controlled by the matched design. Two different approaches were used to analyze matched data. First, conditional logistic regression models were estimated using the PROC PHREG procedure in SAS. This procedure is appropriate for analyzing matched data as well as sparse data. The 4:1 matched cases and controls were treated as strata. The match ID was used as the strata variable. [39][40][41] Second, generalized estimating equation models were computed using the PROC GENMOD procedure in SAS with a repeated statement by match ID. --- RESULTS Findings from the full cohort sample are presented in Table 1, including demographics, health, and social stressors by gender identity. Participants ranged in age from18 to 70 years, mean age 32.0 years. Overall, 78.6% were White ; 1.2% identified as transgender. Table 2 shows data from the nested matched-pairs subsample for transgender cases and cisgender controls . Participants ranged in age from 19 to 70 years, mean age 39.7 years. Overall, 86.5% were White ; 20% were transgender by design due to the matching algorithm. Table 3 presents health indicators and social stressors comparing models estimated for the full cohort to those estimated for the nested matched-pairs subsample using two different analytic techniques. Analyses from both study designs produced virtually identical results . First, the prevalence of HIV, substance abuse, and smoking did not significantly differ for transgender and non-transgender patients in either study design 1 or study 2 . For example, 54.8% of transgender respondents reported ever smoking cigarettes compared to 49.6% of cisgender patients in study 1, and 59.7% of cisgender matched controls in study 2. Second, transgender patients were significantly more likely to endorse a lifetime suicide attempt and suicidal ideation compared to cisgender patients in both study designs. Overall, 29.0% of transgender patients sampled had ever attempted suicide, compared to 8.5% of cisgender patients in study 1 and 12.9% of cisgender patients in study 2. Third, transgender patients disproportionately reported social stressors relative to cisgender patients in both study designs. By way of example, 54.8% of transgender patients reported childhood abuse age < 15 compared to 19.5% of cisgender patients in the full cohort and 25.0% of cisgender matched patients in the nested design. --- DISCUSSION The current study found that data from a cross-sectional, clinic-based sample of transgender and cisgender patients can be used to document health disparities by gender identity. A health disparity is a "particular type of difference in health…in which disadvantaged social groups-such as the poor, racial/ethnic minorities, women, or other groups who have persistently experienced social disadvantage or discrimination-systematically experience worse health or greater health risks than more advantaged social groups". 42 Reducing health disparities is a core aim of Healthy People 2020. 43 Differences in health and social stressors in this study were found by transgender identity in patients engaged in care at an urban health clinic, particularly on mental health, victimization in childhood and adulthood, and discrimination. These results replicate findings from prior research in clinic and non-clinic based samples. 2-9, 17-21, 44 We also found that a nested matched-pairs design that utilized <10% of the full cohort sample performed equally as efficiently and, for a fraction of the time and resources, produced virtually identical findings as the full cohort sample with regard to comparing transgender and cisgender adult health. Specifically, even after matching the demographic variables for which a statistically significant difference existed in the entire cohort , overall, findings of health indicators did not change in the nested study. Findings suggest that a nested design offers a potentially effective method of using patient data to study transgender health with relatively minor loss in statistical efficiency compared to the full cohort study. Some transgender health disparities shown to be present in the peer-review literature, 5,6,11,14 were not evidenced in our sample, likely given the specific composition of the patient population utilized for this analysis. For example, the prevalence of HIV, substance abuse, and smoking did not significantly differ for transgender and cisgender patients. The lack of health disparities in Fenway's patient population by gender identity can likely be attributed to several factors. First, the organization was the largest primary care provider for HIV-infected people in New England during the time of the initial data capture. Additionally, the organization is a LGBT health clinic and it is well-documented that other sexual minority groups have higher rates of HIV, substance use, and smoking compared to the general, non-sexual minority population [45][46][47] thus, differences in these conditions may not be seen between transgender and other sexual minority patients sampled. Second, we did not have information on assigned sex at birth or gender identity vector . Given the differential distribution of HIV among transgender women relative to transgender men, not stratifying by natal sex/gender status may obfuscate differences that may be present when, for example, comparing transgender women and cisgender women. Third, measures were dichotomously assessed to minimize patient burden. The dichotomous smoking variable is particularly problematic. A lack of disparity in ever smoking may not reflect the nuances of patients' smoking histories . Nonetheless, disparities between transgender and cisgender patients were seen with regard to lifetime suicide attempt, suicidal ideation, childhood abuse, intimate partner violence, victimization, and employment and healthcare discrimination, highlighting the need for targeted individual, interpersonal and structural-level interventions that aim to eliminate social stressors and improve the mental health of transgender individuals in clinical settings. Several limitations warrant consideration. The sample was taken from one of the world's largest LGBT-focused health centers. While generalizability to transgender and nontransgender patients of other health centers is unknown, our findings are consistent with those documented in prior studies using clinic and non-clinic based samples. 2-9, 17-21, 44 Nested matched-pair data cannot make up for the limitations of the original design which was a cross-sectional sampling of patients who presented for care and who were willing to fill out a survey. No information on the response rate of participation represents a significant limitation and an area for future improvement of rigor in this line of research. An additional limitation is that no data were available on assigned sex at birth or specific gender identity . The distribution of health concerns differs within different subsets of the transgender community; therefore, not being able to stratify by natal sex/gender status may conceal some health differences. Also, the childhood abuse screener did not distinguish between emotional, physical, or sexual abuse experiences in childhood. Lastly, data used for this analysis were more than ten years old; however, the focus of this report is on methods and analytic techniques. Clinic-based samples and patient-related data are under-utilized sources of information about transgender health, particularly in community-based, urban health centers that typically serve a large pool of transgender patients, many of whom come from marginalized communities. Well-designed studies that sample transgender patients, and a subset of matched cisgender patients, can provide rich information on transgender health disparities. Matching can be conceptualized as stratification in the design phase to form strata that are balanced and that provide for an efficient statistical analysis. 40,48 It is especially useful with small sample sizes and when a random sample is difficult to obtain, as well as to control for confounding factors that are difficult to measure. A limitation of matching worth consideration is the inability to examine the risk factors associated with the matching variable. 49 Just as retrospective and prospective chart review studies of clinic populations contribute to science in transgender health by reporting on provider-reported patient medical record data , 10,15,50,51 cross-sectional point-of-care convenience samples are also advantageous as they offer the opportunity for clinic patients to self-report on ancillary information that may not be collected in a routine medical encounter . 52,53 Certainly there are disadvantages to utilizing clinic-based samples, the most obvious being selection bias . 23 Additionally, self-reported data are also subject to bias, particularly questions that are sensitive or subjective in nature . Ultimately, however, clinic based samples, and in particular data gathered outside of a patient-provider encounter , provide clinicians and researchers with the opportunity to supplement patient medical records with additional social and behavioral sciences data which can serve to advance the body of knowledge regarding specific marginalized patient populations as well as improve the quality of patient care. 52,53 Clinic-based settings offer the unique opportunity to investigate health disparities by gender identity by comparing the health and wellbeing of transgender and cisgender patients. 26 This study assists in filling a gap identified by The Institute of Medicine's 2011 groundbreaking report in LGBT health which stated: "All aspects of the evidence base for transgenderspecific healthcare need to be expanded. Research methods that will yield the data needed to inform decisions about transgender-specific health should be developed" . 1 Gender identity should be routinely collected as part of patient demographic characteristics in clinical settings and will allow unique opportunities to conduct research in transgender health and build the knowledge-base for transgender clinical care. Multi-site Note: Among the 2,622 cisgender patients, 33.4% were female and 66.6% were male . No data were available on assigned sex at birth or specific gender identity of transgender patients. LGBT Health. Author manuscript; available in PMC 2014 November 04. LGBT Health. Author manuscript; available in PMC 2014 November 04. --- studies of community-based health centers that pool transgender patient records are needed to replicate findings from this formative study.

Purpose-U.S. health surveillance systems infrequently include measures to identify transgender respondents or monitor the health of this underserved and marginalized population. Methods-From 2001-2002, transgender and non-transgender adults were sampled at a Massachusetts clinic. Health differences were formatively examined by transgender identity using a cross-sectional, clinic-based sample (n=2,653); and a nested matched-pair subsample (n=155). Results-Both designs produced virtually identical findings: (1) the prevalence of HIV, substance abuse, and smoking did not differ significantly for transgender and non-transgender patients; (2) transgender patients were more likely to endorse a lifetime suicide attempt and ideation compared to non-transgender patients (p<0.05); (3) transgender patients disproportionately reported social stressors (violence, discrimination, childhood abuse) relative to non-transgender patients (p<0.05).suggest that a nested design may provide an effective methodology for using clinical data to study transgender health, and underscore the need for routine collection of gender identity in clinical settings.

Introduction IoT describes a different world of heterogeneous objects such as sensors, smartphones, and actuators in which everything, even objects, has an independent identity [1]. They have distinct features, such as different operating systems, platforms, communication protocols, and related standards, but all these differences are ignored when interacting with each other [2]. Therefore each device needs to communicate with other things around it to meet the needs of its users. In recent years SIoT is a new described expression in the literature that comes from combining SNs with IoT, which is related to the interaction between things and the Internet as a network substrate, and its functionality like information exchange, behavior, and relationship are independent of the human intervention [3]. Moreover, objects have this capability to establish a social relationship autonomously, and the communication between them can be varied from simple-to-complicate, for example, as simple as using some smartphone applications like Waze to route the shortest path or as complex as the communication infrastructure of a smart city [4]. SIoT used all interconnected things all over the world to create a social network based on common interests and motivation to provide better services to end-users [5]. The contribution of SIoT is for different objects to cooperate effectively and securely to gratify end-users' desires to satisfy some main parameters such as reliability, safety, time, cost-effectiveness, and availability. Many survey papers have not focused on SIoT comprehensively to discuss it in all its aspects in detail, which motivated us to provide a new systematic literature review article because of: -Lack of comprehensive knowledge of the SIoT system that examines and compares it in detail from all aspects provides accurate statistics and results. -Lack of sufficient information about the structural and communication details of objects on the SIoT can be an excellent guide to exploring and finding its nature and challenges. -Lack and non-occurrence of resources are used in summarizing the results and evaluations performed, which has led to less accuracy of the results. -Lack of a clear method research and paper selection approach helps other researchers gather their references and related information. -Lack of transparency of the structural issue, especially in detail such as platforms, datasets, object relationships, human roles, components, and in the context of existing and forthcoming challenges, and lack of possible solutions to solve or anticipate them. Therefore, the main fields of our research include the following: • Present a taxonomy to conceive a better view of SIoT main elements. • Display a comprehensive study on the main principles and concepts of the SIoT to evaluate and examine its primary structure and cover the lack of sufficient information and awareness. • Illustrate all aspects of SIoT holistically include the main features, evaluation parameters, and simulation tools for each component of the SIoT. • Display of platforms and datasets used in evaluating SIoT samples of studied literature. • Provide meaningful statistical and technical information by investigating and analyzing many papers. • Provide major SIoT challenges for a network of smart objects. • Propose open issues and future guidelines for solving problems. The structure of this article is as follows: "Background" section provides a brief background on Social IoT and presents recent studies. In "Research method" section outlines our research methods and questions. "The fundamental components of the Social Internet of Things" delineates the main structure of SIoT and highlights taxonomy and analytical comparison of recent literature. Consequently, in Section five "Discussion" and Section six "Conclusions" are presented. --- Background The population of humans on the earth is about 7.4 billion. This population volume needs to connect to the objects around them to improve their quality of life by sending or receiving various services. The number of objects is estimated at around 75 billion by 2025, much more than the human population. This rapid and dramatic growth [6] has caused problems such as scalability and feasibility. Therefore, different communication types between objects are necessary for our daily life demands regardless of their kind. The first idea of forming SIoT comes from [7], which Kleinberg introduces the small world phenomenon. SIoT mimics the structure of the human SNs structure and describes how to communicate, the friendship selection, or choose Friends of A Friend between different objects to help provide a range of certain and high-quality services [1]. Each object that is considered a friend of another object must have different parameters such as scalability, interoperability, and trustworthiness to form a healthy friendship based on evaluating these factors. Therefore, each object can communicate with other objects to improve the interaction between humans and objects based on a set of rules defined by their owners, such as how objects communicate, choose a friend, and provide services [8]. Each device we use soon becomes intelligent and social, and with the help of various sensors, can receive or send information. Socialized objects provide more accurate and faster responses to complicate requests than objects that work individually [9]. Hence we can conclude that the smarter things become, the more social the IoT becomes. Some articles concentrate on the human SNs structure, while others consider the SN of objects. Figure 1 presents the articles that recently reviewed the network structure between human SNs and things SNs structure. Tripathy et al. [9] highlight a history from the origin of SIoT that developed from IoT to SIoT. They point to some of the critical features of SIoT, including availability, privacy, and security. They explored future open issues consisting of analyzing service discovery in network navigability, analyzing a massive amount of data to describe human dynamics, and proposed new related models. The constraint of this paper is that it is not apparent how they extract studied articles. Abdelghani et al. [10] point out to trust management in SIoT and describe it as one of the main issues to build a reliable, and secure data exchange regarding QoS offered services. Therefore, they provide a comprehensive surveyed article in trust. They discussed the primary factors and present a classification of trust types, models, related properties, SIoT limitations and requirements, and challenges. A safe relationship has two main elements as trustor and trustee, formed based on mutual benefits and affected by many parameters like location and time. Despite the comprehensively of this article, they do not consider open issues in SIoT. Roopa et al. [11] present a comprehensive systematic review based on SIoT and review the existing related articles to satisfy some main scopes in SIoT. These include trustworthiness and object relationship to improve link connection between friend objects in large scale networks and listed the essential perquisites of the SIoT system. Furthermore, this article's main advantage is the complete analysis of the trustworthiness and summarized various aspects of SIoT relationships. However, the methodology and how they select studied articles are not clear, and the lack of a complete taxonomy on the main issues is apparent. Table 1 summarizes the related studied papers that systematically reviewed the SIoT regarding various aspects and survey articles that have been presented. The article's main content and covered years are listed briefly in this Table. --- Research method All we try to do is display a new correct systematic literature review from the SIoT. So we try to find the latest articles published between 2011 and December 2019 in reputable ISI journals and conferences as our search engines, as shown in Table 2. Then we considered these interesting articles to classify them into relevant groups based on search terms and keywords. Finally, after a long review of articles, 55 articles were accepted in this research to categorize relevant articles to explain the main structure of SIoT. We used six --- Social network structure Things Humans Nitti et al -2014 [25] Yang et al -2015 [26] Eddy and Ousmma -2018 [29] Ma et al -2017 [8] Atzori et al -2011 [22] Xiao et al -2015 [38] Bao et al -2013 [44] Nitti et al -2015 [93] Kowshalya et al -2017 [43] Chen et al -2014 [39] Bao and Chen -2012 [53] Guo and Chen -2015 [42] Kim et al -2017 [20] Atzori et al -2012 [21] Fig. 1 A comparison of the human social network structure and objects' social network structure studied papers AND 2 Objects AND 3 Components AND 4 Relationship AND 5 Trust AND 6 Service and 7 Data AND -The following selection criteria are applied to the final result: -Articles published between 2011 and December 2019 -Articles are related to IoT and SIoT -Articles that provide high-quality approaches and ideas for SIoT have analyzed SIoT analytically and statistically, especially in a particular domain like objects relationship -The following elimination criteria are applied to the final result: -Articles not in the English language -Articles not index in ISI -Articles not access to full-text documents -Articles present an overall survey and systematic review -Articles do not consider SIoT analytically and statistically -The scope of candidate articles is entirely irrelevant to SIoT --- Final selected articles Figure 2 shows the refined articles using the method discussed in the previous subsection, which we have grouped according to the main structure of SIoT. Researchers can have a complete and clear view of the final articles and make it easier to decide on reputable articles. --- Academic Literatures --- Architecture Tripathy et al-2016 [9] Ortiz et al-2014 [16] Evangelos et al -2011 [17] Gulati and et al -2019 [23] Gulati and et al -2019 [24] Dutta and et al -2015 [97] Kim et al -2017 [20] Atzori et al -2012 [21] Atzori et al -2011 [22] Fu et al -2018 [28] Eddy et al -2018 [29] Wei et al -2018 [30] Atzori et al -2011 [31] Figure 3 presents a comparison between total studied papers over time based on their publishers, as mentioned in Table 2. --- Related research questions This article tries to respond to the following Research Questions : -RQ1: What target platforms, datasets are there for SIoT? -RQ2: What evaluation parameters are essential for the core elements of SIoT? -RQ3: What simulation tools are used to simulate the main structure of SIoT? -RQ4: What are the most important challenges of the SIoT? --- The fundamental components of the Social Internet of Things In Fig. 4, we have designed a holistic taxonomy to describe SIoT, which consists of six main components: Architecture, Relation Management, Trust Management, Web Services, Information, and finally, SIoT tools that include datasets and platforms. --- Architecture Although there is no standard architecture for SIoT, most articles proposed a fourtier architecture that includes devices , global connections, platforms, and applications based on IoT architecture [13]. Devices need to be connected to the internet or a gateway directly or indirectly to send or receive information from a particular platform or user application. Global connections are responsible for connecting objects to each other and playing a communication layer between platforms and devices using communication standards, gateways, and protocols to read and exchange information over the Internet between objects as a middleware. Applications are used to capture all user application services, such as smart homes, smart health, industrial, and mobile applications [14]. This architecture defines IoT basics, but it is not perfect for deploying SIoT architecture, so we have shown a five-layer architecture for SIoT in Fig. 5, which includes the entity, component abstraction, communication, social interaction, and application layer. The entity, communication, and application layer functionality are similar to IoT reference architecture, but we have specified the component abstraction and social interaction layer to the SIoT architecture. -The Component abstraction layer: this layer contains the key elements of the SIoT that are essential for communication between objects that include ID 1 , OP 2 , and OC 3 to perform service discovery, trust management, relationship management, and service selection processes. -The Social Interaction Layer: this layer is used as an interface and social agent to establish social communication between smart objects and humans. A comparison between SIoT architecture given in recent contributions is presented in Table 4 in detail. Table 5 presents a summary of studied articles based on their domain and the type of architecture (layered, generic, relational and unified that layered and relational is clear but generic is general and public IoT architecture that allows allocation and change in different areas such as the smart city, agriculture, health, energy [15], SIoT and WoT [16]. Based on different IoT architecture, various IoT applications are created. Their basic architecture is the same as the IoT architecture, but it has been customized on various applications. The unified means integrating architectures based on platforms, devices, smart objects [17] to create a better architecture to solve IoT problems, for example, integrating the edge and cloud computing concept or IoT architecture of platforms with building a suitable software infrastructure based on the micro-services to develop IoT solutions and create smart products [18] or create a middleware integrated architecture for IoT security for empowering IoT applications [19]). Ortiz et al. [16] introduced the main dimensions of SIoT as Social role , intelligence , social conscious devices and everything as a service . Accordingly, they developed a hybrid architecture of IoT and WoT, includes actors such as objects and users, an intelligent system for managing and coordinating transactions between objects including searching and discovering for services, exchanging information, an interface for activating interactions between objects, including control commands and service management and finally the Internet as a communication platform for transferring object services to users. This study's advantage is that they consider the challenges and open issues deeply and provide a detailed description of SIoT while the deficit is that they do not introduce any suitable platform to implement their method. Kim et al. [20] proposed a new architecture for SIoT end-users based on a threelayered Socialite system: Socialite Client application, Socialite Server, and Databases. Socialite Client application has two components, first, control or access to devices remotely, and the next, allowed them to define their own rules for communication by end-user programming. Socialite server is used to access overall access to devices that may have different creators, and databases are used to process and store data. Atzori et al. [21,22] proposed a two-tier social-based architecture for SIoT: one is for SIoT server which contains three sub-layers: The base layer, the component layer, and the application layer, which is the base layer responsible for storage, data management, database and communication, and the component layer contain the core elements of the SIoT system , and the application layer also contain applications, API and interfaces. The other side is the client that has three sub-layers including objects layer, object abstraction and the application layer also includes two parts, including service management , and social agent that is used to communicate between objects and the SIoT server. In the base layer, two types of components are introduced, including key and lateral components. The key SIoT components include Id management that assigns and manage an id to identify each object, object profiling, which includes static and dynamic information about each object, and owner control, which includes a set of rules set by the object's owners and other components including service discovery, service composition, trust management, and relationship management. This study's strong point is the Integration of human social network architecture with the SIoT concept, but there is no simulation in this article. Gulati and et al. [23] introduced four-layer semantic-oriented platform architecture for SIoT as a reference model, including four-layer: objects, communication, SIoT management, and application layers. The layer of objects refers to the objects that are placed on the SIoT, and communicate through local networks and sensors, and is the lowest layer. The communication layer includes protocols, gateways, and technologies that are used to communicate between objects. The SIoT management layer is related to SIoT platforms and services that include essential components such as id management, object profiling, owner control, service discovery, service composition, trust management, and relationship management that communicate with the application layer via the API. The application layer includes various applications such as mobile applications, web applications, etc. So they have devised a complete architecture for SIoT, but they have not explored the details about their implementation. Gulati and et al. [24] proposed a three-layer architecture for industrial IoT that by using semantic technologies, they have investigated the relationship between different objects. They have seven design considerations for their SIoIT architecture: resource management, relationship control, data management, interfaces, scalability, interoperability, trustworthiness, security, and privacy. Their proposed architecture has the following layers: -The base layer consists of two sub-layers includes the infrastructure and communication layer. -The middleware layer responsible for interactions and transactions between different components in SIoIT, and it acts as an interface between the hardware and the application layer. -The application layer accesses to user applications. This article's strength is effective communication management and the combination of the SIoT concept with the industry. The disadvantage is that it is not dynamically tested in a real industrial environment and remains as future work. --- Relation management There are many interesting papers mentioned to RM in SIoT. For example, according to [9], every object on the social network receives more accurate responses to their requests than objects that work individually. Therefore, these objects need to communicate with each other to form a social community to request or respond to relevant services, and the SIoT works based on these friendly relationships. Besides, the main purpose of creating SIoT is to separate things from humans to build their social networks autonomously. Thus, RM shows the hidden intelligence of objects that help them decide to start a friendship, update, or end it [25]. Also, each object must choose its type of relationship with other objects [26] to share relevant resources, information, and services. Therefore, the management of their relationships is based on the choice of relationships they have with other objects around them. The relationships between different things start when they find social nature. Therefore, according to recent articles [20,21,27], we can divide the relationship between different devices into five main groups as presents in taxonomy in Fig. 4, namely Parental OR, 4 Roopa et al. [11] delineated and classified new relationship types for SIoT based on two main categories: UO 5 Relationship, which includes four main sub-categories as OOR, SOR, SIBOR 6 , and GSTOR 7 . OO 8 Relationship . Atzuri et al. [22] provide a relational architect for the social structure of objects in SIoT to address many of IoT's problems, like service discovery and service composition. 10:52 Therefore, they have simulated objects' mobility to evaluate the underlying relationships between objects and provide statistical analysis on SIoT to illustrate the similarity between human and object social networks. This approach's strength is to improve network navigation capability, but there is no simulation or suggested algorithm in this study. Fu et al. [28] have presented a theoretical model-SSIoT -that works on four real datasets: Facebook, CDBLP, fsrWeibo P2P, to show that search engines are influential in SIoT social relationships. They first considered the search engines as an interface between social networks and IoT, and then evaluated six performance factors that included: degree distribution that strengthened power-law, and then assessed network diameter and network distance , and network density, network stability, and user betweenness are increased in SSIoT model. The strength of this article is the presentation of appropriate statistical analysis about the performance metrics. This model's limitation is that the propagation of malicious codes in the SIoT network is much more than SIoT. Eddy et al. [29] proposed a reference architecture using the SIoT concept to exchange services in industrial properties for object analysis and evaluation. For this, they have compared human communication based on the Fiske model in verses of objects communication to point out that these communications are similar to each other and mentioned to SIoT protocols and related features to form a community of social communication industrial objects. The benefit of this study is that they considering IIoT scope based on SIoT. However, the service evaluation cost was not considered. Wei et al. [30] introduce physical objects as social communication concepts to describe the relationships of the physical objects accompanied by IoT to distinct the concept of social relations for physical objects from humans or smart objects. Therefore they proposed a category based on spatial-temporal features of social relationships. This paper's profit is that they accelerated the finding of relevant physical objects between huge amounts of them. In Atzori et al. [31], an explanation is given for the various communications required in SIoT based on the communication models in human SNs. Also, the type of necessary objects transactions were shown in a Table and an app example, but there is no clearness in the object's interaction. --- Human role on the SIoT system The social network of objects is modeled on human beings' social behaviors so that they can communicate with each other and form their social network. So how humans communicate with each other is the key to creating a social network of objects. The vast impact of IoT on human life is undeniable, which has led to a two-way relationship between humans and objects. The SIoT, as a part of IoT for the proper function and production of content, requires human beings and the laws that determine it. On the other hand, SIoT is a system that can obtain the information it needs from the environment, is an essential factor in the accuracy and speed of transactions. For every object that is placed in this system, three main features must be considered [21,29]: Object ID Management , Object Profiling , and Owner Control • ID Mng: each object has its identifier with which they are identified throughout the SIoT system. • OP: it contains static and dynamic information from each object in the network that acts as their profiling information for other objects. • OC: is a set of rules by object owners to control the behavior of objects, such as how the objects communicate with each other, start, end, or change the state of themselves in the network. After identifying objects, each object is allowed to be present and active in the SIoT under the permissions given by its owner. The OC is responsible for providing these permissions that greatly influence the way objects behave, such as determining the maximum number of members, the conditions that objects can establish, terminate, or change their status. Therefore, humans have a key and decisive role in the formation and management of SIoT systems and ensure their efficiency and effectiveness with their determining rules as owners of objects. Hitherto, a unified model of human social communication was provided by Alan Fiske in 1992 [32], which shows four elementary relational models based on human communication in social networks to analyze and evaluate their communication behaviors. His model was presented in Fig. 4 as a taxonomy that includes Communal Sharing, Equality Matching, Authority Ranking, and Market Pricing. -Communal Sharing: it's for people who belong to a particular community, such as family and friends, to communicate based on the shared profile or the information they need or offer. -Equality Matching: based on the similar connections between different members of a particular group, while retaining their own identity and keeping load balance in the data exchange in the involved objects . -Authority Ranking: humans are classified hierarchically according to different levels and with different complexity. Each human being has access to a level based on his score, which leads to their asymmetric categorization, so anyone with a lower rank should be a service provider to someone with a higher score. -Market Pricing: communication between people is based on common benefits and social privileges such as wages, prices, and values. Table 6 summarized the articles studied on the SIoT that provide RM. The article's primary content, the strength, weakness, and new finding of this article was listed for each topic. Table 7 lists the main evaluation parameters related to RM based on the studied literature. --- Trust management Many articles are discussed TM in SIoT. For instance, trust is one of the most critical topics in today's technology, such as IoT [33] and cloud computing, which deals with how objects interact together [34,35]. Lack of trust in objects that work together socially leads to some problems such as loss of privacy, safety, security, access, and alteration of the information by unauthorized persons or things. Besides, object owners may carry out harmful attacks based on their interaction with other objects, such as Bad-mouthing, Self-promoting, and On-Off attacks; hence evaluating trust between them is essential for SIoT to identify the best interaction between customers and providers. Anything that needs to establish a trustable connection needs to have sufficient confidence in the other things that want to be connected. It can make a higher trustworthiness communication to provide some requests and separated malicious things in the network from trustable things [25]. Confidentiality is one of TM's main issues, which means that information is available at the right time by the right person because each object has its vulnerabilities Kim et al. [20] - - √ - √ - - Atzori et al. [21] - - - - - √ √ Atzori et al. [22] √ - √ - - √ √ Fu et al. [28] - √ - √ - - - Eddy et al. [29] - - - - - - - Wei et al. [30] √ - - - - √ √ Atzori et al. [31] - - - - - √ - and attacks. Therefore, a control system is required to prevent unauthorized access to data and network resources. Hence, it is vital to introduce a security policy in the access control system to limited access to the network from attacks [36]. Also, things adjacent to each other contain much crucial information that they can share with other things in a distributed social manner to provide some qualified services, which should be based on the right choice of things as friends because each object can discover relevant services by inquiring from its friends or FoAF to decrease searching area. After all, it is vital that every object only exchange data and services with those things that are trustable and provide a healthy and reliable communication to respond to requested demands, which leads to increased security and safety of SIoT networks. --- Trust types The most of related articles such as [10,11,25,[35][36][37] mentioned trusting main types as: -Transitivity: trust between different things is based on indirect known, so that for example, if has confidence in and has confidence in , so must trusts to . -Direct: it is based on direct perception between two things. -Indirect: it is based on other objects, recommendation, and reputation. -Local: it is different for any object. For instance, trusts to , but does not have any trust to . -Asymmetric: there are different levels of trustworthiness between two friend objects, for example, has a high trust to , but is not as same as the . -Subjective: it is based on individual opinions like [11,25]. -Objective: it is based on each object's quality of service properties like [11,25]. -Context-dependent: it is not equal for the same devices and related to objects context like [11,38]. -Composite: it is a composite of friend opinions and recommendations which leads to trust or distrust. -Personalized/History: it is based on the prior background of things to each other; for this reason, may two separate things have different trust levels to each other. -Dynamic: it means that the trust value is not static throughout time and may be different over time, whether the conditions altered like [11,[39][40][41]. Abdelghani et al. [10] presented two types of trust between objects, including quality of service trust and social trust . Kowshalya and Valarmathi [35] explained a dynamic trust model to consider the degree of resiliency of trust in SIoT against On-Off selective forwarding attacks and, for the first time, categorized trust except Direct and Indirect into centrality , energy and service score [43] . --- Trust models In the sense that there are several models for building trust in SIoT and all models need to be evaluated for correctness and accuracy, so evaluation and model concept is intertwined. There are several models for building trust in SIoT, all of which must be assessed to use trust evaluation models instead of trust models. Many articles mentioned to these evaluation models such as: Nitti et al. [25] described a dynamic trust model based on the level of honesty between different objects. Two models for the trust were supported, including subjective and objective. The subjective model has a slower transitory response than an objective model; besides, objective processing and trust information storage are done in a DHT system observable by all network objects. The strong point of this study is that they can separate distrust objects from network effects. Also, technical analysis for TM is done between related objects that cooperate, such that each object can be malicious if it is strange in the network. Other objects meet it rarely, and it did not become active in social activities. On the contrary, other objects can befriend. However, this article's weakness is that they do not use direct observation to evaluate trust metrics and just assess indirect trust observations. Xiao et al. [38] propose a SIoT guarantor and reputation trust evaluation model based on objects' behavior to propose a proper service response and used some effective techniques like credit rating and reputation rating to evaluate the degree of trust between objects. Any object that provides a complete service gets a higher rank than those who do not collaborate or do not prepare any appropriate service; finally, the objects with lower rank known as malicious objects. This work is defined as a suitable approach to determine dishonest objects, but they do not consider all main trust aspects in large scale networks like scalability. Chen et al. [39] present three kinds of social trust factors based on owner's interaction, including Friendship, Social Contact, and Community of Interest relationship, which is based on common interests, and the system resiliency against opportunistic service attacks were considered. The limitation of this work is that they do not consider attacks approaches. Chen et al. [40] proposed an adaptive TM protocol which is based on main TM features to assess trust response: Honesty , Cooperativeness and Community of Interest ), but the deficiency of this study is that they do not consider dynamic environment issues. Sharma et al. [45] suggest a cooperative trust relaying and privacy-preserving model that, to the best of our knowledge, is the first attempt that used Fission Computing and edge-crowdsourcing network to evaluate privacy and trust. They have used several theoretical analyses and numeric simulations to implement their proposed model and numerical simulations to implement their proposed model and have also helped evaluate the proposed method's effectiveness from real data. Truong et al. [46] delineated a general description for trust in all SIoT components according to trust information used to evaluate the amount of trust in SIoT. Moreover, they have developed a platform to assess trust services in SIoT, which has three main parts as Reputation , Recommendation and Knowledge . Also, they use a car-sharing service use case to implement and consider their approach. In Table 8, a summary of the related studied papers that considered SIoT concerning TM parameter has been presented. In Table 9, based on the studied articles, a brief explanation of TM's main evaluation parameters was listed to conceive them better. --- SIoT trust attacks Many security attacks affect the SIoT system, which is listed below. However, many attacks face unauthorized access and suspicious behavior [47] due to a lack of authentication infrastructure [48]. The others try to present a proper perspective of themselves to gain other objects and users' trust. So first, we need an appropriate mechanism to verify the authenticity and confidentiality of information in unauthorized access to the data and network resources [36]. To this end, we can use effective techniques to predict the behavior of objects such as machine learning, decision tree [49], and deep learning approaches. They separate malicious objects from reliable ones by predicting their next behavior to make better decisions about exchanging data. The importance of one object to another can be assessed to prevent suspicious communication between them [35]. It is also possible to design a trust management system or consider an effective encryption mechanism for encrypting data in data transmissions and sharing so that information can be exchanged securely [50]. Also, centralized privacy-preserving communities of objects can be created to form a trustable data-sharing model [51]. Also, like [25], we can use subjective trust approaches. Each node calculates its friends' level of trust based on its history or objective methods that use distributed DHT structures to securely use information from its friends. • Bad-mouthing: misleading the background of the trustable objects to decrease its selection chance by cloud services [10,40,42,44,52,53]. • Slandering attack: misleads trust feedback to change its results in service providers [54]. • Self-promoting: promotes itself position by offering good feedbacks about itself to alter trust services for service providers [40,42,44,[52][53][54]. • Whitewashing: a malicious object clean its bad background by separation from the application and, after that, join it again to increase its chance for selecting by service providers [10,40]. • On-Off Selective: an object goes to the Off state during the primary transaction to consume a higher amount of energy until it comes back to On state [35,42,52]. • Opportunistic service: a malfunctioning object provides good services to enhance its popularity to collaborate in some attacks like bad-mouthing [10,42]. • Discriminatory: attacks to those objects without strong community interaction with other objects [10,40]. • Ballot stuffing: provides good recommendations about misbehaved objects to increase the selection chance of them to choose by service providers [10,40,42,44]. --- Web Services As web services are one of the main challenges for developing many recent technologies like IoT, SIoT, cloud, and fog computing, we concentrated on the primary process from the beginning that a particular service is requested until an object responds to it. Therefore, a web services process scheme was illustrated in Fig. 6, a combination of three main sub-processes, in order as Service Discovery, Service Selection, and Service Composition. Service discovery is the process of searching for objects that can provide the desired service. Service selection is a selection process of appropriate service matched to requested services. Service composition is the process for combining different services to offer an appropriate response to requested services, which results in the desired quality of service, better functionalities, and feasibility because a service cannot meet all user's needs. So, choosing suitable services and a combination of them can be one of the main concerns in the SIoT environment. Kouicem et al. [55] propose a dynamic framework based on multi-agent structure implemented on cloud computing and used large-scale composition techniques according to real use cases-in smart cities monitoring scope-to enhance performance, QoS, and contextual metrics of service selection and composition approach. Multi-agent architecture is responsible for the services' core operations, divided into plan composition, service selection, and service orchestration to describe effective services and abstract services. They also implement three algorithms: an algorithm for plan generation, an algorithm for the best selection, and an algorithm for updating QoS metrics after selection execution. Finally, they evaluate plan generation time, service response time, and success rate of services in comparison to three similar approaches and demonstrated the scalability, flexibility, and adaptability of their approach. However, they do not consider any contextual models, and there is not any exact statistics for their QoS evaluation metrics. Using a three-dimensional structure and RESTful, Chen et al. [56] provide a distributed social structure for service discovery and selection based on object interconnections. Greg et al. [57] present an integrated framework for semantic web service discovery, selection, and run-time integration for non-functional evaluation factors, like response time and throughput, to satisfy user's requirements. They used a HIS-Healthcare Information System-use case to present the usability of their method. Their experiments are based on using a greedy search technique to combine the service response process at an optimal execution time. They provided a comparison between their method and QoSbased and hybrid approaches . They demonstrated that this approach is better in scalability and performance and is optimum in execution time rather than a hybrid approach. Butt et al. [58] proposed an adaptive, context-aware, and scalable protocol for service discovery in IoT. They use Trendy techniques to improve service discovery's main factors like reducing service invocation delay, the amount of energy consumption during the network life cycle and packet overhead, and improved network scalability by reducing the number of packets. Trendy employs an efficient discovery technique with a context-aware selection in IoT web services. Wei and Jin [59] proposed a context-aware architecture and a capable ontology-based context model to provide an efficient service discovery support for IoT. The main goal is to describe the role of context in IoT and remove unreliable context because it has an essential role in making a smart discovery process regarding minimum human intervention. This study's main point is that they present a reliable and efficient discovery process, although there are no analytical statistics to prove efficiency. Xia et al. [60] point out to SLSA-an efficient and scalable service discovery for SIoT concerning three main performance metrics including the success rate of queries, an average number of relay nodes, and average path length of searches to provide an efficient service search for secure and fast web services discovery. They compared their proposed mechanism with three others and classified 500 web services into 50 groups that each of them has 10 types of a specific service. Finally, they conclude that SLSA has high search efficiency in queries' success rate, much more energy efficiency in the average number of relay nodes, and decreased average path length of the discovery process than the other compared mechanisms. This study's strong point is that their evaluation is based on a dynamic behavior in a large scale network, and based on their results, SLSA can address resource limitations in SIoT. Klauck and Kirsche [61] presented a DNS-SD approach for discovery web services for resource constraint devices in IoT. Their approach is based on integrating three protocols called uBonjour to improve interoperability 12 An opensource OS. 13 Multicast DNS standard for finding other devices to share their information. between resource constraints and non-constraint devices to discover desired services. This approach's benefit is to promote the self-configuration and service discovery process, though they did not mention any possible flaws. Also, their implementation testbed is small. Shamszaman and Ali [62] present an effective solution for sharing resources and choosing the right object at the right time to deliver a service, taking into account QoS criteria in SIoT. It has used the Integer Programming Solutions to implement it, and the strong point of this article is that their work has high flexibility in resource sharing. However, they do not consider the entire system in terms of QoS and keep it as future work. Baker et al. [63] proposed E2C2-an energy efficiency SC algorithm for IoT-based services regarding integrating the minimum number of web services to manage the amount of energy consumption for IoT devices to satisfy user's requirements. Their implementation is based on the performance comparison of E2C2 with existing approaches like Base Cloud, All Cloud, Smart Cloud, and COM2. The critical point of this research is the high performance and energy-awareness of the E2C2 approach, but they do not consider SS in this article to complete their SC method. Pallec et al. [64] proposed a service composition approach according to the available recommend-based IoT services to satisfy the user's requirements. They used TAG framework to classify the recommendation of available services to facilitate the smart objects' communications. The important part of this study is that they used a physical-interface-based algorithm to calculate the service signature for enabling service interactions. However, they do not implement any simulation for their approach. Ma et al. [65] presented a new hybrid approach that combines Genetic Programming with a greedy search approach on built test suit benchmarks like WSC-2008 and WSC-2009 on web services to enhance some QoS main factors like availability, reliability, and. They compare their hybrid approach with GP-based without greedy search techniques to improve the search process in a large service repository to provide a more correct and accurate composition approach. Besides, they test their approach for QoSaware service composition in dynamic and static behaviors. This article highlights that using a greedy search decreases the number of web services for the GP-based composition approach to reduce search time and increase search efficiency. The weakness of this study is that they offered no implementation to develop their approach. In Table 10, a summary of the related studied papers that considered SIoT regarding the web services process has been presented. In Table 11, based on the studied articles, a brief explanation of the main evaluation metrics was listed to analyze them. Figure 6 presents the web services process from the first. The user first requests a service. The discovery process begins to find the best services that meet the user's demand in the service repository. However, extracting an appropriate service from the service repository needs to select a specific service based on its functional attributes. Service selection needs to check whether this service description matched with the end-users requests or not. If this service is found, then selection started, and if multiple matched services were discovered, then based on the network and user context information, the most appropriate service was selected [58]. Otherwise, it creates the appropriate service with the composition process and at the end, try service filtering to remove impertinent services to provide an absolute service regarding QoS factors and context-aware properties like location or context of services like its functionality descriptions that most of the user requests are concentrated on it [57]. --- Information The data exchange in SIoT is according to the three main characters of Big Data notion, namely Volume, Velocity, and Variety , which have the potential to manage effectively, collect, maintain, and process to be able to analyze this massive data for evaluating and predicting objects behaviors, information [16]. Another issue in this domain is related to the massive volume of data, which was integrated into the interaction between different devices and things, so we need a specific solution to aggregate and collect data and need particular software to process this huge amount of data [66]. For example, we can present SIoT Big Data as a four-tiered framework includes: --- Table 11 A comparison between the important parameters of evaluating web services -A collector server: to collect the sensed data from a common community of objects in our environment. -A storage server: to store different types of data that are aggregated from various devices. -A filtering server is a primary processing unit to remove noisy data from raw data then send it to the process server to accomplish in SIoT interactions. -A process server: to analyzed and processed stored data. Lakshmanaprabu et al. [67] implemented a Java-based SIoT data classification with an optimal feature extraction framework that presented a comparative analysis between the proposed framework with other algorithms like PSO, ACO, and GA in some performance metrics like accuracy, count of features, throughput, energy consumption and time based on four different datasets including Indoor User Movement Prediction from RSS, Water Treatment Plant, Hepatitis and Twitter Dataset for Arabic Sentiment Analysis. Therefore, they use MapReduce and Hadoop framework to improve the efficiency of the proposed framework. They filtered data by Gabor filter to reduce irrelevant data and remove unwanted noises. It is derived that by this approach, the improved accuracy is 98.2%, specificity , and sensitivity . The strong part of this research is that they provide technical analysis compared to other existing algorithms, and the drawback is that they do not consider evaluation cost. Hasan and Fadi [68] proposed a fault-tolerant routing topology for exchange data in the SIoT network by using the PMSO 15 algorithm and compare it with FPMSO 16 and CPSO 17 methods to consider the connectivity between humans and devices, furthermore; by using PMSO, they can create, maintain and select k-disjoint paths that can tolerate possible failures in data transformation paths and implement a high-performance routing protocol. This research's important point is that they analyze a model for each mentioned parameter like energy consumption, average delay, and throughput evaluated by fitness function with the different k-disjoint multipath routing protocol. Ahmad et al. [69] proposed a four-layered Java-based architecture based on ABC for Big Data to extract attributes, MapReduce Hadoop, to analyze a large amount of data, and Kalman filter to remove undesired noise from data processing. They compare ABC with PSO, ACO, and GA algorithms to demonstrate its high accuracy and efficiency in feature selection for Big Data notion. Therefore, they used some test datasets to evaluate the performance metrics like throughput and their algorithm's processing time. The weakness of this study is that there is not any statistical analysis of main performance metrics. Alam et al. [70] presented a comparative analysis of eight well-known algorithms 18 on data mining scope to consider each algorithm based on accuracy, processing time, and execution time. Finally, compare them based on the results together. According to this, they find out that C4.5 and C5.0 have better accuracy and processing time rather than other algorithms. Also, they have better average accuracy , 15 Bio-inspired Particle Multi-Swarm Optimization. 16 Fully Particle Multi-Swarm Optimization. 17 Canonical Particle Swarm Optimization. respectively. The weak point of this study is that they do not evaluate large datasets and keep it as a future experiment. Ahmad [71] proposed a parallel processing architecture for analyzing and processing huge amounts of data in SIoT systems by simulating big datasets like Vehicular Mobility Traces. To do so, they evaluated performance by using the Apache Hadoop framework. Their approach is divided into three functional scopes as an object, SIoT server, and application domain. The main advantage is that they demonstrate a direct proportional between data size and throughput, either between data size and processing time . In Table 12, a summary of the related studied papers that considered SIoT regarding information has been presented. The article main content, the strength, weakness, and new finding of these articles was listed for each topic. In Table 13, based on the studied articles, a brief explanation of the primary evaluation was detailed to analyze them. --- SIoT tools --- Platform IoT platforms are those software's that can connect anything in an IoT system to each other through cloud services [72] to create new applications [11]. Hijazi et al. [72] were categorized IoT platform components in four subcategories as Sensing Component, Communication and Identification Component, Computation and Cloud Component, and Service and Application Component. There are many platforms specifically for Big Data in IoT, such as Apache Hadoop, 1010 data, Cloudera, SAP, HP-HAVEn [73], besides AirVantage and IBM Watson used for Big Data management [74]. Table 14 presents some main SIoT platforms in detail, including many scopes such as related interfaces, data manipulation protocols, tools, databases, the platforms functionality, and the URL that can use some of them like ThingSpeak, Paraimpu, Lysis, and the others for social purposes. Moreover, many other platforms such as IFTTT, 19 Belkin, 20 Quit, 21 NaturalFuse, 22 Cityscripts, 23 and PhpBB Social Network 24 that are not listed in this Table . Byun et al. [75] proposed Lilliput-an ontology-based SN platform for IoT-to illustrate an ontological and socialize perspective for interaction in IoT concept between humans, devices, and the places by using a social graph model. Also, they proposed an environment model to show the communication of smart devices and people with together that divide into three main groups as object-to-objects, humans-to-humans, and places-to-places interactions that can have many bidirectional relationships among this main classification, for example, they can have an object-to-place communication or place-to-humans or objects-to-human and vice versa. The advantage of this study is that they consider all aspects of the relationship between people and smart objects and implement efficient hybrid architecture for both IoT and SNs. The limitation of this study is that maybe increased the cost of utilizing machine learning methods. Pintus et al. [76] develop a scalable and user-friendly platform -Paraimpu-for a large scale of the SWoT paradigm to add, share, compose and interact with the HTTP smart virtual objects with together that provides data adoption in the heterogeneity nature of devices. The restriction of this study is that they do not consider the social relationships between devices. Girau et al. [77] discussed the first SIoT platform based on Thing speak to manage social interactions in a centrality manner. However, they do not consider any PaaS model for this issue, so that [78] illustrates a cloud-based platform called Lysis that uses a semantic search approach and has four main elements as SVOs that act as social agents to help for making a social interaction in a community autonomously to improve SD and scalability, a PaaS service model to use SVOs in the cloud to facilitate the management and development of the application by programmers and non-programmers effectively, Cloud-Storage that use a user cloud space to store devices generated data and information, Reusability to make this possible to create object templates to use in a common community. Beltran et al. [79] introduce a semantic web service platform among humans, devices, and web services by using SNs as an SCE where the area surrounding that users can make their desired web services privately on the devices to enhance the cooperation between semantic web services and SIoT. The excellent point is that they combine Web with IoT effectively, but there is some limitation to describe more details in their implementations. Helal et al. [80] introduced a new tool called Persim 3D to generate a real dataset from the usual activities in a smart environment like smart homes. Cicirelli et al. [81,82] proposed a new smart city agent-based platform based on SIoT attributes called iSapiens that is a Java-based platform for designing a smart environment and using edge-computing, out-of-the-edge computing, and IoT to build a pervasive intelligent environment to form a cyber-physical paradigm for implement smart city applications. This platform has two main elements as Virtual Objects to manage the heterogeneity of objects in IoT and software service agents to online and offline analytics by edge computing and out-of-the-edge services on computational objects. This study's benefit is improving the fault-tolerant and main QoS factors and using them from a real case study to exploit the architecture. The deficit is that there is not an implementation or algorithm in this study. Chen et al. [83] presented new hybrid architecture to implement a platform for the SIoT paradigm based on the heterogeneity of IoT devices and the concept of SNs by using related protocols like MQTT, CoAP to analyze the collected information from these devices besides the SNs that built on Hadoop and Spark cloud servers. Sanchez et al. [84] presented a three-layered architecture for a real use case-Santander city in Spain-includes objects, gateways, and IoT servers to monitor environmental variables like pollution, weather conditions, etc. IoT objects are some resource limitation devices that sense the environments. Gateways build a connection among IoT objects and the network infrastructure. Finally, the server acts as a data repository to aggregate objects sensed data. The strong part of this study is using a large dynamic scale IoT testbed to evaluate smart environments. However, because of the ultra-large-scale of this scenario, maybe some maintenance and scalability problems occurred. Zhang et al. [85] presented a platform for social WoT to interact between humans and objects by SNs. Table 15, illustrates a summary of some studied articles about platforms that cover the main article topic, platform name, and the scope of using platforms is presented for each article briefly. --- Dataset Table 16 presents different types of datasets based on their features and URLs. -CRAWDAD: this dataset is a community to aggregate wireless network data resources and stores this data from many different locations to collect and analyze them. 25 Many contributions to this dataset are explained in Table 16. -Reality Mining Dataset: this dataset is a collection of one hundred human subjects in nine months and presents information on users' location, communication, and device usage behavior. 26 -Washington State University's CASAS Dataset: the CASAS project described the real smart environment and sensors as intelligent agents that can provide a safer and more comfortable life in our environment by gathering vital information from our surrounding environment by using these smart devices [86]. This dataset is divided into many different categories based on their usage like Real-Time Smart Home Stats, Assisted Care Apartments Real-Time Activity Update, etc. -SNAP: Stanford Large Network Dataset is a huge collection of different datasets in various scopes, for example, social networks dataset that is related to online SNs interaction between different users and has many subsets like Brightkite or Epinion. --- Discussion In previous sections, we provide a comprehensive description of all aspects of SIoT that, based on them, some research questions were presented. In this section, we answer these questions based on statistical and comparative analysis of these technical questions. RQ1: What target platforms, datasets are there for SIoT? Based on "SIoT Tools"section, we provide a comprehensive overview of all IoT and SIoT platforms and datasets. We present their features and targets absolutely in separate Tables . RQ2: What evaluation parameters are essential for the core elements of SIoT? Figure 7 presents the main evaluation metrics in RM and TM that, based on it, in RM scalability is the most considered metrics by about 29%. Then navigability has 22%, time and adaptability have 14%, and finally, the stability, connectivity, and efficiency by 7% for each of them are evaluated more than other factors and in TM accuracy and resiliency are the most considered metrics by 25% . Then adaptability has 19%, and reliability has 13% that they are considered more than others, and finally, scalability, cost, and time have 6% that was assessed as less concentrated in literature. According to Fig. 8, web services were considered based on their essential evaluation factors, which is the most important factor by 23%. Then scalability has 16%, and availability has 14% that are considered more than others. Reliability and cost have 12%; adaptability, flexibility, security, and interoperability by 9%, 7%, 5%, and 2% are at the next levels. Due to this, the primary evaluation parameters in SIoT information was investigated and according to it, throughput and time are the most important metrics that have 25% for each of them, accuracy and efficiency have 19%, and finally, energy is at the last position by 12 percent. RQ3: What simulation tools are used to simulate the main structure of SIoT? Figure 9 shows the statistical percentage of the simulation tools applied in the studied literature. The Java-based tools like Eclipse have the most percentage in simulation tools in reviewed studies with 28%. The SWIM simulator has 13% usage for simulation , and Apache Hadoop and MATLAB toolkit has 10% of usage commonly in literature. --- Challenges of SIoT Many challenges in the SIoT can be significantly crucial for researchers to address new solutions for them. In Fig. 10 we listed the main challenges with their possible solutions --- Possible solutions Challenges Fig. 10 SIoT main challenges and possible solutions that they are not considered exactly in the reviewed articles, and we illustrate all of them as follows: RQ4: What are the most important challenges of the SIoT? • Heterogeneity: SIoT is made up of millions of objects with different features in sources, platforms, protocols, and standards, and all objects and data must be recoverable [87]. These differences have led to the creation of a heterogeneous network of objects that affect their interaction and compatibility directly with each other and increase complexity [87], so that the heterogeneity nature of objects leads to many important problems like interoperability and compatibility that need solutions to solve them [9], such as: • Using POR objects relationship: the parental objects can be used in such systems to ensure compatibility between them because all objects are homogenous and compatible with each other. • Using middleware for non-POR objects: in contrast for other objects that are not homogeneous, can be worked on a suitable middleware as an interface to communicate and adapt them by enhancing the communicability, applicability, practicability, and operability of devices without considering models or manufacturers [88]. • Discovering object identification policies: We need new addressing solutions for finding device identification [87]. • Mobility and Dynamicity: smart objects placed in a dynamic environment are constantly changing their location, which leads to problems such as lack of effective object search for selecting and providing services. The other important issue is the dynamic behavior of objects and environments, which leads to a change in the state of objects. So, the objects alter their status in the network. Some solutions proposed for these issues, such as: • Create objects communities: for solving mobility, groups of objects can be formed into communities based on distinct features like their movement, social behaviors, social similarities [89], and common interests to interact with each other [90]. When an object changes its position, the structure of the community alters based on this displacement. Therefore, we can use some functions like Euclidean, adjacent matrix, or Global Positioning System to get the current location of objects for calculating their distance or using the SWIM 27 model for modeling their mobility [90], which is used to generate location-based traces for the SIoT objects [91]. • Manage dynamic behavior of objects: for solving this issue, objects need to dedicate some main rules and protocols by their owners to manage these changes to prevent from changing the network topology, although adaptability is the other issue that arises from this dynamicity because an object needs to adapt itself with these frequent changes. 10:52 • Tracking objects: one of the main issues in SIoT and large scale networks that were considered rarely is tracking objects, interactions, and activities. We can propose some solution for solving this problem such as: • Using a graph model: we can present a graph model for smart object interactions based on following objects' social behaviors. • Determine rules: some rules must define to construct, update, predict, or remove the edges between two objects. Each object can be considered the main node. Their connection with together creates their edges. Their behaviors add weight to it according to the type and feature of a pair of objects relationship like common interest, providing a specific service/services, same location, etc. • Using objects movement patterns: Zhiyuan et al. [92] proposed Resource Discovery based on Preference and Movement Similarity for tracking objects movement patterns using GPS and three-dimensional location approach. • Security, Trust, and Privacy: because of a huge connected environment of devices, opportunistic services, and users in SIoT [51], security is an essential key to share information guardedly. So, contrary to many types of research that have been done in this field, it remains as one of the significant challenges and requires handling for the survival of the system in the face of various attacks to have security, reliability, availability, resiliency in interactions [9]. So there are some solutions to this problem, such as: • Access control system: we need a control system to prevent unauthorized access to data. • Effective encryption mechanism: We can also use an effective encryption mechanism to encrypt data reliably or use cost-effective and self-synchronizing models [50] to secure real identities from attacks. • Trust management framework: moreover, we can work on the SIoT trust management framework [43] to provide new applications for making trust between objects and users. • Safe data sharing model: We can also create privacy-preserving user and object communities based on policies to form a dynamic and safe data sharing model for more security and privacy [50]. • Predicting node behaviors: We can use some techniques like machine learning [26] and decision tree for predicting nodes behaviors [95]. • Resource-constrained devices: even though the SIoT is a system with limited resources and this issue has a direct impact on the life of the network and the exchange of information, there is still no optimal solution to address this issue by considering energy limitations at all levels of design to have more effective interactions [9]. Therefore, we need more studies and research to create an effective resource management system [93] on the SIoT to get the most out of such a system. • Effective resource management system: due to the mobility nature of objects, which leads to much more computation power and energy in this system, we need an effective resource management system for solving it. • Efficient service search and discovery: the large number of objects in SIoT led to lower scalability and navigability for selecting friendships, searching services, and making an appropriate connection between objects. This issue increases the overhead at all levels of the SIoT system. So one of the main issues is the effective search and discovery of the services, which leads to a reduction in system overhead, provision of services with greater efficiency [52], better response time, decrease objects transactions time, and increase the navigability and scalability of the network [94,95]. So there are some solutions to this problem, such as: • Using a mechanism to control overhead: we need a mechanism to control overhead and effectively search for services and resources, such as creating communities of objects based on main parameters such as social similarities [90], providing similar services, resources, other similarity measurements [49]. • Using new searching algorithms: in [90], they used intracommunity and intercommunity service search algorithms between detected local and global communities to discover services efficiently [90]. --- Conclusion This article delineates a comprehensive systematic literature overview of SIoT. First, we provide a brief background on SIoT. Second, we have collected data from a large collection of previous works. We propose working methods, paper extraction methods, and research question design to explore open topics and challenges of evaluating the data collected. We reviewed related articles published between 2011 and December 2019 in English and ISI journals. In the end, about 55 articles were selected for statistical and technical analysis, focusing on the main structures of SIoT. In the third, we developed a technical classification of the core structure of the SIoT ecosystem, which includes six sub-categories: architecture, relation management, trust management, web service process, related information, and tools and shows the elements one by one perfectly to highlight its strengths and weaknesses and express its main idea. Also, this study can be created and used for future work to test more closely and accurately the critical elements of SIoT, such as trust management, relationship management, and friendship selection. It could also explore it for future research in promising areas such as smart grids, smart cities, and smart industries. There are many challenges in this area that need to be thoroughly explored, and we will keep them as future work. --- --- Information --- Platforms Fig. 2 A taxonomy of reviewed academic literature based on SIoT main elements --- --- Funding No funding was received. --- Competing interests There is no conflict of interest. ---

IoT describes a new world of billions of objects that intelligently communicate and interact with each other. One of the important areas in this field is a new paradigm-Social Internet of Things (SIoT), a new concept of combining social networks with IoT. SIoT is an imitation of social networks between humans and objects. Objects like humans are considered intelligent and social. They create their social network to achieve their common goals, such as improving functionality, performance, and efficiency and satisfying their required services. Our article's primary purpose is to present a comprehensive review article from the SIoT system to analyze and evaluate the recent works done in this area. Therefore, our study concentrated on the main components of the SIoT (Architecture, Relation Management, Trust Management, web services, and information), features, parameters, and challenges. To gather enough information for better analysis, we have reviewed the articles published between 2011 and December 2019. The strengths and weaknesses of each article are examined, and effective evaluation parameters, approaches, and the most used simulation tools in this field are discussed. For this purpose, we provide a scientific taxonomy for the final SIoT structure based on the academic contributions we have studied. Ultimately we observed that the evaluation parameters are different in each element of the SIoT ecosystem, for example for Relation Management, scalability 29% and navigability 22% are the most concentrated metrics, in Trust Management, accuracy 25%, and resiliency 25% is more important, in the web service process, time 23% and scalability 16% are the most mentioned and finally in information processing, throughput and time 25% are the most investigated factor. Also, Java-based tools like Eclipse has the most percentage in simulation tools in reviewed literature with 28%, and SWIM has 13% of usage for simulation.

Magura and Laudet, 1996; Wasserman and Levanthal, 1993) . According to Watson and Parke , there is a great deal of research on gender in relation to substance use disorders, but there is a small amount of recent qualitative research in this area looking at the experience and perceptions of the individual. In contrast to the substantial amount of literature on how individuals adjust to substance abuse, there has been little work examining how drug dependence affects the parental functioning . Heroin-consuming mothers frequently express inadequate feelings toward parenting, with higher probability of using negative discipline . Haight et al. report that mothers, who are substance abusers, assume that their problem has an impact on their children's physical and psychological well-being, which includes: the exposure to domestic violence, to adult substance abuse and to other anti-social behaviors, as well as the loss of important relationships. In general, mothers who use drugs are commonly believed to be selfish, uncaring and neglectful with their children, although some studies have revealed that these mothers felt capable as parents, with a high valorization of motherhood and holding firm standards for childrearing . However, most of these studies do not report if these mothers are the primary caretakers nor what is the domain of care that they mainly provide. There is also a lack of information regarding the family conditions in which their children live, that is, family structure, number of adults taking care of the child . Because of the lack of information on parenting of drug-addicted mothers, we decided to conduct an exploratory study, using Grounded Theory. Few qualitative studies have been used to study parenting although these methods, particularly Grounded Theory, are extremely adequate to identify relevant concepts and hypotheses, and to assess social interactions and processes in a dynamic way . The objective of this study is to contribute to a theoretical model about parenting of drug-addicted mothers, on a methadone program, with children under two years old. The empirical research questions, included one main question: 'How is it to be a drug-addicted mother?', and some subsidiary questions: 'How do they feel as mothers?', 'What are their main problems and/or concerns?', 'How do they deal with those problems on a daily basis?' --- Method Participants Twenty-four Portuguese women, attending a supervised methadone program participated in this study. Their ages ranged from 23 to 42 years old , and they had one to six children with the younger child between 1 and 32 months . Twenty-one mothers were unemployed and all had consumed drugs and methadone while they were pregnant. --- Procedure The data were collected in a Maternity and at an Addiction Treatment Center, in Lisbon, Portugal, after having obtained official authorizations to conduct this investigation by both institutions. After being informed about the purpose and relevance of the current study, all participants signed an informed consent form. The interviews were conducted at the Maternity and at the Addiction Treatment Center, in private rooms. All were conducted by the same interviewer and audiotaped, taking from 60 to 90 minutes. The interviews started with an open question in order to motivate and allow the participants to talk freely about their own parenting experience -'How has it been to be mother?' Each interview was transcribed and analyzed before the next one was made. The information gathered during the interview or during coding allowed the elaboration of new questions, that were used in the next interview. So, the data collection, the coding and analysis were concurrent. --- Data analysis The Grounded Theory is a general research methodology that uses a latent pattern analysis. It can be used with both qualitative and quantitative data, and its purpose is to produce empirically based grounded prepositions. This methodology derives from a 'discovery model', rather than a 'verification model' in which data are gathered in order to verify previous hypotheses . In this study, a 'discovery model' seemed appropriate to fulfill its main purpose: to deepen our knowledge about the particular experience and different perspectives of mothers who are addicted to drugs and resulting in a smooth uninterrupted emergent analysis and the generation of a substantive or formal theory from data collection . Since substance abuse can be a field of preconceptions, it seems appropriate to choose a methodology that keeps preconceived ideas from emerging, letting concepts slowly emerge from data, differently from Qualitative Data Analysis. Through the codification process, the 'constant comparison method' and memo writing, a core variable or Basic Psychosocial Process , emerged: Functional Parenting. Several relations were found between this core category and other categories, allowing the creation of a theoretical model. The model emerged gradually, enabling the identification of contextual categories, that characterize the risks of maternal drug addiction. The theory thus explains sufficiently how people continually resolve their main concern with concepts that fit, work, have relevance and are saturated . --- Results --- Circumstances for substance abuse during the pregnancy There are some circumstances that contribute to create a favorable context to the difficulties felt by mothers who were substance abusers: a significant menstrual irregularity or even an absence of menstruation, which, along with a depreciation of sexual relations, a contraceptive negligence and heedlessness toward their own body rhythms and signs, make pregnancy an unexpected condition. Pregnancy is usually very late detected, increasing insecurity feelings, making the physical and psychological preparation, and adaptation, more difficult, and increasing the inadequacy or failure of the prenatal medical care. One mother said: 'I didn't have my period for several years, I thought I couldn't get pregnant ... never took the pill ... I never thought I could get pregnant' . So, in 16 cases, the pregnancy was not planned, something mothers did not expect in their lives, and some experienced it as a shock, as they do not feel prepared for motherhood. Some had voluntary abortions of previous pregnancies, as a way to avoid and/or delay motherhood; others did not act on it, but they thought of aborting before they made the decision to keep the baby: 'I didn't know what to do, I thought of aborting as I did before, but then I decided to have this baby' . However, some of these mothers actually wanted to have a child. This seems to be related to a greater capacity of developing stable relationships, more often when the partners are not substance abusers or when both are undergoing treatment. In these situations, the pregnancy is planned or, at least, 'allowed'. They say they were aware of the probability of pregnancy, having done nothing to avoid it. Despite the fact that pregnancy is usually not planned and detected later than usual, a medical supervision usually exists in order to decrease the ominous effects of drug dependence on both the mother and the child, especially because these pregnancies are frequently considered at-risk pregnancies. Pregnancy in drug addict women frequently coincides with the beginning of treatment , by their own initiative, but in other cases it is suggested by doctors who accompany them. The pregnancy is perceived as a moment of change, representing a reason to stop consuming drugs and to begin recovery treatments. However, in spite of this wish for a change, their recovery attempts sometimes fail, resulting in relapses and substance abuse during pregnancy and even after the child's birth. The fear of consequences for the child is often present: 'I was always thinking ... if she was ok, all I wanted was that she'd be a healthy baby' . --- Ambivalence between addiction and pregnancy/parenting Pregnancy and motherhood is characterized by a strong ambivalence between addiction and parenting. Contradictory feelings and thoughts tend to arise: on one hand deep feelings of despair, anguish and anxiety emerge, on the other hand, pregnancy and motherhood are simultaneously moments of hope, in which mothers stop to think, and begin to see their child as a salvation: 'I didn't know what to do, I felt I couldn't stop taking drugs, how could I be a good mother? Afterwards I could see things more clearly: I could stop it for this child, my baby would save me and it gave me strength enough to carry on' However, they often see the child as an extension of themselves, where they place expectations and roles that the baby cannot accomplish: 'My son helped me very much in my addiction. I felt I was more responsible and he gave me strength to keep going on with my life and to recover from drugs' . These women usually feel they cannot cope with the ambivalence, entering an exhausting cycle of feelings of guilt, which we called guiltiness/non-guiltiness. They frequently blame themselves for consuming during the pregnancy/motherhood, for consumptions prior to the pregnancy and for relapses. They fear for serious consequences for the baby's health, and await the child's birth with significant anxiety. After birth and during infancy, eventual addiction's consequences on the child make these mothers feel guilty. One of them said: 'I used drugs when I knew I was pregnant, sometimes I relapsed and I felt very scared and guilty for the consequences it could bring to my child' . Nevertheless, they concomitantly reveal a need of counteracting this guiltiness by justifying their need for taking drugs. So, they speak about being impotent toward the need of consuming, feeling incapable of control. Some of them use 'normalization' as an argument, stating that no drug addict would be able to avoid consuming, as a way of relieving themselves from the responsibility. When symptoms arise in the newborn, women try to minimize them as much as they can, using normalization, that is, considering themselves as similar to any other woman, in a constant search for self-valorization. They repeat the same process with their children, who they consider normal by comparison to other children . --- Functional parenting In most cases, especially regarding the mothers who began treatment a year or two before the pregnancy, the parental role is developed with satisfaction. These mothers enjoy being with their children and even miss them when they go to kindergarten. However, some mothers continue consuming during the pregnancy and even after giving birth. Also, the relationships with the partners are sometimes disruptive, leading to the separation of the couple. In these families, the initial period of parenting is difficult. Faced with a constant ambivalence, there is very little space for the mother role, and a central process that we have called 'Functional Parenting' arises. This means that, in these women's personal universe, parenting represents a functional part of their lives, an almost distorted role, seen as a mere function. In this way, the 'Functional Parenting' becomes evident in some strategies and behaviors: mothers describe several difficulties regarding the maternal role . In the mothers' words: 'She is very stubborn, she has a strong personality, she seldom does what I tell her to do!' or 'I get lost when she cries, I don't know what to do!' . Confronted with these difficulties they make a huge effort to keep on doing what they think is essential, in their role as a parent. They try not to deprive their child from the satisfaction of basic needs because of their drug addiction, and the child is even referred to as a priority . Sometimes, these mothers consume minimum dosages to prevent them from feeling physical and psychologically bad and to allow them to be vigilante and take care of their child, dealing with his/her most immediate needs and with any eventual accidents that may occur at home. The hangover, when it occurs, is sort of a controlled hangover, in which mothers wait, suffering, for a new chance to consume. In their descriptions of the parent-child relationships, they mostly refer to the basic care and the effort they make to meet these needs: 'I have given my children all they need, although I could be relapsing and making a huge effort to keep on taking care of them' . They refer to a high level of fatigue and a lack of patience/ availability, trying to assure basic care taking of the child as soon as they get home. This lack of availability and patience seems to increase when the child gets older and starts having a more interactive and demanding relationship with the surrounding environment, and the people in it. These mothers also reveal a great incapacity in dealing with the child's emotional needs and demands, and in coping with behavioral problems. They also mention difficulties on limit-setting, and their constant drought of patience sometimes has serious consequences on the child's basic care. Some mothers see the pregnancy and motherhood as a motivation/salvation, in which the pregnancy time and the children are part of their recovery, becoming somewhat of an extension of themselves and they feel it as a fulfillment: 'She was my life-boat, a motivation to begin my recovery' . They say that the new situation is 'a sign' and a motivation to make changes in their lives. While prior to the pregnancy they were only focused on consuming, and were overcome by the addiction, now they feel the need to think about their children, who become drug substitutes. Despite the ambivalence they feel regarding the time they have to devote to the child, and cannot dedicate to themselves any more, they believe that it is precisely that amount of time that will keep them busy and their minds off the drugs, in an efficient enough way so that they do not go through relapses. The feelings of ambivalence and turmoil disappeared in most of the cases, leaving only some remainders a few months after birth, which depend on some intervening factors, which seem to have influence on the change from an ambivalent mothering into a healthier parental role. --- Intervening factors: family support to parental role In these situations, due to the common instability involved, it is frequent for the parents of the addicted mothers to take care of their grandchildren, and provide a home so that the mothers have 'somewhere to live'. In 11 cases, grandparents were taking care of the child most of the time. The grandmother is often the primary caretaker, carrying out the mother's role: 'When she starts crying I call my mother, she feeds and bathes her' . This allows the mother more availability, however it also may have its problems, such as excessive control by the grandparents , which may result in a loss of independence and autonomy for them, a lack of space and frequently generational conflicts. Besides the support given by the grandparents, some women refer to other support given by other people . This other support is essentially financial and material, as most of these mothers are unemployed, but also includes psychological support. The partner's support is identified by these women as one of the most important types of support. Some of the partners, whether they are consumers or not, help these mothers during the pregnancy and feel happy with the child's birth, which is very important for them. This support is also provided by the trust partners place on these women, during the initial period of difficulties. Drug-addicted mothers who have their partners' emotional and/or financial support, say that it is a big help in all the processes of change, especially when their partners shared responsibilities and helped taking care of the children. Addicted partners can also be a good help when both decide to start the treatment, when they know they are going to be parents, which is a double stimulus to stop consuming drugs. However, not all partners are this supportive; some women refer to a lack of support, usually in the context of a difficult marital relationship, that results many times in separation. In these cases, there is a growing lack of communication and support during pregnancy, at the delivery moment and in helping with the child's care. If partners are consuming during pregnancy and after birth while women are undergoing treatment, their relationship becomes impossible, triggering the separation: 'When I decided to begin treatment, we both did. We had tried before, each one alone, and we had failed, it's impossible to begin treatment when one is doing drugs and the other wants to leave that life' . Without this support mothers feel a stronger social isolation, becoming less available and patient toward their child, and completing the treatment with success, as well as recovering, is more difficult. Social isolation and loneliness also arise, because substanceabusing women who stop consuming move away from their social group, as a way of avoiding relapses. Furthermore, the healthcare professionals' support is also of great importance. These mothers value doctors and nurses who try to protect, help, support and understand them, making them feel that they are good mothers, and that they know how to take care of their children. They make a clear distinction between these 'helpful professionals' and the others, who are considered incompetent and discriminating professionals, that treat them as if they are impaired in their maternal role, and looking at them differently, as people who do not deserve their attention, thus increasing the burden of social stigma: 'I'm tired of being discriminated ... from the moment I stepped in the emergencies service and they knew I was a drug addict, they treated me differently' . Other aspects that seem to transform these parents' initial ambivalence into parental healthier behavior are: starting the treatment during or even before pregnancy; consuming during pregnancy and infancy; the child's and mother's ages. As ambivalence decreases as the children get older, a healthier and more adapted parental role seems to slowly emerge. According to these results, mothers of children older than six months, who began the treatment early, before or during early pregnancy and did not have relapses, are more adapted to the parental role, and more attached emotionally and available. Although they see motherhood as a transition, they face it as a good responsibility and a fulfilled purpose. --- Discussion Drug-addicted mothers clearly reveal a strong ambivalence between dedicating themselves to their children and continuing consuming drugs, usually associated to a feeling of guilt. In those periods, primary care seemed to be assured, however motherhood seemed reduced to functional aspects. A small group, especially those who did not have social support, continued revealing difficulties, concerning social and didactic care, as well as in imposing limits. On the one hand, the results revealed that these mothers adapted to their parenting role as time went by, on the other the turbulence that most had been through could not be ignored. Facing a constant ambivalence, little room was left to fulfill their role as mothers. The Primary Maternal Care -an expression first used by Winnicott to describe a temporary phase of extreme sensibility toward the baby -seemed difficult to take place. Thereby, a central process, which we called Functional Parenting, emerged. This process is characterized by a distortion of the parenting role's characteristics that are seen merely as functional, expressed through many of the mother's behaviors . Actually, these mothers do not see Functional Parenting as something negative, since they believe that basic care is all they have to do, neglecting the possibility of a more emotionally engaged relationship. These results are consistent with other studies that identified drug-abusing mothers as less emotionally engaged and responsive to their infants , showing a less positive involvement or a lower maternal involvement . We believe our study may help to understand neglect situations that occur in these families, as well as why the child is neglected. Rather than a victim of physical or sexual abuse, the primary reason for intervention by social services in these families where one or both parents are drug addicts is neglect. The fact that these mothers attend methadone programs, may explain why we found no cases where basic care was absent. During periods of intense drug use children can be victims of different types of situations, such as improper feeding, clothing or caring , lapses in hygiene and inconsistent regard for the child's safety and supervision . In this study mothers frequently referred to the consequences of drug abuse on their children: abstinence syndrome; low weight or even prematurity; respiratory problems; feeding and sleeping difficulties; as well as crying, agitation and restlessness. However, they do not acknowledge the existence of relationship problems with their children, denoting their difficulty to consider anything beyond basic care as a main aspect in the dyadic relationship. This does not mean that the child becomes harmed, since sometimes other relatives replace the mothers in their parenting functions. The recovery attempts are so invasive, taking most of their strength and availability, that whatever is left to dedicate to their children becomes minimal. While trying to suppress their own difficulties, they ignore some of the child's needs. Not all mothers reveal the same degree of ambivalence, or the same difficulties in the transition from pregnancy to the beginning of motherhood, so it was not possible to find out what actually differentiates the mothers' behaviors. It was possible to identify some factors involved, such as the period of abstinence, if it started before or after the pregnancy; the presence or absence of a partner, kind of social support they receive in the everyday care of the child. However, it is not yet possible to understand the importance of each one of these factors. Although we found great internal difficulties in the process of becoming a mother, when recovering from drug addiction, these difficulties do not seem to have much of an external impact on parenting, in most of the mothers in methadone programs, confirming Hogan and Higgins' statement when they say that, during periods of relative stability , the impact of parental drug use on the family functioning might be minimal. Nevertheless, it seems to us, that the minimal impact on parenting occurs in situations of longer and more stable engagement in recovery programs, and mostly results from the intervening factors like social support and partner support, which seem to have a main role in the process of becoming a mother, allowing these mothers to overcome difficulties in their relationships with their children. Conceptually understood as a development crisis, motherhood, in addicted mothers, can be of extreme importance to eneehance a change, such is the opportunity that pregnancy and motherhood represent to them. With adequate clinic intervention, this moment can be of great motivation to change these mothers' consumption patterns, and their whole lifestyle, motivating them to enter a treatment program. We would like to emphasize that the results of this analysis are unique to the particular participants of this study. It was possieble to identify the factors involved in the ambivalence and difficulties in the transition to motherhood, such as the length of abstinence, if this abstinence started before or after pregnancy; the presence or absence of a male partner, or what kind of support, namely by grandmothers, do they receive in day-to-day care. However, it is not still possible to understand the importance of each factor. It would be interesting to investigate in future studies their family history, as well as the bonding characteristics experienced with their own mothers. It would be interesting also to observe parenting behavior in a longitudinal study in order to complement the retrospective approach of the present study. --- Competing Interests None declared.

The parent-children early relationship, how it evolves and its quality are mains aspects in the psychological, social and emotional development of the child. When parental behavior is not adequate and does not correspond to the child's solicitations, it can represent a risk to his/her physical and psychological health. Being a drug-addicted parent may set a highly problematic context for parenting since it may be a risky context for the child ( Barnard and Mckeganey, 2004). In recent years, there has been a growing number of drug-addicted women of a fertile age, with a consequent growth of the number of children that are exposed to drugs in the prenatal period (Pajulo et al., 2006). Drug-addicted women are more likely to have multiple, intertwined psychosocial risks that may result in maladaptive parenting practices, with possible impact on the developmental outcomes of the children involved (Derauf et al., 2007). The time spent with children is less among drugabusing parents (Tarter et al., 1993) and the risk of child neglect increases (

Background Internationally, there is a growing call to include informal carer costs and benefits in economic evaluations to evaluate the broader impacts of health and social care services [1,2]. In this context, informal carers provide care beyond normal expectations within a pre-existing relationship such as assistance with personal care, household activities or practical support, and generally do not receive payment for the care they provide [3,4]. Over the last 15 years, three carer-specific preferencebased instruments have been developed to measure outcomes for economic evaluations, the Adult Social Care Outcomes Toolkit for Carers [5], Care-related Quality of Life instrument [6] and Carer Experience Scale [7]. Whilst the number of economic evaluations including carer effects has grown in recent years, few have been conducted in the Australian setting [2,[8][9][10]. In Australia, there are about 2.8 million informal carers with over a third acting as the primary carer, i.e. the person who provides the majority of care [11]. In 2020, informal carers provided an estimated 2.2 billion hours of care, on average 786 h per year or 15 h per week [11,12]. If services were purchased from formal care providers, the replacement costs would be $77.9 billion, almost 40% of the total spending on health in Australia in the same year [11,13]. Given societies implicit reliance on carers' willingness to fulfil this role and the economic consequences if this situation should adversely change, it is crucial that carer costs and effects are considered in economic evaluations [11]. In turn, this would also better inform healthcare decision-makers on actual societal costs, increasing the chance that welfare optimising decisions are made. Two recent studies have investigated the relative construct and discriminative validity, test-retest reliability and responsiveness of the European-developed ASCOT-Carer, CarerQol and CES in a survey of Australian carers [14,15]. Studies in England have also compared ASCOT-Carer, CES, CarerQol and EuroQol-5 Dimension-5 level [14,16]. These studies indicate that the instruments tap into different constructs of carer-related QoL and caring experiences, reflecting the original purpose of the instruments and suggests the ASCOT-Carer, CES and CarerQol cannot be used interchangeably [16,17]. The ASCOT-Carer was developed to measure social care-related QoL and support of carers in the setting of policy and formal support interventions [5,18]. Whereas the CarerQol measures the impact of informal care on carers' QoL, combining the burden of caring and valuation of their well-being in the context of an evaluation in health care [6]. The CES captures the caring experience rather than carer's QoL per se [19]. Validation has been investigated for the constructs in each of the instruments: the ASCOT-Carer with carers in England [5]; the CES with carers of older people in England [19] and the CarerQol with carers in eight European countries [6] [20]. Each of the instruments has preference weights, allowing the calculation of a summary score which reflects carers' preferences for difference aspects of carer-related QoL [7,[21][22][23]. An Australian population was only included in one of the CarerQoL validation studies [22], with the general adult population using hypothetical carer scenarios. Content validation of instruments assesses relevance, comprehensiveness, and comprehensibility of the questions and the overall instrument; ensuring interpretation is as intended, all aspects important to the specific population are included and that the instrument's constructs/domains measured as proposed [24]. Cross-cultural adaptations of instruments are important to capture differences in linguistics, colloquialisms, context and culture, even if translation is not required [25]. Qualitative research is ideally placed to validate the face value and content of instruments by exploring these social and cultural variables that may differ between informal carers in different countries [26]. Given the differences in health and social care support between countries [27][28][29], it is important to evaluate the applicability of these instruments in an Australian setting [25]. Consequently, the aims of this analysis were to identify aspects of carer QoL important to Australian informal carers and explore how well the ASCOT-Carer, CarerQol and CES constructs capture these aspects in the Australian context. --- Methods --- Study design An online questionnaire was administered to a sample of informal carers in Australia between June and September 2018. Participants were recruited through Carers Victoria, a state-wide not-for-profit organisation supporting carers to improve their wellbeing, health, resilience and capacity [30]. This analysis was part of a larger study that investigated the psychometric properties of the carer-related preferencebased instruments [15] and exploratory factor analysis [17]. --- Setting and participants Adults , Australian residents who self-identified as primary, informal carers and able to read the English written study questionnaire were invited to complete a web-based questionnaire. An email invitation was sent to all Carers Victoria registered informal carers who had previously consented to contact for research purposes. The online questionnaire link was also advertised in the Voice: Carers Victoria ebulletin which is distributed to all informal carers registered with the organisation and in researcher's social media posts. Informed consent was collected before starting the questionnaire and a $10 gift voucher was offered to all participants as an acknowledgement of their contributions. --- Instruments --- Adult Social Care Outcomes Toolkit for Carers There are seven domains in the ASCOT-Carer, a preferencebased instrument of carers' social care-related quality of life including; control over daily life, occupation , social participation and involvement, personal safety, self-care, time and space to be yourself and feeling supported and encouraged [5,21]. The content of the ASCOT-Carer was developed from a literature review, focus groups and interviews with carers and care managers [31], and semi-structured interviews with carers [18,32]. --- Care-related Quality of Life CarerQol contains two sections; the CarerQol-Visual Analogue Scale , which measures wellbeing and the CarerQol-7D which measures subjective burden [6,33]. There are seven dimensions in the latter; fulfilment, support, relational problems, mental health problems, problems combining daily activities with care, financial problems and physical health problems. The content of the CarerQol was developed from a survey of carers in the Netherlands and a review of eight popular burden measures [6,23]. --- Carer Experience Scale There are six dimensions in the CES, a preference-based instrument of caring experiences; activities outside caring, support from family and friends , assistance from organizations and the government , fulfilment from caring, control over the caring and getting on with the care recipient [7]. The content of the CES was developed from semi-structured interviews with carers in the UK and a meta-ethnography of qualitative studies on caring. --- Questionnaire The questionnaire was developed online using Qualtrics®. It was piloted to refine wording and comprehension of the activities with a convenience sample of Deakin University Health Economics and Faculty of Health staff members and informal carers . Study participants had the option to complete the questionnaire over multiple sessions and all questions were voluntary. Figure 1 shows the sequence of instruments, randomisations and the question wording. Firstly, contextual questions about personal characteristics, caring situation and characteristics of the care recipient were asked. Followed by the randomised ASCOT-Carer, CarerQol and CES instruments to minimize potential ordering effects [15]. To achieve the aims of this paper, the first thirty participants were asked two separate free text questions about the greatest positive and greatest negative effect on their quality of life as a carer in the past week. To minimise survey burden, all participants were randomised to one additional activity: instrument plus free Fig. 1 Questionnaire Flow. *Positive/Negative questions: Please describe the things that have had the greatest positive effect on your quality of life as a carer in the past week?; Please describe the things that have had the greatest negative effect on your quality of life as a carer in the past week? . ** [Instrument] followed by question: Thinking about the statements included in this completed questionnaire, please describe any other things that affect your quality of life as a carer that were NOT mentioned. . *** [Instrument] and following each question: How important is this to your quality of life as a carer? . ˆCRA = Caregiver Reaction Assessment; Social Isolation = Three item UCLA Loneliness Scale; AQoL-8D = Assessment of Quality of Life-8 Dimensions text question on whether any aspects of carer QoL were not mentioned in the instrument ; instrument plus a 5-point Likert scale of the importance of each domain within the instrument s ; or complete the instrument only. Finally, all respondents were invited to rank the top five most important aspects of caring from most to least relevant . Other instruments were also completed by all participants in the final part of the questionnaire , for use in the broader project [15,17]. --- Data analysis Analyses were conducted in STATA Statistical Software: Release 17 [34], QSR NVivo software© [35] and Microsoft Excel [36]. Instruments were scored using preference-based weighting of the respective instruments. To ensure consistency, the UK value sets were used for all three instruments. Results were reported in a previous paper [15]. --- Population Descriptive statistics were generated for the demographics, caring situation and care recipient characteristics. --- Relevance and comprehensiveness --- Relevance Responses to the Positive/Negative questions and [Instrument] missing domains questions, were imported into QSR NVivo software version 11 [35] for analysis to identify specific factors influencing respondents' care-related QoL. A four-stage content analysis procedure guided the coding of the open-ended responses: decontextualization, recontextualization, categorisation, and compilation [37]. Conventional content analysis was used in the development of the coding framework and themes were inductive, data driven, and with researchers avoiding using preconceived categories [37,38]. Responses for each instrument and question were coded separately. Coder one spent time noting any preliminary ideas, codes and themes before building categories and with these, a coding structure. Where responses contained more than one theme/sub-theme they were coded into each. Coder two reviewed the coding structure and relevant-free text responses. The two coders discussed differing views on codes and discrepancies were settled by authors LE and NM before finalising the coding structure. Proportions of the [Instrument] domain importance question rated as unimportant, neutral or important to respondents' CrQoL were calculated and compared with the Chi-Square test. In addition, for the ranking activity, descriptive analyses were used to determine number of times domains were ranked number one and also the number of times chosen in the top five ranks. --- Comprehensiveness The coding structure of the analysis for questions [Instrument] missing domains was reviewed by authors JB & LE to identify any key aspects of carer QoL that participants identified as missing from the instruments. --- Results --- Population Online Appendix 1 shows the sociodemographic characteristics and caring situation of the informal carers and care recipient characteristics of the total questionnaire sample and for each of the subgroups that received and completed the additional questions that contributed to this analysis. The total sample size was 500 participants with a mean age of 52 and the mean age of the care recipient was 45. Majority were female and had completed undergraduate and postgraduate education . Just over half of the participants were employed , while the other half were retired or engaging in housework duties including caring . Just over half of participants were sole carers and a quarter providing care to multiple recipients. Most participants shared a household with the care recipient and had been caring for > 24 months . Relationships with care recipients included children , parents , partners and other family members or friends with their most common medical condition being chronic diseases or disabilities followed by mental health problems . --- Relevance The open-ended responses resulted in 244 units of data and identified eight themes: Behaviour-Mood of care recipient; Caring Responsibilities; Finances; Health; Own Life; Perceptions of carers; Relationship with care recipient; and Support. Although the qualitative positive/negative and Instrument [missing domain] question responses were coded separately, strong similarities between the coding frameworks were very apparent with the same themes present for each question, differences only in sub-themes. Each question appeared to elicit responses that indicated what matters to carers, so the results have been presented together in Table 1 which summarises the themes and sub-themes presented in Online Appendix 2. There was a wide diversity in the openended responses. The number of responses ranged N = 24-33 for each open-ended question . Likert responses to the importance of each domain were categorised into: Not Important , Neutral and Important and aggregated within each instrument. The total instrument importance shows that carers in our sample judged all three instruments as important and ≤ 7% not important. There was no statistically significant difference between the ratings across the three instruments . Similarly, the importance of individual domains within each instrument indicated that all the domains for each instrument were important aspects of CrQoL . Domains most frequently considered important were the CES domain of getting on with the care recipient and activities outside of caring and CarerQol's Mental Health Problems . Domains deemed least important were for ASCOT Carer's personal safety , and control over daily life and CarerQol's financial problems . Table 3 summarises the instrument domain ranking activity. The top five ranked domains were the same using either analysis method , only the third and fourth positions were reversed. --- Comprehensiveness After completing the instrument, very few participants reported that the carer-related instruments covered all aspects affecting care-related QoL , CarerQol and CES . Content analysis of this question for each instrument shows that all themes in the coding framework were identified as missing by our sample of carers. This is a particularly interesting result, as many of the themes are constructs measured by the instruments. --- Discussion This analysis identified aspects of carer QoL important to Australian informal carers and explored how applicable the constructs of ASCOT-Carer, CarerQol and CES were to this population. Behaviour-mood of care recipient, Caring Responsibilities, Finances, Health, Own Life, Perceptions of carers, Relationship with care recipient and Support were identified as aspects of caring that affect carer QoL in Australian. Comparing domains most importance in our sample with previous studies that developed preference weights and tariffs for the instruments showed mixed results. Occupation and control over daily life for the ASCOT-Carer where the most preferred among English carers [21], whereas our Australian sample found self-care and time and space to be yourself to be the most important. Suggesting that, with further investigation, Australian preference weights for the ASCOT-Carer may be in need of development. In our sample the importance of domains in the CarerQol and CES were in line with instrument tariffs [7]. CarerQol tariffs, developed for Australia, indicated mental health and combining caregiving with other activities as the most and least preferred [22,23]. Whilst in a sample from United Kingdom, the CES found activities outside of caring and getting on with the care recipient as most preferred and control over caring the least [7]. Almost all participants reported aspects of carer QoL not captured by the carer-related instruments. Many of these aspects that were perceived as not covered by the instruments, could have been included in the domains. This was also the case in a study looking at patient, self-reported, QoL aspects not captured by EQ-5D-5L [39]. As caring experiences are subjective and responsibilities and challenges vary greatly between carers, they may have felt the domain did not completely encompass their experience of carer QoL. This reflects how some aspects of QoL, which can be important to individuals, cannot necessarily be translated into a question for a QoL instrument. Particularly when required to be applicable to a broad range of carers , align with the construct of the instrument, and also fit with other considerations . Alternatively, respondents may have interpreted the questions differently or focused only on certain portions of the question . A content analysis of the end-of-life patientreported outcome measure showed that interpretation of questions is related to individual circumstances, where in financial matters varying themes of money, investments, funeral arrangements and wills emerged [40]. This could also explain why similar domains across instruments were treated differently. The domain of support is present in all three instruments, however, respondents reported different types of support were missing in each instrument . Comparable results were also found in more detailed studies of the exploratory factor analysis using this same dataset, where only a moderate correlation was for my son so that I can attend a meeting' 'There is no support for carers that I know of. In the early days after my husband's stroke there was some support, but everything is capped at a certain amount of visits. And after that you don't meet the criteria anymore so you're on your own. I have not had a proper break from my caring duties for more than 2 years. I care for my husband 24/7. No one works those hours in paid employment. I love my husband and wouldn't have it any other way but to have support and know help is only a phone call away would be great.' 'When we look like caregivers who look like they have all the bases covered, you receive less support' 'this program has now been defunded so I won't have the opportunity to meet with the other carers monthly/bimonthly and the support worker has lost his job so I will no longer have him as a support/source of assistance with my caring responsibilities' 'Frustration with Government Departments and Caring Organisations who because they are underfunded, over committed or don't/can't do their jobs effectively. I am constantly told they can or will help then simply do not carry out their promises and or don't follow up as promised' *Government financial assistance coded to 'Support' theme 1 3 found between CarerQol and CES support items and also for relational problems [17]. Content and/or face validation of the three instruments has not been performed with Australian carers, so detailed information of how each instrument's questions are interpreted and understood by this population is not known. The broader project performed a content comparison of the three instruments showing they each perform well in measuring their relevant domains with Australian carers [15,17]. However, the qualitative component of this study suggests that some differences in question interpretation may exist. The majority of sub-themes related specifically to the care recipient. However, only two domains include aspects of carer QoL that are influenced by the care recipient . Consideration was given to a similar domain in the development of the ASCOT-Carer, however, it was omitted because it did not fit with the construct of the instrument [18]. This relationship between the care recipient and carer has previously been proposed as an advantage of the CES in capturing broader aspects of caring [7,15,16]. Given the possible interdependence of care recipient and carer QoL [41][42][43], instruments capturing both could be included in economic evaluations of carer and patient interventions to fully capture the effects of an intervention [44]. However, consideration also needs to be given to the type of evaluation being performed, the perspective taken and the possibility of double counting which could overestimate the benefits of an intervention [2,45]. Keeping these factors and participant burden in mind, an appropriate combination of instruments may be used in measuring carer QoL in studies focused on informal carers. Some of the missing aspects of carer QoL in the instruments, as reported by our sample, are intentionally not covered by the instruments as they each have been developed with different intentions and measure different constructs of CrQoL. The CarerQol was developed and intended to measure the impact/burden of caregiving on QoL and so, appropriately, does not include any themes specifically about the care recipient [6]. Similarly, the CES missing themes of finance and health are reasonably missing as the instrument's purpose is to measure the experience of caregiving. Health problems are not directly measured by CES, as qualitative research indicated that this was linked to other attributes included in the instrument [19]. And the ASCOT-Carer does not measure finance and health, as the instrument was developed as a measure of social care and support services on carer QoL. Financial hardship due to caring and health were considered in the early development of ASCOT-Carer [31], however, it was excluded as it was outside the scope of the instrument's purpose. Although health was not considered as a separate domain in the ASCOT-Carer it is captured by the lowest QoL response option for each item and indicates that the carer has high-level needs that, if unmet over time, put the carer at risk of poor physical and/ or mental health. --- Strengths and limitations Content analyses have the potential to be influenced by researchers' experiences and preconceptions. Coding framework along with transcripts were reviewed by a second researcher independently and collaboratively discussed, reducing the impact of coder bias. Quotes and sub-themes were classified to themes based on consensus and the coding framework is presented to demonstrate how the data were categorised so that other researchers can consider how their interpretation aligns with the researchers' views. The recruitment of study participants and completion of the questionnaires occurred towards the end of the roll-out of a new government support system, National Disability Insurance Scheme , which replaced the existing system of 1 3 disability support. The NDIS caused changes to administrative processes in receiving financial and formal support and may have been particularly front of mind for carers having to navigate this new system. Further, specific issues may have arisen directly due to changes in the systems. The cohort included a greater number of sole carers , a greater proportion of female carers and a higher percentage of carers providing more than 30 h of care per week compared to the Australian population of primary carers [11,12]. Income and employment, relationship to care recipients and sharing household with care recipient were similar to the Australian population of carers [12]. Uniquely, participants included carers of people with multiple health conditions. The study included informal carers in the Australian setting only and therefore results may have limited generalisability to other settings. Study participants were self-selected via newsletter advertising and, to reduce burden, randomly allocated to one additional activity described in this paper . This resulted in a different sub-group completing each . Subgroup characteristic differences include the Positive/Negative question participants containing only females, being less employed, less likely to be married and more likely to have been caring for > 24 months and > 4 h of care per week and the CES missing domain participants being less educated and caring for more recipients with mental health problems. The open-ended components of this study were embedded in a larger quantitative study [15,17], so there was no opportunity to apply qualitative techniques such as faceto-face interviews or focus groups to explore responses in more depth, clarify the views' expressed or to measure comprehensibility of the instrument questions. Cognitive interviewing, in checking respondent's understanding, mentally processing and response to materials would help with understanding these differences. Cognitive interviewing evidence is present for the ASCOT-Carer in England [18,31,32] and during its translation into German [46,47], Japanese [48] and Finnish [49], as well as for the CES in England [19]. Australian evidence would provide a much greater understanding of the comprehensiveness and comprehensibility of the three instruments in this setting. As questionnaires were completed anonymously, researchers did not have an avenue to discuss findings with participants and receive feedback on the themes and analysis. Importance of domains in this study, were assessed by a sample of informal carers. Whereas carers also participated in the development of preference weights for the CES using a best-worse scale valuation exercise [7], the ASCOT-Carer and CarerQol used the general population imagining a hypothetical state of being an informal carer using BWS exercise and a discrete choice experiment respectively [21,22]. The inconsistencies with our sample may be due to the differences in sample and also between stated different methodology using preferences or revealed preferences or due to different analyses. Three different approaches were used to investigate how well the instruments capture aspects of CrQoL important to Australian informal carers, strengthening conclusions concerning coverage. Some potential cultural/ethnic difference appear in the free text responses. However, with a very small number of participants born outside of Australia and/or speaking a language other than English these differences were not explored in this paper. A real opportunity exists for future research in this area of an Australian population. --- Conclusions Open-ended responses and quantitative data collected from a sample of Australian informal carers, suggest there are multiple aspects of caring that impact carer QoL. Consideration should be given to measurement of care recipient health and well-being and spillover effects affecting carer QoL, with thought to the risk of double counting. The ASCOT-Carer, CarerQol and CES appear to be relevant for an Australian informal carer population and include most of the aspects of quality of life important to them. The interpretation of questions may differ in Australian informal carers which requires confirmation with a content and/or face validity assessment. The findings support previous research that the selection of an instrument should take into account the aim, purpose and constructs of the instrument. --- --- --- Ethical approval The questionnaire and methodology for this study was approved by the Deakin University Faculty of Health, Human Ethics Advisory Group, Burwood, Australia and through the Carers Victoria's Carer Participation in Research process. Consent to participate Informed consent was obtained from all participants before commencing the questionnaire. Consent to publish Patients read and accepted, informed consent regarding reporting of research findings in a de-identified format. 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:// creat iveco mmons. org/ licen ses/ by/4. 0/. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Purpose Identify aspects of quality of life (QoL) important to Australian informal carers and explore how well the Adult Social Care Outcomes Toolkit for Carers, Care-related Quality of Life instrument and Carer Experience Scale capture these aspects in the Australian context. Methods Online questionnaires were completed by Australian informal carers. Socio-demographics, open-ended questions: positive/negative aspects of caring and QoL aspects missing from the instruments, and ranking of the instrument domains was used to explore the content of the instruments. Instruments were scored using preference-weighted value sets (reported in another paper). Content analysis was used to analyse the open-ended responses. Chi-squared test looked at differences in domain importance. Descriptive analyses summarised all other information. Results Eight themes were identified: Behaviour-mood of the care recipient, Caring responsibilities, Finances, Health, Own life, Perception of carers, Relationship with care recipient and Support. Many aspects of carer QoL mentioned as missing in the instruments appeared covered by the domains, of which all were reported as important. The highest ranked domain was relationship with the care recipient. The influence of the care recipient specific support, behaviour/mood and health on carer QoL appear absent in all instruments. Conclusion The content of the three instruments appears relevant in an Australian setting. The influence of care recipient's health and well-being on carer QoL should be considered, along with spillover effects. A content and/or face validity analysis is required to confirm differences in item interpretation in Australian informal carers.Informal care • Outcome measurement • Carer-related quality of life • Preference-based measures * Jessica Bucholc

Background The role of high depressive symptoms and depression as risk factors of substance use is well established 1 . Numerous studies have shown that individuals with depression are also more likely to use a wide range of substances 2 . The bidirectional link between substance use and depression suggests that substance use may also operate as a risk factor for depression 3 . However, very little research has explored racial differences in the links between depression and high depressive symptoms and smoking 4 and vaping electronic cigarettes 5 . Between and within racial group heterogeneities may exist regarding the comorbidity between smoking and depression 6 . People from diverse demographic backgrounds may exhibit distinct motivations for substance use or presentation of depression 7 . Black and White individuals may also differ in smoking behaviors 8 , depression 9 , and how they cope with stress 10 , as well as likelihood of seeking care if they are depressed 11 . For example, some research suggests that Black and White individuals may differ in the connection between depression and smoking cigarettes or vaping e-cigs 12 . While the overall connection between depression and smoking 6 and vaping are known, few studies have explored Black-White differences in this regard. Furthermore, while cigarette smoking is declining 13 , e-cigs popularity is increasing 14,15 , thus there is a need to investigate not only conventional cigarettes but also e-cig use, as correlates of depressive symptoms 16 . Given the unique and different characteristics of users of e-cigs compared to conventional cigarettes 17,18 , some results may apply to conventional nicotine delivery systems but not to other alternatives 19 . --- Aim Built on our past work on the bidirectional links between depression and substance use in Black individuals 20 , we conducted this study to fill the gap in the literature. Our study had two aims: to test the association between depression and smoking conventional cigarettes and vaping e-cigs and to explore group differences in these associations. Given higher chronicity of depression in Black communities 21,22 , in part because of low health care use 23 and high stigma 24 , we hope to gain a deeper understanding of how racial group membership impacts the need of diverse racial groups with and without depression for cessation programs for smoking conventional cigarettes or vaping e-cigs. The findings of this study may have significant implications for tailoring interventions and prevention strategies that simultaneously address smoking/vaping and depressive needs of diverse populations. --- Methods This cross-sectional study utilized data from the Health Information National Trends Survey . HINTS is a nationally representative survey conducted in the United States designed to assess various aspects of health communication, cancer knowledge, health care use, smoking behavior, e-cig use, and depressive symptoms 25 . The survey is administered through computer-assisted telephone interviews, ensuring a standardized data collection process 25 . The study included a representative probability sample of US adults obtained from the HINTS survey. The sample consisted of both men and women, covering a diverse range of demographic characteristics. The participants were selected through a multistage sampling design, incorporating both landline and mobile phone numbers to ensure adequate representation across the population 25 . Our analytical sample was composed of all HINTS participants in the year 2022, regardless of their race/ ethnicity and smoking status. The variables of interest in this study included race/ethnicity, smoking status, e-cig use, depressive symptoms, demographic, and socioeconomic status characteristics. race/ethnicity were self-identified and as follow: Non-Hispanic White , Hispanic White, Non-Hispanic Black, and other/mixed/multiple race. Other/mixed/multiple included any group which was not White or Black. Smoking status was assessed using self-reported measures, distinguishing between current smokers and non-smokers. Separate questions were asked about conventional cigarettes and e-cigs. E-cig use was evaluated by identifying individuals who reported current or ever use of e-cigs 26 . Depressive symptoms were measured using the Patient Health Questionnaire-4 27 , a validated screening tool that assesses symptoms of depression. A threshold of 3 or more was considered mildto-moderate, and a score of 7 or above was considered severe depression 28 . Race/ethnicity was the moderator. Age, gender, and SES indicators were included as covariates in the analysis. Statistical Package for the Social Sciences software was utilized for data analysis 29 . Univariate, bivariate, and multivariable models were employed to examine racial differences in the variables of interest. For multivariable modeling, binary logistic regression models were employed to assess the associations between depressive symptoms and smoking conventional cigarettes or e-cigs. We ran binary logistic regression models overall without and with interactions terms between depression and race. The models were adjusted for covariates, including age, educational attainment, marital status, and employment. Smoking/vaping and depression were the outcome and predictor variables, respectively. Odds Ratio, 95% CI, and p-values were reported. Statistical significance was set at p < 0.05. --- Results Table 1 summarizes the data by demographic factors, socioeconomic factors, and ever and current use of e-cigs and conventional cigarettes. 6154 participants entered our analysis. Participants ranged in age from 18 to 99, with a mean age of 55.60 . Table 3 shows the summary of the logistic regressions for the associations between depression and ever and current use of e-cigs and conventional cigarettes. While depression was associated with higher odds of ever and current use of conventional cigarettes, the association between depression and smoking was dependent on race/ethnicity . Regarding the magnitude of the difference in the effect sizes for Blacks vs Whites, for eversmoking, on average, this effect was 1.637, and for Whites, this effect was 1.519, and for Blacks, this effect was 1.78 larger than Whites. This magnitude of difference was more for current cigarette use (2.275 higher for Blacks than Whites, 2.048 on average, and 1.659 for Whites. Table 4 shows the summary of the logistic regressions for the associations between depression and ever and current use of e-cigs. While depression was associated with higher odds of ever and current use of e-cigs, these associations did not depend on race. --- Discussion The findings of this study revealed that depression is associated with smoking conventional cigarettes and e-cig use. We also found moderating effects of race-specifically, there is a stronger association between smoking and depression for Black adults compared to White adults. The results of our first finding is consistent with previous research that has suggested a link between smoking/vaping and depression 6 . There are, however, very few studies on group differences in the association between smoking and depressive symptoms. One study reported a stronger link between depression and smoking among women compared to men 6 . We are unaware of any past work on racial variation in the comorbidity between depression and tobacco use in forms of smoking or vaping. The stronger association between depression and smoking in Blacks than Whites may be because depression is more chronic in Black communities. There are several reasons for higher chronicity, severity, and consequences of depression for Black individuals compared to White individuals 21,22 . First, systemic racism and social inequality can significantly impact the mental health of Black communities. Structural and institutional racism, which limit access to quality healthcare including but not limited to depression treatment within Black communities, along with higher rates of poverty, may collectively contribute to increased stress and lower overall chance of depression treatment in Black populations. Chronic stressors can exacerbate depressive symptoms and make treatmentseeking more challenging for Black individuals. As a result, Black individuals with depression are less likely to seek care and receive adequate mental health care 23 and treatment of depression within Black communities 23 . Stigma surrounding mental health issues can lead to underreporting or misdiagnosis of depression, as seeking help may be viewed as a sign of weakness 24 . Traditional cultural norms and expectations of strength and resilience may prevent individuals from seeking necessary support and treatment, leading to a chronic and untreated state of depression 24 . Trust is also lower toward mental health treatment in Black communities. Additionally, there is evidence to suggest that Black individuals may experience racial discrimination in the healthcare system, which can negatively impact their mental health treatment outcomes. Finally, disparities in access to quality mental health care can also contribute to the increased chronicity and larger consequences of depression in Black than White communities. Lack of culturally and structurally competent mental health services, shortage of Black mental health professionals, and disparities in insurance coverage and affordability of care also reduce the chance of receiving timely and effective treatment for Black patients. These combined can result in delayed or prolonged periods of untreated or under-treated depression, leading to a more chronic and severe course of the depression for Black communities 21,22 . As a result, we may observe higher prevalence of depression in Black smokers. This study bridges the gap in the current literature by documenting universality of the associations between conventional cigarette smoking, e-cig use, and depressive symptoms. The results pave the way for universal approaches to effectively address smoking and depression in clinical and public health settings. Lack of race, ethnic, or gender differences identified in this study may have important implications for approaches to addressing smoking cessation and mental health interventions for people across all groups 30 . Understanding that Black people are more susceptible to the adverse effects of depressive symptoms on smoking or vice versa can inform tailored interventions for Black populations. It is essential to develop tailored strategies that address depression comorbidities in smokers of White and Black people 30 . Additionally, research on the stronger link in Black people is needed as well. To further enhance our understanding of relationships between smoking, e-cigarette use, and depressive symptoms, we explored racial variation in this regard. Our findings illustrate this relationship to be non-linear, nuanced, and complex. The results showed an increased need for addressing smoking conventional cigarettes and vaping e-cigs among adults with high depressive symptoms among all groups. However, the need to address smoking in depressed individuals is more prominent for Black than White people. That is, Black adults with depression may be at an even higher risk for smoking than their White counterparts. Several directions for future research in this area can be pursued. Longitudinal studies, for example, are needed to explore the temporal dynamics and causal relationships between these variables, shedding light on how changes in smoking behavior and depressive symptoms can influence each other over time. Secondly, mechanistic studies should delve into the biological and social mechanisms underlying the observed links. Additionally, exploring the intersectionality of race, ethnicity, and gender with socioeconomic status, and sexual orientation can provide a more nuanced understanding of the associations. Furthermore, identifying mediating and moderating factors, including stress, social support, coping strategies, self-esteem, and body image, can illuminate the mechanisms through which the smoking-depression relationship emerges. Given the limited research on group differences in the association between e-cig use and depressive symptoms, future studies should focus on examining this relationship in depth, considering group differences in motivations, perceptions, and patterns of cigarette smoking and e-cig use. Incorporating diverse cultural contexts and populations in research can also shed light on the influence of cultural norms, values, and social factors on the relationship between smoking, vaping, and depressive symptoms. By addressing these research directions, we can advance our knowledge and develop targeted interventions, policies, and public health strategies to reduce the burden of smoking-related risks and promote mental well-being among diverse populations. This study has a few limitations that should be acknowledged. Firstly, we relied on self-reported measures of depressive symptoms as well as tobacco use, which are subject to recall and measurement bias 31 . Additionally, due to the cross-sectional study design, our ability to establish causal relationships between depression as a risk factor and tobacco use as an outcome is limited 32 . Additional studies with longitudinal design are needed to provide a more comprehensive understanding of the complex dynamics between smoking, vaping, and depression across various social and demographic groups. Furthermore, the study focused on U.S. adults and the generalizability of the findings to other age groups is limited. In addition, current users of e-cigarette group had a small sample size. This may have contributed to no observation of race x depression effect on e-cig use. --- Conclusions In conclusion, this study expands the existing literature on the association between smoking, vaping, and depressive symptoms. The findings suggest that Black adults may be more vulnerable to smoking if they are depressed. This observation emphasizes the need for race-based tailored approaches to promote mental health of men and women who smoke cigarettes or enter cessation programs. Future research should explore the underlying social, behavioral, psychological, economical, or biological mechanisms driving racial variation in the associations observed here. It is also unknown why Black-White differences in the association can be seen for conventional cigarettes but not for e-cigs. For e-cig users, universal programs may be better than tailored interventions because various racial groups may similarly need depression treatment if the individual uses e-cigs.

Background: While previous research has established the role of depression as a correlate of tobacco use, limited research has been conducted on potential racial variations in these associations. Aims: This study had two objectives: (1) to investigate the associations between depression and smoking conventional and vaping e-cigarettes, and (2) to explore the role of race in modifying these associations. Methods: Data from the Health Information National Trends Survey (HINTS-2022) survey, which is a national probability sample of US adults, were analyzed. Variables of interest included lifetime and current smoking status, e-cigarette use, depression (PHQ-4), socioeconomic status (SES), and demographic characteristics. Binary logistic regression models were applied to assess the associations between depression as the independent variable and current and ever smoking conventional cigarettes and vaping e-cigarettes as dependent variables, adjusting for covariates (age, gender, education, marital status, and employment). We tested potential interactions between race/ ethnicity (non-Hispanic Whites, non-Hispanic Blacks, Hispanic Whites and any other/mixed race/ethnicity) and depression on our outcomes.The study findings indicated significant associations between depression and use of conventional cigarettes and e-cigarettes. Although individuals who were experiencing depression were more likely to use conventional cigarettes (p < 0.05) and e-cigarettes (p < 0.05) compared to those non-depressed, the positive associations between depression and lifetime and current smoking were stronger for Black than White individuals.This study suggests that depression may have a more salient role in smoking cigarettes and e-cigarette use for Black than White adults. It is unclear if Black adults are more likely to smoke if they are depressed, or if Black adults who smoke cigarettes are more likely to be depressed because of higher stigma and lower access to mental health care services.

Sears, the pageant coordinator, moves toward the microphone. 1 She welcomes the audience and explains the contestants' weeklong schedule. The list of places holds little interest for most people, as they carry on chatting to their friends. For those involved in the pageant, however, the activities hold a different meaning: a young girl who learned how to say "hello" in Yup'ik during a trip to the Noel Wien library, an elder who shared memories about life growing up in his village during the Fairbanks Native Association luncheon, or a young person struggling to overcome loss who asked for advice at the Presbyterian Hospitality House. The casual observer does not see the hours of community service events that have been carried out all week, work that has given these five women the opportunity to "learn leadership by doing leadership." 2 The Miss World Eskimo-Indian Olympics pageant is a national contest that takes place annually in Fairbanks, Alaska, within the Indigenous Olympics. WEIO started as a local celebration and continuation of arctic games-such as the Ear Pull, the One-Foot High Kick, and the Nalukataq -and now attracts audiences in the thousands. 3 Each July, athletes and dancers travel from across the state of Alaska and as far away as Greenland to come together and compete in the Indigenous games. The Olympics are organized in heats, with many finals happening in the evenings. Some of the categories are separated by gender, such as those in which competitors would be advantaged by height. 4 The evenings start with traditional song and dance performances from different Alaska Native cultural groups, as the games are presided over by a "lamp tender," usually a male or female elder. Within this culturally based activity, since its inception, there has been a contest to choose a woman to act as an ambassador during the games and for the subsequent year. 5 Starting in 1961 with seven participants, the beauty contest took on the standard Western form of the time: women parading in black one-piece swimsuits, voiceless, with the winner chosen by the cheers of the crowd. 6 Bud Hagberg and Frank H. Whaley created the contest at the same time as organizing the Olympic Games. They put word out to the villages, asking them to send in their "queens" to compete along with dancers and athletes. 7 The idea was not to disenfranchise women from competing in the games; women were as much a part of the games as men. 8 Nor was this a bid to cast women as the "culture bearers of the Nation," as was the notion in other Indigenous pageants. 9 Transmitting culture was at the forefront of WEIO, as the organizers were trying to promote cultural revitalization through the games, the dancers, and the values that WEIO promoted. Both women and men are dressed in traditional regalia and take part in the dance groups. 10 While we do not know why Hagberg and Whaley only created a female contest, we do know that Miss WEIO was started as a venue to honor a woman as the figurehead of the games. 11 Within five years, the Western value systems of measuring beauty were abandoned after Daphne Rylander Gustafson, the winner from 1962, took over the running of the contest. It was from that point onward that a focus on community service and leadership started as the contest evolved into a cultural pageant. 12 Today, the focus is no longer on voiceless beauty but on strength and leadership, with the women's "talent" now focusing on a contemporary topic relating to the needs of the villages. 13 Most of the Miss WEIO contestants grow up attending the annual Olympics event, and several regularly win the games as athletes. Before the athletes arrive and the games begin, however, the Miss WEIO contestants travel to Fairbanks to start their six-day program of community service activities. It is these behind-the-scenes activities that will be the focus of this article. Pageants are highly contested spaces. Scholars and media critiqued pageants' objectification and gender socialization during the second wave women's liberation movement and pageants' exotification of women of color during the latter half of the twentieth century, and they continue to criticize pageants' commodification and sexualization of children today. Miss WEIO started as just another swimsuit pageant in the 1960s and quickly evolved as Alaska Native women took control of the contest. Today, Miss WEIO is a space for Indigenous women to have a voice. Miss WEIO empowers Indigenous youth as a role model and illustrates to the outside world that Indigenous peoples are contemporary peoples through the social justice issues raised by the participants. While there are problems that need to be overcome in all pageants-how the pageants are judged, the danger of women being objectified, women being cast as biologically and culturally reproducing the nation yet not politically able to do so-Miss WEIO has created and molded an Indigenous space to celebrate Indigenous women on their own terms. This research, therefore, explores this adaptation of a traditionally objectified women's space into an arena for community leadership. Within Indigenous women's leadership theorists from Canada and America, there is a growing body of literature focusing on well-being and healing through women's communitybased leadership. 14 As Verna Kirkness states in her foreword to Living Indigenous Leadership, "Leadership . . . is not the purview of the educated or the elected. . . . [We need to] rethink leadership, to give thought to the original philosophies and practices of our people and to give voice to these invisible leaders." 15 In other words, leadership is practiced by more than just politically elected people. In fact, the very notion of a centralized, self-nominated, and elected political leader is a Western construct, with Indigenous communities traditionally having many situation leaders for different purposes who are appointed by the community, with women oftentimes acting as the final gate-keepers of decisions. 16 While it is problematic that the political and the community have been gendered into male and female leadership, this article aims to illustrate that Indigenous women's leadership takes a holistic look at community well-being. Part of the reason why Indigenous women's leadership literature focuses more on community than politics is the lack of women in political leadership positions in the United States and Canada. 17 Indigenous nations have had female presidents, unlike the United States. However, the patriarchal systems of repression, that have been generated from federal government policies such as the boarding schools, have caused many nations to reinforce the gendered spheres of leadership, resulting in women struggling to be elected into positions of power. 18 Historically, women in Indigenous nations have held positions of power that are often higher than positions held by males; however, once US systems of government became commonplace in Native nations and elected positions became the norm, women became relegated to community leadership. 19 Jennifer Nez Denetdale writes about this problem in relation to Indigenous pageants when she argues that the very notion of the Miss Navajo Nation pageant relegates women's bodies to both biological and cultural reproduction while simultaneously disallowing political leadership, a position that Diné history warns against. 20 This issue is complex, and a thorough analysis of all Native nations' pageants would have to be conducted to see if this phenomenon occurs in every Indigenous nation across the United States. For those pageants that are linked to a cultural event such as a powwow or the Olympics, however, the political is entwined into the cultural. Within Alaska, arguably the gendered roles of leadership are broken, with women and men partaking in both community and political leadership. The role of "culture bearers of the Nation" is not solely reserved for women, as all genders embrace culture in their leadership. 21 This article will therefore focus on this use of this cultural leadership. Wilma Mankiller offers us a model for Indigenous women's leadership that centers the community at the heart of the decision-making process and provides a way to explore Indigenous youth leaders. 22 Although Mankiller's model was based on political leadership, this article aims to illustrate that it works equally well for community leadership. While the Miss WEIO contestants are not necessarily entering the pageant in preparation for political leadership, 23 there are similarities in the way women lead, regardless of whether the aims are political or community based. The winners of Miss WEIO all share a common feature: they want to improve well-being in their communities, and they view the pageant as an opportunity to do so through the networking and subsequent financial support available to title holders. This article therefore analyzes existing leadership literature by exploring the case study of community service events featured in the Miss WEIO pageant contest between 2011 and 2013. 24 This study combines Mankiller's applied theory of Indigenous women leaders with the community well-being model and Alaska Native values to explore how the Miss WEIO pageant promotes community-based leadership that can subsequently be used to achieve community well-being. 25 While it is not within the scope of this article to give an in-depth analysis of the pageant as a gendered space based on Western constructs , it is worth noting that this pageant takes place within a larger pan-Alaska Native cultural activity within which men, women, and children act as culture bearers and role models and therefore must maintain sobriety. 26 The research that follows begins with an overview of Mankiller's community-first leadership model as it relates to the Miss WEIO contestants, gives a definition of community well-being, and introduces the community well-being model. It continues with the importance of Alaska Native values in Alaska Native society. The community service events from the Miss WEIO pageants between 2011 and 2013 are subsequently analyzed to demonstrate that the contest provides an invaluable opportunity for experiential leadership, as the participants are placed in situations in which they are a voice for their people, communicating both crossculturally and cross-generationally. --- Community Well-Being Mankiller explained that there are philosophical and policy differences that distinguish Indigenous leaders from non-Indigenous leaders. Women in general, she stated, have an interconnected approach to leading in governments. For example, women perhaps link the economy to health and to children; their leadership style is collaborative and based on teamwork. Differentiating between Indigenous and non-Indigenous women's leadership styles, Mankiller explained that while non-Indigenous women recognize the interconnectivity of issues, they do so with the "family unit" or the "individual" as a central focus. In comparison, Indigenous women focus holistically on issues relating to whole community health for this generation and the generations to come: community well-being is the central focus. 27 Fig. 1. The six indicators for measuring and monitoring well-being A project led by an international group of scholars and researchers to monitor the wellbeing of the people living in the Arctic resulted in a report titled Arctic Social Indicators. The report suggested six separate indicators for measuring and monitoring well-being. These indicators were arranged into broad topics : health and population; material well-being in the Arctic; education; cultural well-being and cultural vitality; contact with nature; and fate control. Out of these six, learning your heritage language was critical to maintaining well-being in two categories, cultural well-being and cultural vitality, and fate control. The culture team identified the element of maintaining cultural identity as important for well-being. Defining culture further, the report identifies seven dimensions: "language ; knowledge ; communication ; spirituality, such as religion and ritual; sociocultural events and media; economic and subsistence practices; social organization, institutions, and networks." 28 These dimensions are mirrored in the pageants of Alaska, as the contestants must be confident in their knowledge of Alaska Native cultures. --- Fig. 2. The community well-being model for leadership Thinking about Mankiller's ideas on the interconnectivity of women's leadership and her community-first approach to Indigenous women's leadership and reflecting on how we can measure well-being in communities, I propose the community well-being model as a model for leadership, promoting and understanding well-being in any community. 29 The community wellbeing model accounts for the interconnectedness of language, ceremony, place, and sacred history, otherwise known as the "peoplehood matrix." 30 For example, in the ideal situation, the community's heritage language, part of their culture, would be spoken in the formal and informal education systems, the local shops , and in health facilities . The option for children to learn their heritage language would tie into fate control, having agency over one's future. Ceremonies that are practiced by the community and that can be used to restore balance to the earth might also be practiced in one's heritage language. Thus, language becomes integral to all aspects of the community well-being model. In the same way, land, sacred history, and ceremony are all interconnected to language and to the six factors needed for a healthy community. If we look at the examples from the six-day program of community service activities that the Miss WEIO contestants participate in, then we can see the community wellbeing model enacted in both their conversations with the public and their social justice platforms. --- Experiential Youth Leadership Within the Miss WEIO pageant, there are several contestants for whom the pageant is merely the next step in their ongoing leadership experience. In my interviews with former Miss WEIOs, the women described their previous leadership positions and experience, such as being the ambassador for the Institute of American Indian Arts, a student government member, an attendee at the Alaska Federation of Natives , and the keynote speaker at the AFN Conference. 31 Carole A. MacNeil, national director of the 4-H Youth in Governance Initiative, coined the term "youth in governance" to explain that "youth leadership development" is problematic because it assumes that leadership is a future goal and does not consider those youths who are currently serving as leaders. MacNeil explains that "youth in governance" is a better phrase, as it accounts for the development process as leadership in progress. 32 The Miss WEIO pageant is thus best understood as a week-long youth in governance program. The Miss WEIO organizers arrange a dense schedule of events, including orientation; speeches and photographs with tourists at the Riverboat Discovery and the Alaska Salmon Bake restaurant; a photo shoot by a local photographer ; interviews at two local radio stations, KFBX and KIAK; a visit and questions with patients at the Denali Center; the chance to serve food to and learn from elders at the Fairbanks Native Association luncheon; a question-and-answer session with the local children at the Noel Wein Public Library; a chance to meet the WEIO board members at a potluck; a chance to share experiences with youth at the FNA Graf Rheeneenhaanjii; an opportunity to take part in the WEIO opening ceremony; a chance to formally introduce and share one's platform with Fairbanks locals at the Golden Days senior citizens lunch; a meet and greet with residents and the mayor of the North Pole at the North Pole hotel; an opportunity to give advice to troubled youth at the Presbyterian Hospitality House; a meet and greet with customers and staff at GCI phone company ; the WEIO coronation; and a classic car ride during the Golden Days parade. The youth who enter Miss WEIO are repeatedly given the opportunity to learn through this array of community service events, which have been chosen by the Miss WEIO committee to develop the contestants' confidence and communication skills and prepare them for the judged portions of the contest. 33 According to MacNeil, "Learning leadership happens experientially, through involvement in opportunities to practice the skills, experiment with approaches." 34 Carmen Sears, who has been planning the Miss WEIO pageant since the late 1990s, creates opportunities for the contestants to gain experience. She explained that she incorporates community events into the pageant because the contestants "come from smaller villages" and are unable to practice leadership skills in front of large audiences. 35 Flora Roddy, who has been working with Miss WEIO for the past twenty years, confirmed that the development of skills and confidence in the contest is clearly visible by the end of the week. 36 The pageant organizers choose events in which the contestants can have an impact in the community. While the first few days introduce the contestants to the Fairbanks community, the judged portions of the contest occur in the latter part of the week, thus separating out the community service events from the competition events. 37 Youth who practice leadership must have authentic experiences. 38 The pageant week provides such experiences at the community level, enabling the contestants to speak to people from the area. This experience enables the contestants to network with the Fairbanks community in an authentic way, furthering their leadership positions by becoming ambassadors in and for the Fairbanks community . The chance for the young women to consolidate their knowledge and sharpen their communication skills in front of large crowds is paramount to the duties of young leaders in Alaska, who must explain the complexities of village life to people unfamiliar with subsistence living. 39 The pageant does more than just allow the contestants a chance to practice public speaking skills. The contestants are asked political, cultural, and personal questions by diverse community members over the course of the week; thus, they gain experience "where they have voice, influence, and decision-making power" in the answers that they give to the public. 40 The women who enter the pageant gain experiential learning that cannot be replicated through theoretical knowledge . This experience gives the contestants a voice and direct influence in the lives of the Fairbanks community members; they are learning leadership by doing leadership. --- Cultural Leadership: Alaska Native Values Leadership theorists are greatly aware that each culture has its own leadership characteristics and expectations. 41 Native Alaskan cultures are extremely diverse, with communities having a list of values, a set of ideals that some Native Alaskan peoples strive toward. Such traditional values have endured through time. While each Native Alaskan group has a unique list of values, there are common themes throughout all these communities that can be recognized in oral traditions. The amalgamation of Native Alaskan cultural values are as follows: It is with these values in mind that cultural differences become apparent in leadership. With culture and values being the benchmark for leadership within Miss WEIO, the process to enter the pageant may start several months, even years, before the initial application to the event. The participant's preparation includes developing mental, physical, and practical skills such as language fluency, history lessons, cultural awareness, and a demonstrable culturally relevant skill. Applicants also need to have a general understanding of Indigenous issues in the twenty-first century both for Alaska and for the United States more broadly. Family and friends all help, and in some cases the whole community comes together as the women get ready for the event. Items of regalia are gathered, and information is sought to accurately portray a community's culture to the wider audience and to the outside world. 43 Family and, more importantly, community support therefore becomes an invaluable part of the pageant process, again illustrating the practical application of Native Alaskan cultural values through the contestants' preparations. WEIO observes the cultural leadership model by placing community in the center of the women's youth leadership experience and through adherence to these values in the community events that take place throughout the week. For the young women who are leaders in their Indigenous communities, behaving in accordance with cultural values is an essential element of leadership. Former Miss WEIOs have shown that family and traditional values are incorporated into their lives and their leadership, as can be evidenced by Miss WEIO 2012, Kelsey Ciugun Wallace: [I have been able to continually build upon] representing my people in the right way. . . . Before when I was running for Miss Kuskokwim . . . I knew where I was from . . . what we did . . . but along the way I've been taught . . . [so much. For example,] traditionally back when my elders were living the traditional lifestyle . . . how they were able to incorporate those values into everything that they were doing. That's something that's really changed for me. . . . [Now when it comes to] the really big decisions that I have to make, I remember [the] traditional ways that I've been taught. 44 As can be seen in Wallace's example, culture becomes integrated into the decision-making process and therefore one's leadership style. --- Putting the Needs of the Community First Wilma Mankiller stated that Indigenous women leaders put the community first. In a similar way, the servant-leadership approach, coined by leadership theorist Robert K. Greenleaf in the 1970s, is used to describe leaders who draw their inspiration from "the people". 45 If we choose to think of "the people" as one's community, we can see that these two ideas are intrinsically linked. While the servant-leader approach has undergone many developments, Greenleaf's definition is still the most commonly used in the field. 46 The desire to serve is the root of servantleadership, which mirrors Mankiller's community first approach. After the initial desire to serve comes the aspiration to lead, so that servant-leaders can help their people. The pageant therefore becomes the venue for participants to enact their community leadership. 47 Within the Miss WEIO contest, the women competing choose to enter because of an interest in helping their Native Alaskan communities; therefore, we can surmise that the contestants are being community leaders by adopting the servant-leader approach. As with many pageants, each contestant has a platform that she highlights throughout the week. From year to year the platforms have included suicide prevention, language revitalization, subsistence hunting, drug and alcohol prevention, and domestic violence eradication, to name a few. These platforms have been chosen because of the lived realities of Native Alaskan life. The women often campaigned about these platforms before they entered the contest, and they look at the pageant as an opportunity to gain experience, organizational backing, and networking to continue their work as servant-leaders. Alaska Native values have been correlated to Robert Greenleaf and Larry Spears's ten characteristics of servant-leadership and used as a tool for analysis of the community service events during the Miss WEIO pageant week. 48 The following table illustrates the previously mentioned Alaska Native values and their corresponding leadership characteristics. 49 Native In the examples that follow, community service events are cross-referenced with both the Native Alaskan values and the servant-leader characteristics to illustrate that the contestants are following a servant-leadership model of Alaskan Native youth leadership. 50 Holistically, the use of values in these events show that the contestants are putting the community first, as highlighted by Wilma Mankiller, and practicing a community well-being model of leadership by integrating the six interconnected aspects of well-being. --- Honoring Your Elders The Native Alaskan value of honoring your elders can be illustrated through listening at the FNA luncheon, during which the contestants took the time to sit with elders. The contestants over the past few years have had the opportunity to meet with elders as they served traditional food to approximately twenty Fairbanks residents. After lunch, the young women introduced themselves and conversed one-on-one with the elders. This was a time to ask questions about environmentalism, natural resource management, climate change, culture, history, ancestry, songs, dance, and language, as well as to sit and listen to what the elders wanted to share. The conversations varied depending on the level of commonalities and shared experiences of the contestants and elders. An elder who was born in one of the contestants' villages especially enjoyed catching up on events back home, which led to a history lesson about his village and his ancestors. Another elder was keen to hear what the contestants had been doing during the week. Yet another was sharing her experiences of her early years in Fairbanks. The elders who shared their knowledge, histories, and culture gave a cross-cultural and crossgenerational learning opportunity that could not be replicated through books. Here we can see an example of the integrated approach to community well-being that the elders embraced in their conversations with the youth. The discussions were not focused on one specific topic but integrated aspects of well-being-the economy, health, education, the environment-to give history lessons and talk about contemporary issues. Getting to know elders from various Alaskan communities and having the opportunity to sit and listen are central to both servant-leadership and Native Alaskan values and were promoted throughout the week on various occasions. The listening trait is explained as a "learned discipline that involves hearing and being receptive to what others have to say." 51 Perhaps the most important element of sitting with elders and listening during WEIO is to hear what concerns the elders have about the future. One elder shared her conservation concerns, that there were few people left who could sing and dance the traditional songs and that someone needed to record them before they were gone. For the contestants to have the chance to hear such concerns and act upon them thus becomes central both to their roles as servant-leaders and to the cultural leadership model of Alaskan Native youth leadership. For a young woman who comes from an Athabascan village, the opportunity to speak to a Yup'ik elder and learn the differences between various Alaskan environments and cultures is an invaluable experience. These opportunities may be a normal occurrence for some young people, but for others, this is a new and exciting opportunity. In many Indigenous cultures, the elders are the fluent language speakers who know dances and songs, who understand the land and seasonal cycles, who know the necessity of maintaining communities. Given the cultural knowledge that elders have gained throughout their lifetimes, they are ideal people to provide context in any given situation. Throughout WEIO, the contestants therefore gain advice and knowledge from the cultural experts in their communities. Native Alaskan communities acknowledge the importance of elders within their societies and actively seek elders' advice on important issues affecting the community. Listening to elders is a community value that was and still is central to community survival. 52 Many leaders, especially youth, have a plethora of ideas aimed at bettering their communities but may lack the practical experience and institutional knowledge to put these ideas into practice. To overcome this disjuncture, a meeting was established within Alaska-Denakkanaaga, the Elders and Youth Conference-to enable both elders and youth to share experiences and knowledge with one another. 53 The elders have a chance to mentor Native Alaskan youth and holistically examine issues of concern to Native Alaskan societies, considering problems from a wider perspective. While the conference involves more youths than the Miss WEIO contest, the principle of youth and elders meeting and sharing ideas remains the same. The transference of information from elders to youths is a vital link that was at one time the target of destruction by the federal government. 54 Starting with the federal boarding schools of the 1800s, educational policies that were in place until the 1970s forced children to attend schools away from home. 55 In an attempt to strip away every part of their culture, children were punished for speaking their language, causing intergenerational trauma and a break in cultural continuity. 56 When the children finally went home, they had missed vital years of education in their home community, they did not know their elders, they had lost fluency in their heritage language, and they had missed important religious ceremonies; while they had been away, the rhythms of life in the community had changed. 57 The federal government has systematically tried to break cultural continuity in the hope that the younger generations would suffer an irreconcilable loss from which they could not recover, thus ending Indigenous culture in the United States. 58 In spite of the government's attempts, Indigenous nations found new and intuitive ways to continue teaching their culture to future generations. Organizations such as WEIO, with the yearly Olympic Games, are one such way that the transmission of culture continues in today's societies. The contestant's actions during the community service events demonstrate Greenleaf's theory of listening being the first and foremost characteristic of servant-leadership. The FNA luncheon is just one example illustrating WEIO's youth leadership; however, all the examples that follow start with the act of listening before the contestants reflect, offer advice and cultural knowledge, and share their own experiences. Listening is the cultural anchor throughout the week, shaping the perceptions of the youth who decide to run for this pageant each year. --- Taking Care of Others The Native Alaskan value of taking care of others can be likened to empathy-seeing a situation from another's perspective-and healing-helping that person to overcome difficulties. Each year contestants have two or three opportunities to offer emotional support to troubled youth of all races. During a visit to the FNA Graf Rheeneenhaanjii and the Presbyterian Hospitality House, the contestants, chaperones, and organizers shared their culture and knowledge with local youths, men, and women between eleven and eighteen years old who were encountering physical and emotional problems. The youths listened to each of the contestants introduce themselves and then had an opportunity to ask questions and advice from the young women. This was a very emotional time, as everyone talked about issues close to their hearts such as alcoholism, drug abuse, suicide, and cultural loss. The level of mentorship varied from one-on-one conversations about overcoming personal problems to group question-and-answer sessions or talking circles with shared inspirational stories. The depth of expertise differed with each contestant based on her prior experiences and the level of comfort in sharing her personal experiences with others. Servant-leadership explains that the healing process works both ways: "In helping the followers become whole, servant-leaders are themselves healed." 59 While the scope of this research cannot confirm or deny this aspect of servant-leadership, the emotional atmosphere in the activities was one of release and calm once the conversations had taken place. Throughout the week, there was resounding agreement from the contestants and the organizers that positive role models had made a difference in their lives, helped them overcome problems, and helped in their personal growth, inspiring their interaction with the youth at the FNA Graf Rheeneenhaanjii and the Presbyterian Hospitality House. Thus, the healing process came full circle as the contestants gave back to the local community. In today's global society, young people are facing unprecedented issues as the discrepancy between what is realistically achievable and what the media dictates as "ideal" continually widens. For example, in terms of body image, record cases of anorexia nervosa and bulimia in both women and men are being reported in the United States. 60 At the same time, obesity and diabetes are on the increase, causing health-related problems and fueling speculation that the parents of today's youth will outlive their children. 61 Among Indigenous youth in America, suicide is the second leading cause of death. 62 Illegal drug use is on the increase. 63 Today's youth who opt to matriculate at university face a lifetime of debt and job uncertainty. Thinking again about the holistic approach of the community well-being model of leadership, the youth participants draw from various themes to talk to the people in a bid to encourage a positive lifestyle. The younger generation would benefit from strong role models, preferably peers who have been through such pressures and can offer advice, reassurance, and coping strategies. Although this role of caregiver could be interpreted as a gendered behavior and could be interpreted as perpetuating women in maternal or nurturing roles, this activity is one part of the community service activities that take place all week. In this capacity, the women are working as educators, ambassadors, and role models as opposed to political leaders. Miss WEIO gives women the opportunity to learn leadership by doing leadership; therefore, if we use the values as a basis for culturally based leadership, then we cannot differentiate them based on Western notions of materialism and nurturing. Breaking down the binary of male/female, Indigenous nations have been more fluid about who is categorized as a woman and a man, with third-gender peoples being deemed powerful, as they can negotiate across gendered boundaries. As such, it would be problematic if men never took on the roles of carriers of culture, taking care of others, or community healing. There is evidence, however, of men taking on such roles in today's societies, so we can assume that men are also following the values of their ancestors. 64 These characteristics should be perceived as useful to community leaders, regardless of gender. Throughout the week, it became evident that the contestants were demonstrating empathy and healing, but not just through the advice given during the two events listed above. Each of the community service events required an element of empathy with an audience who may not have known about Native Alaskan culture and who may have asked inappropriate cultural questions; this is especially true of children. While the women patiently answered questions and engaged in conversation, one can argue that healing and forgiveness were taking place in a truly crosscultural sense. The act of gaining understanding and apologizing can be seen on a national level in Australia with their National Day of Healing, aimed to "focus on the healing needed throughout Australian society" as a result of the Stolen Generations. 65 Just as Australia needs to heal as a country for its citizens to be able to move on, so Alaska and the rest of the United States need to heal from the genocidal policies that took place against Indigenous peoples from first contact and that still exist today. While the empathy and healing during WEIO take place on a much more intimate basis, the shared cross-cultural knowledge that occurs between the contestants, the Fairbanks community members, and the tourists to Alaska creates a space for cultural understanding, which in time can lessen the power of stereotypes, lead to empathy, and start the long process of healing across the country. --- Show Respect to Others and Share What You Have The Native Alaskan values of show respect to others and share what you have are demonstrated when the contestants meet the public at the Riverboat Discovery, the Alaska Salmon Bake restaurant, and the Noel Wein Public Library. Within the servant-leader approach, awareness includes "understanding oneself and the impact one has on others" through being "acutely attuned and receptive to their [the leaders'] physical, social, and political environments." 66 As mentioned earlier, during the Miss WEIO contest the women are aware of their positions as ambassadors for their peoples and for the WEIO games. Throughout the week, the women meet with a cross section of the Fairbanks community, as well as tourists who have traveled to Alaska from the Lower 48 and beyond. Every time the contestants meet with the public, they make sure that they are representing their people in the right way. 67 As ambassadors, the servant-leader trait of persuasion, which is "a clear and persistent communication that convinces others to change," becomes apparent. 68 While the women are not political leaders, they do become cultural spokespersons, enlightening and educating the people they meet on Native Alaskan physical, social, and political matters. 69 The Native Alaskan values of showing respect to others and sharing what you have are demonstrated when the contestants meet the public. Throughout the encounters with the public at the Riverboat Discovery,the Alaska Salmon Bake restaurant, and the Noel Wein Public Library, the women patiently answer questions, have photographs taken, and use the opportunity to express their platforms to an eager public made up of tourists and local Fairbanks residents of all ages. On the first day, after orientation, the contestants go on a short cruise on the Riverboat Discovery, which is followed by a meal at the Alaska Salmon Bake restaurant. These two events are predominantly tourist attractions, so there are a disproportionate number of photographs taken both with and of the women in their traditional regalia. Unfortunately, this activity illustrates the audience's lack of political understanding of Indigenous nations and highlights the "exotic other" concept that the women experience from the tourists. Before, during, and after the photographs, the holiday-makers ask questions relating to the women's culture, for example, "What are your parkas made from?" Some tourists, however, take a more political stance with their questions, asking, "How many of you speak your heritage language?" They want to know about language policies. Although short in nature because of the structure of the event, these questions spark discussions of a physical, social, and political nature that can be heard continuing after the women have left the conversation. At the Noel Wein Public Library's summer reading program, the contestants have the chance to speak with approximately 150 local children, from under one year to fifteen years old, and their parents. One of the librarians will ask the women to introduce themselves, demonstrate their talents, and ask them questions such as "Can you tell us what your gloves are made of?" and "Can you teach us three words in your language?" The contestants will take turns standing up and answering the questions before demonstrating how to use an Eskimo yo-yo or some traditional WEIO games. The young audience can also ask questions. Contestants were put on the spot when a young child asked, "Do you live in an igloo?" "What do you wear under your dress?" and "What do you keep in your handbag?" Again, these questions illustrate the lack of mainstream education about Indigenous peoples, as the questions are always about surface culture. The local newspaper recorded and took photographs of the event, and afterward the eager young audience had their own chance to have their photographs taken with the contestants. Again, this is not without its complications. Young children today have fixations with Disney princesses that create unrealistic images of Native American women and women in general. When elementary and high schools fail to teach Indigenous studies, children turn to television and film to fill in the gaps. While today's multicultural Disney princesses such as Moana are more favorable than the over-stereotyped Pocahontas, who could talk to the wind and animals and had an unachievable body shape, the children who talk to the Miss WEIO contestants can at least see average-sized women of all skin tones who are dressed in jeans or leggings and a top: a realistic and relatable image. Furthermore, this image is not voiceless but has an active voice that answers questions and demystifies Indigeneity. These events become the perfect opportunity for the contestants to show respect and share knowledge and information with audiences who may not have previously been introduced to Native Alaskan culture while at the same time using the servant-leader trait of persuasion to give an accurate account of Native Alaskan culture. 70 For many people, the Miss WEIO contestants are the first Indigenous people with whom they have met and spoken. The contestants understand that their words might be construed as applicable across all Alaska Native nations or even Native American nations; the danger of reinforcing stereotypes is very real. There is a constant pressure for participants to portray their people in a culturally appropriate manner. 71 Americans are fascinated by Indians and Eskimos. The exotic other has long been used by Hollywood and advertising campaigners to sell products. 72 Blockbuster films and popular television series use outdated ideas and imagery to represent Indigeneity. And while a plethora of Indigenous-made films exists, they are rarely, if ever, shown on main television stations or in cinemas. Scholars such as Stephanie Fryberg and colleagues have researched the adverse psychological effects the "stereotypical" Indian has on the self-confidence of Native American peoples, yet that is the image that is still on American butter cartons. 73 It is crucial that Indigenous peoples receive the opportunity to self-represent to future generations to break this false stereotyping cycle. Miseducation about Indigenous peoples starts in the schools and therefore continues in the public. 74 As future generations learn Native Alaskan history and culture directly from the source, they will refer back to this knowledge when they continue into their future professions. Violence against Indigenous peoples and multigeneration historical trauma will not vanish overnight, but the healing process will begin when the injustice toward Indigenous peoples is stopped, and this is only possible through education. While Indigenous pageants are not the prime solution to the miseducation problem, the youth who take part learn through the experience that acknowledging and fighting stereotypes is an integral part of being a Native Alaskan leader in the twenty-first century. 75 Pray for Guidance, Accept What Life Brings, Have Patience, and Live Carefully The Native Alaskan values of praying for guidance, accepting what life brings, having patience, and living carefully are comparable to the two leadership characteristics conceptualization and foresight. Conceptualization can best be described as looking at the whole picture and within leadership can refer to those people who are visionaries. Foresight is "an ability to predict what is coming based on what is occurring in the present and what has happened in the past." 76 These values can be seen as being integral to the community well-being model, as they illustrate the interconnectedness of all the conversations and experiences the contestants have throughout the week. Taken together, these Native Alaskan values and two servant-leader traits can describe the process used by contestants to determine their platforms. As mentioned earlier, each of the Miss WEIO contestants promotes a platform, a social justice issue that she has usually witnessed firsthand and wants to promote during the week. This manifesto becomes a starting point when talking with people in the community, a way of introducing oneself. The philosophies are as diverse as learning to subsistence hunt and fish, overcoming addictions, graduating high school, or simply sitting and listening to your elders. Miss WEIO 2012, Kelsey Wallace, took the opportunity to talk about suicide prevention for her platform. Bethel, Alaska, is said to have some of the country's highest suicide rates. 77 Wallace has seen the devastation caused by suicide in her region and is a promoter of Drew's Foundation, a local organization that is "dedicated to saving lives." 78 By looking holistically at Alaska Native communities, Wallace can see a pattern emerging with youth taking their own lives, and she wants to do something to break this pattern now. Platforms such as Wallace's arise from a desire to see Native Alaskan communities thrive; in other words, the youth leaders have the foresight to see what is happening in their community and have conceptualized a solution which draws from all aspects of life. The messages and examples held in the platforms develop throughout the week as the women have the chance to enter cross-cultural and cross-generational conversations. Having an opportunity to speak in large groups and on a one-on-one basis helps the contestants solidify their messages of hope, survival, and persistence. This is especially true when the contestants speak with troubled youths, as mentioned earlier. After gaining experience in answering questions and learning what answers are the most helpful in any given situation, the contestants feel more at ease and can converse more precisely about their platform and their hopes for their communities. Conceptualization and foresight are shown throughout the community service events by the contestants being positive role models for youths. Throughout the events, the contestants act as role models by sharing stories of hope with youths. 79 This is the perfect opportunity for the women to reach out to their peers and explain how they manage to stay happy and healthy or overcome temptations and deal with personal issues, thus using foresight to stop the trajectory of youths in crisis. The opportunity to interact with youths who are in a desperate situation, think quickly, give advice, and provide a success story from which young people can gain practical advice are indicative of both servant-leadership and Native Alaskan values. --- See Connections and Know Who You Are Throughout the community service events, the contestants use the Native Alaskan values of seeing connections and knowing who you are, to become a steward for WEIO. Being a steward "is about taking responsibility" for the organization or community you represent. Commitment to the growth of people is about helping individuals in that organization or community achieve their best. Combining these two traits together, building community provides "a place where people can feel safe and connected with others" within that community. 80 These traits and values are perhaps more evident when the contestants speak at the Golden Days senior citizens luncheon. To build a community that allows individuals to thrive, you need to have a distinctiveness that sets you apart while at the same time a commonality for people to share: culture. To the outside world, language is an instantly recognizable marker of belonging to a cultural group. Formal introductions are a means of identifying ancestry, thus situating oneself in a larger context and placing oneself within a community. The ability to formally introduce oneself, preferably in one's heritage language, is a cultural leadership skill that many young people need to learn. Although pageants such as Miss WEIO do not integrate a language requirement into the contest, participants often formally introduce themselves in their heritage language, many having learned their formal introductions for the contest. The Miss WEIO contestants have several chances to give formal introductions via public events. Introductions differ, depending on the event and time restrictions, from one's own name and the names of parents and grandparents to a full formal introduction, including the areas from which family members originated. The fluency levels also vary with each contestant. In 2012 more than half the WEIO contestants were taking classes in their heritage language at a university or college. Contestants who have had no opportunity to learn their ancestors' language introduce themselves in English, whereas others whose language is widely spoken in the community give a full, formal introduction in their Indigenous language. Regardless of fluency, all contestants are aware of the significance of formal introductions and the severity of language loss in Native Alaskan societies. The contestants had a chance to formally introduce themselves at the Golden Days senior citizens luncheon. During this event, they stood onstage in front of 340 guests from the Fairbanks area; their only instructions were to introduce themselves and their platforms. Taking approximately three minutes each, the contestants told the audience who they were, where they came from, and what important issue they were campaigning for during the pageant. This was the only time throughout the event in which the contestants were on a raised stage with a microphone in front of a captive audience. The contestants had had ample opportunities to speak in public, consequently developing their public speaking skills, and thus they felt confident speaking in front of large audiences. As the week progressed, the frequent use of introductions enabled a flawless delivery in front of the public. Many of the women did not speak their heritage language as their first language, and their introductions did not come easily in the orientation. The pageant thus offered them a culturally appropriate public venue in which to formally introduce themselves and thus demonstrate a sense of Native Alaskan community. The federal government's assimilationist policies to eradicate Indigenous languages have left many communities with no heritage language speakers. Missionaries, boarding schools, and environmental factors have all led to a decrease in spoken Native Alaskan languages and their transmission to future generations. Public Law 101-477, the Native American Languages Act, passed on October 30, 1990, was introduced in an attempt to halt the deterioration of Indigenous languages. Since then, communities such as the Yup'ik have created immersion schools to reintroduce their heritage languages back into use in the home. 81 Today, out of the twenty-two languages once spoken by people of all ages in Alaska, only two are spoken by children in the home environment. 82 While the women gain respect from elders and community members for speaking their heritage language, the ability to speak a Native Alaskan language also offers a cultural cue to the audience, who automatically recognize that they have met someone from a different race upon hearing an unfamiliar language. Harmful stereotypical imagery is constantly applied to Indigenous people, while the accompanying voice is rarely heard. Often Indigenous people are not recognized as such unless they are wearing traditional regalia or "look" Indigenous. During Miss WEIO the women wear jeans and kuspuks; 83 thus, when the audience hears the accompanying voice speaking a Native Alaskan language, they store a new contemporary image to replace the previous stereotypical one. Creating a sense of community is difficult when there are multiple communities within Alaska. The use of language and introductions by the contestants is one way to show stewardship and strengthen communities from the village through the international level. 84 --- Concluding Thoughts The audience chats away as the list of places visited during the week comes to an end. You can see that the families of the contestants are excited and nervous as they stand ready to take photographs. The winners of each category are announced, and the women come forward to collect their prizes. Cheers from around the arena can be heard from the communities that send their youth year after year to participate in this great honor. The winner, the new Miss WEIO, humbled by being chosen, gives a speech thanking her family, her community, and the people who have supported her journey. The competition is nearly over, but the hard work is far from finished. Although originating from a Western construct of gender, Miss WEIO has been decolonized to create a space for women to work in their communities based on Alaska Native cultural values. WEIO itself places cultural revitalization and maintenance at the center of everything it does. While some may argue that the pageant, because of its very nature as a pageant, reinforces the regulation of women's bodies, one only has to go to WEIO to see that the women look the same and are treated the same as every other Indigenous person who is there. WEIO and Miss WEIO are therefore intrinsically linked, and it is this connection that allows the pageant to emphasize youth leadership. Mankiller provides a theoretical leadership model that places community in the center of leadership. Throughout the weeklong activities, the women center their thoughts on the communities that they meet. In the Indigenous context, leadership skills are not isolated from social content, where people lead for the community good. WEIO follows this model by placing community in the center of each contestant's leadership. The WEIO winners have proven that traditional values are incorporated into leadership; community support becomes invaluable in the pageant process. Within this community leadership, if we examine the events through the community well-being model, we can also see evidence that the contestants are gaining these leadership skills. We see the interconnected approach that the contestants are using in order to integrate their platforms with the well-being of the community. The youth leadership experience does not stop when the winner is crowned with a baleen and ivory crown. The experience is not over when the people leave the arena and head home to their communities. The week has given the newly crowned winner and all the contestants the experience and the confidence to talk to large groups about Indigenous issues, to network with the people in the Fairbanks community, and to speak to the media. The winner must now plan her year so that she can continue her social justice issue, talk with youth and elders, and share her culture with others. The crown and sash now act as a symbol that shows communities near and far that this person has been chosen as an ambassador to represent WEIO and its organizational values. Offers arrive throughout the year for the ambassador to talk to schoolchildren, address the Alaska Federation of Natives , emcee at the Fairbanks Festival of Native Arts, and represent Alaska and WEIO at the Gathering of Nations Pow Wow. The skills that have been continually developed throughout the contest week continue to be used throughout the ambassador's yearlong tenure. Since the first pageant in 1961, the women who have competed have continued to be leaders in their communities. Some went on to become successful businesswomen; some went into governance, becoming regional and village corporation leaders; 85 others went into resource management and academia; and some became keepers of culture. When I talked to previous winners, they attributed their success to some of the key skills gained during the contest week. There is no doubt that the contest gave women confidence to publicly talk about key issues in Indigenous societies. The skills gained placed them in stronger positions to fight for their communities, suggesting that the pageant is doing more than preparing women for gendered leadership styles and responsibilities. It is also preparing women to become leaders equal to men. The youth leadership opportunities provided by the pageant, however, are steeped in cultural relevance. The development of servant-leaders in the context of community service events during the contest week illustrates how contestants can self-develop their own culturally unique leadership style. The competition week for Miss WEIO is intense and exhausting, but those who compete take away an experience that will be etched in their memory for a lifetime. More importantly, they are given an opportunity to learn leadership by doing leadership. Caroline Williams graduated with her PhD in American Indian studies from the University of Arizona in 2013. Williams is currently working at the University of East Anglia, England. As an ally of Indigenous peoples, she examines Native American women's strategies to maintain healthy communities. Focusing on social justice issues through the exploration of role modeling, ambassadorship, and normative beauty ideals, Williams researches this topic through participant observation with contemporary youth leaders and examines their approaches used to promote well-being among Indigenous youth through arenas such as language, culture, and identity. --- Notes This article comes from a larger project that examines the Miss World Eskimo-Indian Olympics from three perspectives: the women who take part, the members of the audience who watch the pageant, 7. The theory of why they wanted a beauty contest is still being researched; however, it perhaps was the idea of Muriel Hagberg based on her experience in the Miss Alaska contest. Or perhaps Whaley, looking back at Muriel's success in the Miss Alaska contest, thought that there should be an Indigenous alternative after pageants in Alaska had become segregated. 8. The Olympics are separated by gender, however, with men's and women's heats. Some games that were traditionally men's games are only open to women if they ask to take part; see https://www.weio.org/gender-the-games. 9. Jennifer Nez Denetdale writes about the disenfranchisement of women in the Diné Nation, as the pageant places women in a maternal role, responsible for transmitting culture, and reinforcing Western gendered ideas of women, which are to place women in a separate sphere from the political realm. These separate spheres are a problematic distinction unique to the Diné Nation, as traditional narratives inform political roles. Thus, the role of Miss Navajo Nation is to embody the perfect Diné women while sitting next to but not participating in the political life of the government.

This research explores the adaptation of traditionally objectified women's spaces into an arena for community leadership. Indigenous pageants offer a place for women to become spokespersons on social justice issues without the sexual objectification associated with beauty pageants. Within Native nations, we see examples of youth creating a better life for their community through Indigenous pageants. There is a growing literature among Indigenous studies scholars on community-based women's leadership. Wilma Mankiller offers a theory on Indigenous women's leadership that centers the community at the heart of decision making. The winners of the Miss World Eskimo-Indian Olympics (WEIO) all share this common feature; they want to better their communities, and they view the pageant as an opportunity to do so. The women who undertake the weeklong experience participate in community service events in the Fairbanks region. These leaders are immersed in experiential learning, an invaluable opportunity that cannot be replicated through theoretical knowledge. The women are presented with situations in which they share cultural experiences and give advice cross-culturally and cross-generationally. Gaining the opportunity to be a voice for their peoples, the women must quickly consolidate their knowledge and sharpen their communication skills as they are repeatedly questioned about Native Alaskan societies. Using Wilma Mankiller's ideas on community-centered leadership, this article explores the case study of community service events featured in the Miss WEIO contest to illustrate that Native Alaskan pageants can provide youth with an invaluable opportunity for experiential leadership that is intrinsically linked to promoting community well-being.

Introduction The increasing ubiquity of information and communication technology in the social welfare sector has led to intensive debate about the development and use of AI-based systems. This debate has become increasingly focused on applications that regulate human interactions and activities. These include applications that affect decisions about people, for example, by predicting the probability of recidivism of delinquency, such as the COMPAS risk assessment software , or by scoring and categorizing job seekers into categories in order "to increase the efficiency of its counseling process and the effectiveness of active labor market programs" such as the AMS profiling system of the Public Employment Service Austria . The Automating Society Report 2020 lists many other algorithmic decision-making systems in Europe that are intended to support and improve the lives of service recipients. Algorithmic decision-making systems "affect almost all kinds of human activities, and, most notably, the distribution of services to millions of European citizens-and their access to their rights" and consequently could pose a threat to existing forms of democracy. Much of the debate has focused on automated decision-making using predictive analytics and both supervised and unsupervised machine learning. Algorithmic systems have been criticized because of serious errors in their design and implementation. Specific criticisms have been the lack of transparency of algorithmic systems ); biased or missing data; the unquestioned assumptions of developer and unintended consequences . The use of such systems brings with it the risk of errors of commission and omission using algorithmic decision support systems . Users may be influenced in their decision-making by the algorithmic system, but they might also manipulate systems through inputs designed to achieve certain outcomes . Algorithmic systems are, of course, also reliant on the data used to develop them. This data may be biased in ways that lead to discrimination, and its use raises questions about privacy and data protection. A fundamental question is the identification of exactly what data an algorithm is designed to use, and what data it can actually access. A preceding question also arises about the nature of such data and the process by which it was generated. The research described here aimed to address this question by exploring the processes by which data is recorded by social workers and decisions are made about what is recorded as data, what is not recorded and the legal, ethical, and professional considerations that arise. 1 The findings suggest that such scrutiny is important to the assessment of the utility of data for the development and use of algorithmic systems. This is of particular pertinence to other professions, such as IT professionals, who have to evaluate the quality of data contained in client information systems used in social work. --- The Debate About Client Information Systems in Social Work It is the intensive and high-level discussion about artificial intelligence that in some cases neglects the fact that the degree of digitalization is low in many jurisdictions. Professions such as medicine , nursing, or social work also have to deal with changing work domains because of digitization processes-including both information and communication technologies and AI . However, discussions in these fields are less concerned with automated decision-making than with supporting decision-making processes and how far processes can and should be digitized . Case recording in terms of what gets recorded has also been the subject of debate in social work. This debate became more focused with the advent of CIS as decisions about what gets recorded, when, how, and by whom were being made by CIS designers and managers with no frontline experience. Debate has also now turned to what we do with all this data, and data science has introduced the possibility of predictive analytics in all its forms. Much of the research and commentary about CIS in social work rests on the assumption that "accurate recording and inter-professional information-sharing can enhance care in health and social services" . A further assumption is that case records should contain all information which is essential to the process of making decisions and taking actions about a case. As such, case recording has been considered as part of the professionalization of social work . Case records can also be used to demonstrate accountability and to evaluate and plan services . They may also be shaped by the criteria set for eligibility for services. Hence, the creation of case records is influenced by the need to meet and anticipate multiple possible uses, some of which may exist in the present and some in the future . Social welfare agencies may have extensive procedures about case recording, and there may also be established routines about what should be recorded but, ultimately, it is left to the discretion of individual professionals to decide what to include and what to omit in case records . This is an example of what Lipsky refers to as "street level bureaucracy," in which social workers act as the interface between state policy and individuals requiring a service. Social workers are expected to match up sometimes general and quite vague social policy with the complex needs of individuals. As Lipsky describes, this may be achieved by social workers "by developing routines of practice and psychologically simplifying their clientele and environment in ways that strongly influence the outcomes of their efforts" . Since the 1970s, there is also critical debate about the constructive character of case recordings. Previous studies highlight the construction of a case during recording and emphasize that this case construction is used to legitimize professional actions . According to Merchel and Tenhaken , records provides "information about the views and interpretations of the person recording, about their selective perceptions, their categories, and their theoretical outlook." Even today, the question of how to evaluate the relationship between records and reality is still being discussed . Gillingham ), for example, speaks of "large gaps" to emphasize the difference between the two. Due to their stigmatizing effects Webb, 2003), records "should treat a client fairly by demonstrating respect for them" . If the intention is to use case records as "data" for predictive analyses, it is important to be aware that CIS might have a significant influence on how case recording is conducted . Scholars report deliberate manipulations and dummy entries , deliberation and interpretative "making […] fit" of given categories , redaction, ignoring, or deliberate expansion of certain defaults , among other things. --- Expectations With Regard to What Is Not to be Found in the CIS The question of whether and, if so, under which circumstances there are omissions in digital records has been addressed in some research ); albeit often marginally. Non-available data are both unknown data and data that are not entered into CIS. Some omissions are the result of limitations in both the hardware and software used in CIS . Furthermore, restrictions due to organizational framework conditions and personal factors ) can influence the quality and scope of recording. However, few studies are also devoted to the phenomenon of deliberate omissions in digital case records . Another reason for omissions in case records is that service users expect that the information they impart to social workers is bound by confidentiality . Indeed, research has found that "[e]thical tensions can arise for health social workers [that means, social worker who are part of an multidisciplinary health care team, DS] between comprehensive recording and protecting client privacy by limiting access to privileged information obtained in the context of a social work encounter" . Stablein et al. report "concealing information and utilizing sets of personal and collective codes designed to alert providers or teams of providers to confidential information within a patient's record." Confidentiality in professional social work practice is one of the core values and a "cornerstone of professional social work relationships" , 2005, p. 7). According to the German Criminal Code , the duty of confidentiality applies, among others, to staterecognized social workers or state-recognized social pedagogues . For professional work, CASW defines that "[s]ocial workers demonstrate respect for the trust and confidence placed in them by clients, communities and other professionals by protecting the privacy of client information and respecting the client´s right to control when or whether this information will be shared with third parties." In this sense, the data collected in the context of case recording usually belong to special categories of personal data that are considered to have a high level of protection . Accordingly, the processing of this data2 in order to decide, for example, on social services, must comply with applicable data protection and data security laws and regulations. At the same time, however, data protection itself plays an ambivalent and ambiguous role in social work practice in Germany . According to Pudelko and Richter , although there is an awareness in German social work of the need to implement legal frameworks to protect personal data, the challenge lies in how exactly to implement these requirements in daily practice. One of the biggest challenges is the European General Data Protection Regulation itself. The application of the GDPR addresses people who are assumed to be capable of informed consent, but this assumption cannot be taken for granted everywhere in the field of social work. Therefore, social workers attempt to compensate for this discrepancy through their so-called double mandate: they not only see themselves as being obligated to the state as a funding agency and principal, but also orient themselves towards the needs and requirements of their clients . In doing so, ensuring the privacy of vulnerable people could play a decisive role in the whole process of social care service planning, including their digital case recording. --- Study Design and Results The current case recording practice within social organizations was analyzed as part of the interdisciplinary project MAEWIN 3 . The aim of the MAEWIN project is to explore the possibility of using automated procedures for text and data analysis to generate evidence-based recommendations for actions in the field of social care services. In addition to assessing what information is available for this process in the database, this research has also aimed to develop an understanding of how it was created. The field of study is the provision of services to people with disabilities, with the aim of maximizing both their self-determination and their participation in society. --- Methodology In contrast to the international discourse in social work, there are no DSS used in Germany that could be investigated for this purpose. The aim of the first stage of 1:26 Digital Society MAEWIN study therefore was to gain an overview of existing and non-existing data points within the CIS and identify those "large gaps" ) that are highlighted in the literature to underline the difference between the recorded content and the actual information available. The prevailing opinion among scholars is that these gaps ensure that the data "is insufficient information to gain any real understanding of either the service user or the response provided" ). Having identified existing gaps for the specific context of social care service planning for people with disabilities, the next step is working out how data might be used to develop DSS and to consider more concretely the potential use of algorithmic systems for decision-making. First considerations and tentative conclusions on this have already been published in Schneider et al. . --- Data Collection, Participants, and Data Analysis Between between January and June 2020, twenty guideline-based interviews with experts working for funding agencies of integration assistance or social service providers were conducted. At the time of the interview, respondents worked in either North Rhine-Westphalia or Berlin and were between 29 and 63 years old. Interviews were recorded and lasted on average 122 min. The recordings were transcribed and then by using qualitative content analysis, according to Kuckartz via MAXQDA . In the following analysis, the following questions from the interview guide are focused on: • Which data are/ should be recorded? • Which data are/should not be recorded? • Why are data recorded? The following presentation of these aspects by the interviewees is summarized in a condensed form. Only in isolated cases, quotes from the interviews are mentioned to better illustrate some aspects. In many passages of the interview, the respondents refer to the specific context of planning social care services for people with disabilities in order to highlight its specificities. This contextual embedding is maintained in the presentation of the results. --- Findings I: Available Data Because social care service planning is subject to the legal regulations of integration assistance in Germany, the necessary criteria for an individual and function-related need assessment can be inferred from the legal text . This includes recording on whether a disability exists or is likely to occur, what effects the disability has on the social participation of the service user, and which goals and aims of participation are to be achieved with social care services. It is also necessary to note which services are to be used to achieve the 1:26 Digital Society goals of the person with disabilities. In addition to recording relating closely to the application for social care services, further case records can be found within funding agencies and social service providers. Most respondents report, keeping a more or less extensive electronic file on their clients. People who are employed by social service provider also record daily events of the clients within the framework of their professional work . The majority of the available data is recorded in semi-structured or unstructured data . Assessing the relevance of information is a difficult undertaking. Respondents had different strategies for determining relevance; for example, some mentioned that information is relevant, if it has an implicit and/or explicit impact on the social workers' mandate to act . Such information can be, for example, new wishes and goals of the service user, or a change in personal circumstances. However, the relevance of these changes is a matter to be decided by the case worker: "When a client falls in love, it's a very, very personal story. When he confides in me: I have fallen in love. Then this is to be recorded if a need for support arises from it. But if that is not relevant-although, with this topic, it has relevance [to work, DS] in many places." In some cases, intimate information that the service user does not want recorded or passed on to third parties may be disclosed. In this instance, professionals must make decisions about whether such information is recorded and whether it is passed on to third parties, despite the wishes of the service user: "And if clients say you don't write that down. Then I have to weigh it up -THEN I have to decide: Is it important that I write it down anyway? Then of course I have to make it transparent and say: Listen, [...] we are obliged [to record, DS]." "That sounds quite nasty now, because it is actually an abuse of trust on the part of the caregiver in relation to the person with a disability. When they [the caregivers, DS] tell us things that the service user does not want us to know. However, these statements are so important, because otherwise we do not understand certain things. Because these are the unspoken gaps." The need for privacy can, therefore, be superseded by the perceived need to record and share relevant information. Relevance can also be ascribed to professional observations and understandings, but they may not fit into the prescribed spaces in a CIS. Case workers find ways to circumvent the restrictions of the CIS: "Yes, we have this standard protocol, which everyone [...] fills out. They all look the same. If there was a conversation about it [about participation restrictions, DS], there is always another blank sheet. Because this protocol does not allow you to record a conversation or record things [for yourself, DS]. So you just have to be inventive." The assessment of relevance can therefore be made for different reasons and, in case of doubt, is at the discretion of the individual professional. Participants also raised the problem that some information recorded in case files can be derogatory about service users or could be interpreted as derogatory. For example, some respondents reported that records included derogatory remarks about the clients , or that the judgments of some of their colleagues were "mean, wrong, unprofessional, foulmouthed." Some respondents concluded that records should be free of personal assessments: "Yes, I think that emotionality or stories like that should be left out of the recording. Because I think everyone feels that differently and sometimes the situation is distorted as a result. Instead, it should be more about the fact that a third party, under certain circumstances, who is doing a substitution or taking over things, is informed without being pushed in a certain direction. Like that. I think that's important. And that it is formulated in such a way that the client can look in at any time without being stigmatised in any way. [...] But that facts [rather than hypotheses, DS] are recorded." "Personal evaluations, especially insults, have no place in the file." A few respondents point out that respectful recording is also challenging for professionals, particularly when the meaning of certain words is ambiguous. Information may thus be quite vague and open to misinterpretation. For some respondents, there is a realistic risk that other professionals or third parties could misinterpret such appreciative descriptions: "Professor Metzler has the word support, doesn't she? This word support. I can classify it both under B, as help, qualitative need for help, and under D. It is a neutral word, a friendly word. The funding agencies of integration assistance always classifies it as B, because there is less money. We classify it as D, for example. And then I say: guys, never use the word support singularly, because there is scope for interpretation. Either you describe the need for help in such a way that it can be assigned to B. If that's not the case, then you have to describe the need for help in such a way that there is no more scope for interpretation." --- Finding II: Non-available Data Respondents first addressed factors that are mentioned in the literature, such as the limitations of hardware and software , as well as the organizational conditions . Participants also mentioned a number of factors that may be present in caseworkers, such as a lack of understanding of the importance of case recording, uncertainty about how to use software, laziness, forgetfulness, and a lack of literacy. "The colleagues say they don't have time to record. [...] They would like to do it, but they don't have the time or the technical equipment. [...] So there is only one PC, three people have to record at the same time towards the end of the 1:26 Digital Society shift. There is a lack of resources. [...] For some [the recording] is an annoying evil. Who don't feel like reflecting on their work or writing it down. Who may find it difficult to write. That is a big problem. Many are also not necessarily familiar with the German language or are ashamed if they have spelling problems." Some participants pointed out that a significant deficit in case files is the service user perspective. This omission arises in part because of the limitations of the assessment tool but may also be because of the limited communication skills of the service user. For example, one respondent mentioned the "power imbalance" that exists in the process of case recording: case records are generally created and held exclusively by social welfare agencies rather than service users. An exception to this general rule occurs with the inclusion in case records of "deputy statements." These may be prepared for people with limited communication skills in order that their wishes and opinions can be included in case reviews. The lack of client perspective in the records is also expressed in the desire of fewer respondents for participatory recording: "I think it would be good to be able to keep the records more flexible in the sense that it is not tied to the office. So, I think it would be better -this is also being discussed at the moment in our organisation -to have portable notebooks, so that I can have them with me and, for example, also record with clients, together. That you sit down briefly at the end of a contact and look: Okay, what was important today? What do we want to record? So you can involve them [the clients, DS]." Overall, participants, for all the reasons mentioned above, considered that case files are incomplete, a partial version of "reality" and require interpretation. Interpreting the meaning of the case recorder, though, may not be easy or precise and so there is a level of uncertainty in using case files in day to day practice. As one participant described, this uncertainty has to be "endured" by social workers and is considered to be part of the professional role of social workers depiction of professional practice in the "swampy lowlands"). This perception is expressed particularly succinctly in the following quote from a respondent: "Paper is patient. You can imagine a lot, but you can also imagine a lot of wrong things." Finally yet importantly, two participants were of the opinion that there should be nothing that should not be recorded. One participant spoke about the need for a "positive fail culture" within social welfare agencies. "So, I think it would be wrong not to record that it also fails once. I also think it's right that you record that you might also be wrong with your assessment." Almost half of the participants reported that they do not record certain topics at all or at least not in detail. In addition to traumas or experiences of violence, respondents mentioned past life events that have no impact on the current situation , relationship problems including sexual relationships 1:26 Digital Society venereal diseases, or frequency of sexual contacts), a client's worldview and psychological problems that require therapy. Participants justified these omissions with reference to the very personal and potentially sensitive nature of some information. One respondent summed it up as follows: "There are definitely stories that do not have to be recorded anywhere." Although respondents acknowledged the needs for high levels of confidentiality, they were also aware that this seems to be not possible in the context of a social welfare agency. In five interviews, the interviewees point out that colleagues could see the data recorded in the CIS. "We can look into the [CIS, DS] at any time. [...] This is used by EVERYONE of us and there is no other way. Otherwise we wouldn't be able to work at all." "Everything we record is accessible to all. So "accessible to all" means: In our system, everyone who has an authorisation to look into these files can see that." Participants were also aware that information recorded in a case file exists permanently. This ubiquitous access led some respondents to reflect on the process of recording and its consequences. "The question is: For example, would I record what happened [in this case] in such a protocol? There is a POTENTIAL danger; it could be that something like this could happen again. Do I have to record that or is that ethically wrong? Because it's something that happened a long time ago [and doesn't currently have a impact on the work with the client, DS]. So, do I maybe not need to write that in my internal protocol? On the other hand, the question is: who reads the internal protocol? So, that's quite less a question of [internal, DS] recording. [...] Which might really only be read by myself as a reminder. Who already knows it anyway. Or my colleague who does the same work as I do." Participants' responses describe two methods in which they respond to ubiquitous access in the records. First, certain topics and information might not be recorded due to ubiquitous access to the CIS: "But I would not record intimate or very private details of a person there [in the CIS, DS] because, of course, that is data that, even if it is protected, may be accessed somewhere [by someone, DS], and that is also stored and exists." Second, participants would only record as much as they deemed to be sufficient for the required work to be done. One respondent describes the limited recording as recording "on a meta-level." --- Discussion The results illustrate that the content recorded depends on the relevance to the case. Relevance is assessed by the individual professionals. Thus, on the one hand, recordings and data transfer to third parties may occur even though the clients refuse this. On the other hand, information may be withheld or only recorded at a meta-level in the CIS. The reasons given by the participants may be understandable and plausible for members of the professions; however, this could be a challenge for people who want to process this data in the context of algorithmic analyses. Therefore, the following section attempts to identify a mechanism for this behavior. For this purpose, I draw on Rössler theory of privacy,4 which links informational self-determination with the relationship level of individuals and extend this theory in light of the above findings. --- Mechanism of How the Data Ends Up in the CIS According to Rössler , informational privacy is not only a yardstick for distinguishing between public and private relationships, but it is also necessary in a functional sense in order to be able to speak of a private relationship at all. Based on the relational theory of privacy, according to Fried and Rachels , she links the concept of informational privacy to "the idea that privacy is a function of relations between individuals" . Therefore, the decision regarding what information to share with others is not only subject to the personal discretion of the individual person, but also based on mechanisms "constituted and governed by standards external to and exceeding individual control" . In the context of planning social care services for people with disabilities, for example, it is necessary to know whether a disability exists; and if so, what effects the disability has on the participation of the service user. Within a professional setting, this information has to be shared with members of funding agencies as well as of service providers. As Rössler points out, the mere sharing of intimate information does not mean that the relationship between the actors involved must necessarily be intimate as well. For example, the relationship with a professional or a stranger is not intimate despite the sharing of intimate information . It is nonetheless common for strangers outside of this professional setting to be unaware of these individual challenges or the resulting social care services. Interaction between strangers is normatively regulated by "respect for the privacy of others and the retention of one's own privacy " . In contrast to both the interaction between strangers and the professional setting, intimate relationships are characterized as spaces in which a person tries out "what a self-determined and authentic life could be, and determines which staging's of selfpresentation would be possible, desirable, authentic, and so forth" . According to Rössler, intimate relationships arise "in a setting characterized by friendly affection or love, by concern and consideration, and by a special form of interest" 1:26 Digital Society translation). Intimate relationships are found "in a particular context of affective devotion, sympathetic attitudes, and commitments entered into between the persons involved" . Classically, the relationship to family, friends, and intimates is private; and these intimate relationships "need to have their privacy protected precisely because one of their functions involves retreating into a private world" . The relationship in social work is also "more dependent on an undisturbed, trusting relationship with the person seeking advice than in almost any other professional field" . Therefore, the following consideration argues that professional relationships in the context of integration assistance can also be potentially one-sided intimate relationships in rare cases. This consideration contrasts with the idea of professional relationships as outlined by Rössler and Roessler and Mokrosinska . Therefore, it is helpful to visualize the particular setting within social work-for example, in the context of planning social care services for people with disabilities. In contrast to the usual setting between a professional and clients that is clearly defined both spatially and temporally, social workers and clients meet in a variety of settings depending on the needs of their clients. In certain situations, it may be helpful or even necessary for the social worker/professional to come to the client's home5 . In assisted living, the distinction between living and working spaces can no longer be determined clearly: what is the living space of one is the working space of the other. This means that some of these social care services take place not only metaphorically, but also actually in a space that is perceived as particularly worthy of protection: one's own home . Even when these conversations take place on the premises of the organization, encounters can occur that, according to Rössler, take place in the symbolic space of intersubjective confrontation. The symbolic space of intersubjective confrontation belongs in the intimate setting because it must be understood as constitutive of the identity and autonomy of the person concerned . According to Appelbaum ), this refers to a deontological argument for privacy: "The ability to speak freely with another person about one's innermost thoughts, fears, and passions is clearly dependent on the belief that one's revelations will go no farther. Creating a space within which this sort of dialogue can occur is likely to facilitate the conscious exploration of alternative modes of thought and behavior on which truly autonomous functioning rests." In such situations, the relationship between professionals and clients can become almost informally friendly. Some social workers even see this as a quality feature of their work relationship . In such settings, it can happen 1:26 Digital Society that the clients undertake such attempts at self-presentation that Rössler locates in the intimate setting. Statements in the interviews that something was told "in confidence" testify to this intimacy of the relationship. The practice of neutral recording also underlines the respect for clients. In this context, publicity does not begin only when information gets into the hands of third parties outside one's own organization, but already at the time of writing it down in CIS, because "certain others" can view this private information about the person in question. This confirms that in certain cases the duty of confidentiality also seems to apply to one's own colleagues. The knowledge gained in such intimate settings must therefore be assessed in terms of its relevance for professional work and then, if necessary, transformed into the CIS . If the person recording receives the information in an intimate setting, recording of the contents is unlikely but possible. It can be assumed that professionals perceive existing records in such cases as, for example, "spite" , gossip, or collusion . If the information provided turns out to be relevant for future professional work, recording becomes necessary. For the person recording, this can be a dilemma: On the one hand, control over relevant intimate information must be relinquished, because there is a certain obligation to disclose within the application procedure in order to be able to demonstrate plausible and comprehensible legitimacy for social care services to provide support for participation in social life. On the other hand, the information cannot be disclosed without jeopardizing the confidential relationship with the client and being guilty of collusion, as is evident in some of the quotes above. As a potential solution, respondents seem to see themselves as so-called gatekeepers or protectors of the clients-not only with regard to strangers or third parties not involved in the planning process of social care services, but also with regard to their own colleagues and the CIS they use in their professional work. One consequence of this way of thinking and acting is, for example, that certain information is withheld to protect privacy and the confidential relationship with clients. Therefore, information is not even recorded electronically , even though it is important for further case processing. Instead, important and relevant information is summarized and noted by professionals "on a meta-level" . Cairns et al. , Huuskonen and Vakkari , and Zanchetta et al. also report both selectively omitting and recording on a meta-level, as mentioned earlier. The above results underlines the consideration that this behavior is not exclusively a response to shared electronic health records , but also occurs when only members of one's own organization have access to this data. Gillingham also pointed out that not all data about someone in the context of social services is available electronically, and therefore all one will ever see in the data is a partial version of what happens in a case. Nonetheless, this behavior may become problematic if the available data in the CIS is to be used to decide on which social care services to deliver for participation. This is because the information that is sometimes withheld may be needed to describe the concrete needs of the client in a comprehensive way-only then can certain services be financed at all. Therefore, full omissions can be understood only as case-specific negotiation processes between all actors involved , and they are therefore probably rather the exception. In particular, the professionals working for funding agencies are challenged: Is the information included in the application for social services in written form sufficient for them to come to a judgment about the decision on social services? On the one hand, if it is possible, the privacy of service users will be protected and guaranteed. On the other hand, due to selective omissions of personal and sensitive client information in the CIS, there will be probably little evidence for an algorithmic system to comprehend specific client situations and reconstruct professional judgments. --- Limitations The research is limited by a relatively small sample of social workers who work in a particular context in Germany, and so no generalizations can be made about the findings. Further research in different social work contexts is required to confirm, refute, and possibly build upon the findings. The research was also limited by its reliance on participants' accounts of their practice. Further research could involve case file audits to determine, in more detail, what is and what is not included in case records and its relevance to the development of algorithmic tools. --- Conclusion The aims of this study were to explore the kind of information that is recorded in a client information system in day to day social work practice and to identify the information that will not be part of records. It was found that the data which is recorded may be stipulated through legislation and internal organizational procedures. There may also be technical limitations associated with both the hard and software used to record information about service users and delivery which prescribe and sometimes limit what can be recorded. At the level of individual practitioners, what is recorded and what is left out are subject to professionalspecific ethics, such as privacy and confidentiality. A key finding was that, particularly in relation to personal and sensitive information, social workers have to decide whether to record information with reference to what they consider to be "relevant." From the data, it appears that "relevance" is subjective and may differ between practitioners. Two significant inferences can be drawn from the findings that relate to the development of algorithmic tool in social work. Firstly, bias does not begin with data mining or analysis, but is embedded in the decisions made by professions about what to record. AI applications are, therefore, characterized by a "double subjectivity" . An important step could therefore be to raise awareness among people who further process the recorded information about how data is recorded in practice. Secondly, the findings show that the challenge of ensuring privacy arises not just in the process of AI data analysis but is integral to the process of case recording. This conclusion, familiar to experts as far back as Garfinkel, is especially elementary for individuals unfamiliar with the customs and record-keeping practices of social workers to understand where the limitations of algorithmic analyses based on case records might lie . As the results illustrate, in some special cases, privacy may be the result of negotiation processes by professionals who choose not to enter information into a CIS. To identify such behaviors, it would be necessary to examine whether and to what extent ethical values such as privacy, data protection, and confidentiality are reflected in the records of specific cases. In this regard, it is not sufficient simply to refer to existing legal regulations and professionspecific guidelines or requirements , because these do not contain any information about the actual recording practice. If we are aware of these two points, it becomes clear that in the current debate on AI, we should focus not only on the ethics of design but also on the ethics of usage of AI. Understanding the record process in social and health care can be one element in getting a more realistic idea of what kind of analysis is not possible due to lack of information . --- Availability of Data and Materials The datasets generated and analyzed during the current study are not publicly available due the fact that they constitute an excerpt of research in progress. In justified individual cases, excerpts of the anonymized data can be made available; the interviews were conducted in German. A complete insight into the data is not possible due to the guaranteed anonymity of the study participants. Code Availability Inspections of the coding data system supporting the results of this study is available in justified individual cases upon request from the corresponding author. --- --- Consent to Participate The participants of the study were all of age and fully capable of giving consent. They had agreed to the interview in writing and after informed consent. The data protection clarification and information on the planned interview was available at least 24 h before the interview was conducted; this included information on data protection aspects, the rights of the participants, and the voluntary nature of participation. The participants had the opportunity to ask questions and to terminate the interview at any time without giving reasons and without adverse consequences. --- Consent for Publication The participants agreed to anonymized publication of individual quotations as part of the informed consent. As the interviews were held in German, the author provides a translation of the meaning in case of quotations. --- Conflict of Interest The author has no relevant financial or non-financial interests to disclose. This article was written as part of the MAEWIN project at the FH Bielefeld University of Applied Sciences. In the meantime, I work at the Fraunhofer Institute for Systems and Innovation Research ISI.

There is intensive debate about the use of AI-based systems in social work, although the degree of digitalization is low in many jurisdictions. In this article, the findings of research about the process of case recording in client information systems for social work are presented. Between January and June 2020, 20 guideline-based interviews were conducted with experts working for funding agencies or service providers. A significant finding was that there are intentional omissions of information within digital client records, despite being relevant for further case processing. This finding and the reasons for omission are highly relevant to the current debate on AI, because it extends consideration beyond the ethics of design to include the ethics of usage.

Background Every year about 10 million dementia diagnoses are confirmed worldwide and estimates suggest that the number of persons with dementia will almost triple until 2050 and then reach about 150 million [1]. In European countries, the need for care and support is increasing simultaneously, largely provided by informal caregivers in home-based care arrangements [2]. This also applies to Germany, where estimates predict the proportion of the population in need of long-term care to increase by 40-70% until 2050. The majority of persons with dementia are living at home and are cared for by informal caregivers [3]. The potentially overwhelming care-related demands and their adverse health consequences are well documented. Informal caregivers of people living with dementia face multiple disadvantages in physical, psychological, social and economic domains [4][5][6][7]. Furthermore, it has been shown that individual domains might reinforce each other negatively. This relates, for instance, to the transmission effect of stigmatization on the perceived psychological burden, when social disadvantages lead to psychological problems [8]. This may also be the case for the interaction between economic and health problems. Another example is chronic stress triggered by long-term financial problems and its adverse health consequences. Therefore, tailored psychosocial support services are of crucial relevance to ease health problems of informal caregivers, to promote quality of life and social inclusion of people with dementia and to strengthen home-based care arrangements [9][10][11]. The role of the dyadic relationship structure between persons with dementia and their caregivers received more attention in recent years. Due to the mutual influence both individuals have on each other, it seems to be reasonable and effective to design psychosocial support in a dyadic manner rather than focusing on one of the two individuals separately [12,13]. Furthermore, research suggests that dyadic structures of home-based care arrangements are heterogeneous and differ with respect to socio-demographic, relational and regional aspects of the dyadic relationship. Those variations in the dyadic structures can cause significant differences in relevant clinical outcomes for both individuals [14][15][16][17][18][19][20][21]. Based on these findings it is reasonable to suggest that there are typical structural features in dyadic relationship constellations and that meaningful subgroups of dyads can be identified along these features. A multi-dimensional representation of dementia dyad subgroups would offer the opportunity to take a differentiated view on life and care situations and to further develop psychosocial support structures in a tailored approach. However, a multidimensional target-group-specific tailoring of support programs to particular subgroups of dyads has hardly been implemented in practice yet [22,23]. The relatively small effects measured so far for psychosocial interventions may be due to insufficiently targeted interventions to subgroups of persons with dementia and their informal caregivers [14,22]. Thus, approaches that focus on particular subgroups might be more effective than so-called one size fits all interventions [22,24,25]. Overall, there is still a lack of knowledge about the heterogeneity of dyadic relationship constellations in home-based dementia care settings. With the German DemNet-D study, data on a large number of communitydwelling dyads of informal caregivers and persons with dementia in Germany was collected in the years 2012 to 2015. Using a model-based clustering procedure, the present study aims to provide an evidence-based analysis of subgroups of dementia care dyads. An increasingly popular method in quantitative health and health care research for identifying subgroups in data sets is latent class analysis . LCA is a crosssectional latent variable mixture modeling approach and serves as an effective procedure to identify a small set of underlying subgroups characterized by multiple dimensions, which may differ for instance in their needs for support [26][27][28]). In LCA the term subgroup is often used as synonym for the term class. The approach has already shown its potential in caregiving research, for example to investigate different classes of service users [29,30], caregiving experiences classes [31] or caregiver profiles [32]. The aims of the present study are a) to identify different classes of dementia care dyads in order to develop and describe a multidimensional typology and b) to investigate differences between the classes with regard to relevant healthcare-related outcomes. The study intends to contribute to the ongoing discourse on the adequate tailoring of psychosocial interventions to certain subgroups of persons living with dementia and informal caregivers [22,33,34]. --- Methods --- Sample and data The analysis was performed using cross-sectional data from the DemNet-D study . This multidimensional and multidisciplinary evaluation study investigated the care and living situation of community-dwelling persons living with dementia and their informal caregivers as service users of 13 regional dementia networks in Germany. Ethical approval was obtained from the ethics committee at Ernst-Moritz Arndt University in Greifswald . The present study includes data from 551 community-dwelling dyads . Originally, the DemNet-D data set contains 560 dyads. We excluded nine of these dyads because professional legal guardians and other formal caregivers provided information. Table 1 summarizes the inclusion criteria for the study participants. Both, persons with dementia and the informal caregiver had to submit written informed consent to participate in this study. Participants were identified and contacted by staff of the dementia care networks. More detailed descriptions of the original study population and the data collected have already been provided in previous publications [35,36]. --- Measures Persons living with dementia: We used the Functional Assessment Staging , designed to assess the stage of dementia . The higher the value, the more severe the dementia [37]. To assess the intensity of agitated behavior in people with dementia we used the CMAI-D . With the CMAI-D 30 typical behaviors of persons with dementia are surveyed, recording their frequency [38]. Everyday competencies were assessed using the IADL scale . A higher value indicates a higher degree of autonomy [39]. Furthermore, we used socio-demographic indicators, including age, gender and social class. The latter, using information by the person with dementia, was operationalized according to the Scheuch-Winkler Index [40,41]. Informal caregiver: As sociodemographic indicators we used age, gender and occupation . The relational indicators are the informal care relationship and the living situation . Furthermore, we included the size of the informal support network . As carerelated indicators we used the time informal caregivers spent for care and support on an average day, an item taken from RUD questionnaire [42] as well as the duration of care . Region: External data was linked via postal codes to examine potential influences of non-personal, regional factors. To analyze regional socioeconomic inequalities in health we integrated the German Index of Socioeconomic Deprivation , which uses regional data to group regions into the three deprivation groups high, middle and low [43]. Furthermore, we created two quotas as proxies for regional nursing and medical infrastructures. For the nursing quotient, we used the number of persons in need of care per employee in outpatient care services per district or city and the national mean value to group districts or cities with a below average or above average rate. The medical quotient was calculated based on the number of contract physicians and psychotherapists per 100.000 inhabitants. Again, we applied the national mean value to form a group of regions above the average and another group below the average. Data was acquired from the German Federal Statistical Office, the Federal Institute for Research on Building, Urban Affairs and Spatial Development as well as the Federal Register of Physicians. --- Outcome measures Persons living with dementia To assess the quality of life of persons living with dementia, the QoL-AD was applied, using the proxy measures provided by informal caregivers . A higher value indicates a better QoL [44]. Furthermore, the Sense of Acceptance in Community Activities tool was used to assess the social inclusion of persons living with dementia. SACA is based on self-assessment by the person with dementia . Higher values indicating better social inclusion [45]. --- Informal caregivers Subjective caregiver burden was measured with four subscales of the field version of the Berlin Inventory of Caregivers' Burden with Dementia Patients designed for counselling services [46]. The four subscales 'burden due to cognitive losses', 'burden due to aggression/resistance', 'personal constraints/health' and 'lack of social recognition' were used instead of a global scale to be able to detect different pathways of perceived caregiver burden. In addition, for the present study the visual analogue scale was used to assess the overall health status of informal caregivers . Higher values indicating better health status [47]. --- Care Arrangement The stability of the home-based care arrangements was assessed by the informal caregivers using a four-level scale developed by von Kutzleben et al. [48]. The caregivers are asked to evaluate their monthly supply situation in general. The scale ranges between the poles "situation is well arranged, no further assistance necessary" and "care at home doesn't work anymore, transfer to a nursing home is intended". --- Service use We analyzed health care services used to identify potential support opportunities. Therefore, the use of medical and therapeutical services within the last 6 months and the current use of nursing and support services was assessed. Furthermore, we examined which sources of information dyads have used so far to gather information about dementia services. If at least one of the services of the respective category was used, the category was rated "use". The use of sources of information as well as the current use of nursing and support services were measured with items taken from the D-IVA instrument [49]. Single items from caregiver questionnaires of the DemNet-D study were applied to assess the use of medical and therapeutical services within the previous 6 months [41,50,51]. --- Statistical analysis To identify subgroups of dementia dyads a latent class analysis was performed [52]. LCA is a modelbased statistical procedure using item response patterns for classifying individuals into different classes that are as homogeneous as possible. The classification is based on observed variables . Relationships between values of the indicator variables are attributed to the existence of an unobserved latent variable . Classes were evaluated based on criteria indicating a good classification and model fit. First, a low Bayesian Information Criterion , where lower values indicate better fit. Second, a good relative entropy , where higher values indicate better classification certainty. Third, a reliable class assignment based on consensus build through discussion within the research team [31]. We used BIC as the main indicator to evaluate model fit and as the main criterion for model selection . Additionally, we used the likelihood ratio test to control model selection through BIC. The entropy of the model was also calculated as a model fitting measure. The main criterion for comparing different class solutions, for instance 4-class model vs. 3-class model, was a decreasing BIC value. With the likelihood ratio test, we subsequently controlled, whether the hypotheses that the 3class model fits the data better as 4-class model can be rejected . Besides the identification of subgroups, latent class analyses allow the examination of the association of a latent variable with other external variables. Given the variables listed in Table 2 and study aims, each external variable was hypothesized to be dependent on the latent classes. To estimate these associations, the three-step procedure was used [53]. We also investigated the dyad classes' potential association with other dyadic variables given in Table 1. The three-step algorithm consists of the following steps: In step 1 LCA model is run for indicator variables and the model parameters are estimated. In step 2 dyads are assigned to the estimated classes based on their posteriori class membership probabilities and in step 3 potential associations between these classes and external variables are investigated using multinomial logistic regression analysis. The maximum-likelihood estimation was performed with the aforementioned adjusted three-step LCA, which is a modification of the Dementia severity 4 Regional socio-economic deprivation Regional care and medical infrastructure three-step procedure and corrects the bias by using the stepwise estimation [53][54][55]. --- Procedure of latent class analysis In a first step, we estimated the number of classes, class size and class structure. A hierarchical list of indicator variables formed the basis for this. The hierarchy was developed based on previous research findings [56] as well as study objectives. Variables were added gradually to the model calculations. The first indicator variables were used to determine the optimized number of classes and thereby the main characteristics of the classes. This number of classes was set despite adding further indicator variables of levels 2-4. Following this procedure, the optimized 6-class model based on primary indicators remained, and further indicators only influenced the class characteristics , if they were significant . To reduce the probability of local optima we used the integrated option in Latent Gold 5.1 software and repeated the algorithm with different starting points chosen at random. The inclusion of lower ranking variables was made after all higher-ranking variables had been tested. Therefore, a first LCA was conducted using the six variables of level 1: sex and age of informal caregiver, sex and age of person with dementia, informal care relationship and the living situation of the dyad. Model solutions were assessed using the BIC criterion, the entropy score as well as a meaningful and coordinated interpretation by the research team. To test whether the influence of individual indicator variables on the model is significant, two statistical tests were used. First, the Wald-Test assesses, if the regression coefficients within all classes are equal to zero , which would mean that there is no significant difference between the classes with respect to the relevant indicator variable. Second, we applied the Likelihood Ratio Test . The LRT tests whether the model with six indicator variables represents a significant improvement compared to models with less variables . The variables of level 2-4 were subsequently added to the model. If a variable significantly improved the fit of the overall latent class model, it was included; if not, it was not considered . Finally, we examined the BIC and entropy to test, if the final model with six classes and all variables sufficiently fitted to the data. The Expectation-Maximization algorithm was applied for the maximum likelihood estimation of the model. The EM algorithm uses no imputation algorithm for missing values. The only assumption is that missing data is missing at random . We used all observed attributes for each individual case and made no imputation and no removing procedures for missing values. The latent class analysis was performed with the software Latent Gold® version 5.1. --- Results --- Baseline characteristics We included 551 home-based care dyads in this study. Table 3 shows baseline sociodemographic and clinical characteristics. Overall, the mean age of the persons with dementia is 79.5 years . More than half of the persons with dementia are female . Median QoL-AD score is 28.7 and the median score for social inclusion, measured with the SACA scale is 24.4 . The informal caregivers mean age is 64 years , 75% are female. Slightly more than 60 % have no occupation . The mean EQ VAS score is 66.4 . The perceived burden due to a) cognitive decline is 8.1 , b) aggression and confusion is 6.0 , c) personal restrictions is 8.3 and d) lack of social support 7.7 . Sixty-one percent of the dyads live together, 50.6% have a couple relationship. --- Latent class analysis Model fit and interpretation suggested that a model with six classes represented the most precise solution . This analysis showed that the 6-class model scores lower on the BIC-scale as the models with 1-5 classes. Furthermore, the model with seven classes does not fit the data better than the 6-class model. Overall, the 6-class model showed the best BIC value in combination with a significant likelihood ratio test compared with the 3-class model and a high entropy score . The latent class analysis yielded a final best-fitting model solution with six classes based on 15 significant indicator variables . Table 5 shows all indicator variables included in the final model, their respective p-value and the corresponding determination coefficient R 2 . For further characterization of the dyad classes, we used only the indicator variables whose variance was best explained by the model : Sex and age of the person with dementia; sex, age and occupation of the informal caregiver; the informal care relationship and the dyadic living situation. Table S1 Additional file 1 summarizes the best fitting 6-class model solution based on most likely class membership. A crucial step in interpreting the results of latent class analyses is to label and describe the statistically identified different classes concisely, using the decisive indicator variables of the model. In this study, the process involved successive discussions within the project team. The next section focuses on describing the key classbuilding indicators of the final 6-class model solution, based on most likely class membership. Class differences are also shown in Table S1 . . Figure 1 provides an overview of the six classes identified . --- Associations with outcome measures Following the identification of the best fitting LCA class solution, the associations between the six classes and several outcome measures were investigated. --- Use of information sources There are no significant differences between the dyad classes in terms of using nursing care services as an information source for care-related issues. Similarly, no significant differences in the use of civil society sources can be observed. The use of medical information sources as well as the use of informal information sources differs significantly between the six classes a . --- Health care service use There are no significant differences between the classes with respect to the use of support services. Significant differences between the classes can be observed for medical services, therapeutical services and nursing care services. In general, it can be shown that the use of nursing care services is more prevalent in adult child-parent relationships and non-family relationship than in couples . --- Quality of life of persons with dementia The QoL of persons living with dementia, rated by the informal caregivers, differs significantly between the classes . Persons living with dementia of classes 4, 5 and 6, who live in intragenerational couple relationships with their informal caregiver, show a significantly higher quality of life than persons living with dementia of classes 1, 2 and 3. The lowest value occurs for caregivers in class 2, which simultaneously express relatively high caregiver burden. --- Social inclusion of persons with dementia Concerning the perceived social inclusion of persons living with dementia no significant Health status of informal caregivers The general health status of the informal caregivers is worse in the intragenerational spousal relationships of classes 4, 5 and 6 than in classes 1 and 2 with adultchild relationship or class 3 with a non-family relationship. Furthermore, it is significantly worse than in classes 1 and 3 with the youngest groups of caregivers. Caregiver Burden Subjective caregiver burden differs significantly between the six classes . The informal caregivers in class 6 and 2 feel most burdened. In contrast, the caregivers in class 3 indicate rather low levels of subjective burden. The burden scores for the other classes are mainly in the moderate range. --- Stability of care arrangement The subjectively perceived stability of the home-based care arrangements differs significantly between the six classes . In class 3, informal caregivers rated the situation as more stable compared to other classes, especially class 1, 2 and 6. Table 7 below summarizes results regarding the clinical outcome measures. --- Discussion In this study, LCA was applied to a large, German sample of persons living with dementia and their informal caregivers. The analyses aimed to a) identify different classes of dementia care dyads in order to develop and describe a typology and to b) investigate differences between the classes with regard to relevant healthcarerelated outcomes. Six meaningful classes of dementia care dyads were identified and differences between them have been examined regarding healthcare-related outcome measures. --- Identification of dementia care dyad classes Informal caregivers and persons living with dementia are often treated as a relatively homogeneous group in intervention research, while their diversity regarding personal, relational and other aspects characterizing life and care situations remain largely invisible [22,25,57,58]. This study shows that informal dementia care dyads differ in terms of factors related to individual persons as well as to relational factors . Our results suggest that these differences among care dyads should be carefully reviewed in the design and implementation of counselling and support services. The key characteristics distinguishing the classes from each other are care relationship between informal caregiver and persons living with dementia and age of the informal caregivers . In addition, the dyad classes differ significantly with regard to gender relations, the living situation and the occupation status of the informal caregivers. Several of these indicators have also been identified as relevant distinguishing characteristics in previous research. For instance, with regard to the age of the informal caregivers [18], the dyadic gender relation [59], the kinship relationship [14], the living situation [20] or the occupational status of informal caregivers a The ranges of the scales used are listed in brackets following the names of the corresponding instruments. The maximum scores are underlined. Interpretation: Qol-AD: The higher the value, the better the quality of life; SACA: The higher the value, the better the social inclusion; BIZA-D subscales: The higher the values, the higher the burden; EQ VAS: The higher the value, the better the health status; Stability of care arrangement: The higher the value, the more stable the care arrangement b Class values presented here are mean values. In parentheses: Standard deviation [15]. However, the present study goes beyond the findings of the above-mentioned studies by exploring the dyadic constellation in a multifactorial way and analyses the characteristics not isolated but in a combined multivariate and model-based latent class analysis. --- Comparing outcome measures between classes Statistically significant differences between the identified dyad classes can be observed both in terms of health care service use as well as in terms of other important outcomes. Regarding the dimensions of health care service use, the use of medical information sources and the use of informal information sources differed significantly between the six classes . Further significant differences were observed in the quality of life of persons living with dementia. Assessed by the informal caregivers, persons living with dementia of classes living in spousal relationships , show a higher quality of life than persons living with dementia in adult childparent dyads do. The lower QoL-AD assessment in classes 1, 2 and 3 might be due to the higher mean age of the persons with dementia, which is in turn common for adult-child informal care relationships [14]. For a progressive condition such as dementia, even if there is no direct causality, typically more severe symptoms arise with very old age , associated with increasing limitations in everyday life [60]. These factors in turn have an influence on health-related assessments made by proxies like informal caregivers [61]. It is noticeable that the informal caregivers in classes 4, 5 and 6 rate their own health status relatively poorly, but at the same time the QoL of persons with dementia were rated highest in these classes. Other studies point to the inconsistent evidence on the influence of proxy health status and assessment of QoL in people with dementia [62]. The results of this study do not provide a clear if and how this might be linked directly. The health status of the informal caregivers is lower in the intragenerational spousal relationships than in the intergenerational adult childparent relationships , and significantly lower than in classes 1 and 3 with the youngest informal caregivers. This is consistent with the results of Pinquart et al. [14], who examine differences between spouses, children and children-in-law in their meta-analysis and find that spouses report greater physical burden than children and children-in-law in the same role. Norton et al. [63] explain those differences inter alia by the own medical co-morbidity, greater risk for functional limitations, and greater likelihood of fatigue at physical exertion. In this context, however, the higher age of the informal caregiver should not be ignored, which often goes hand in hand with the spousal relationship. Studies show that informal caregiver in spousal relationships are usually much older than caring children/children-in-law of persons living with dementia [14]. A large age difference between these informal care relationships is also evident in this study. Most dyads assessed in classes 4, 5 and 6 live together in one household, which in turn could make it more difficult for informal caregivers to distance themselves from the care situation and take care of their own health. The subjective burden of informal caregiver also differs significantly between the six classes. The informal caregivers of class 6 and class 2 perceive the most severe burden but on different pathways. Based on BIZA-D PV subscales used, class 6 caregivers express relatively strong burden due to personal constraints and a lack of social recognition. This subgroup is made up of younger couples that have to deal with dementia at a very early stage of life. Caregivers in those informal care relationships report high levels of stress and burden due to the necessity to readjust work and care and redefine the intimate spousal relationship [64,65]. The dementia diagnosis occurs early in life when it is not expected, which can lead to a financial and psychosocial crisis for the family system [64,65]. For class 2, other mechanisms for increased burden might occur. The persons living with dementia in this group are by far the oldest in comparison with other subgroups. Here, severe dementia-related symptoms are often responsible for higher levels of burden, since informal caregivers have to provide extensive efforts to ensure care [66]. The caregivers of class 3 with non-family relationships and younger informal caregiver, on the other hand, indicate comparatively low levels of burden. One explanation for this could be indirect relief effects of so-called "legitimate excuses" [67]. Legitimate excuses reflect social norms that are widely accepted by society and internalized by acting persons. In the context of informal care, for instance, geographical distance and professional activity are among the legitimate excuses for being less confronted with expectations to take over care tasks. This again may account for the comparatively low burden values of the class 3 and the relatively high values of class 5 and 6, both couple-dyads living in the same household. Similarly, male gender represents a socially accepted legitimate excuse, because men are usually less likely to be confronted with society's expectation of having to provide care for the elderly [67]. The differences in burden between classes 4 and 5 can also be seen against this background. In three out of four subscales, the reported burden values for female informal caregivers of class 5 are higher than in class 4 with male caregivers. This is despite the fact that both subgroups are very similar in other respects. However, another reason for the differences between men and women could be that men are less likely to report burden, which might be due to gendered coping strategies where men are more likely to negate stress or negative feelings [68]. There are also significant differences between the six different dyad classes for the assessment of the stability of the care arrangements. For instance, the stability of the care arrangements was rated better in in class 3 with non-family relationship and younger informal caregivers compared to other classes, especially class 1 and class 6 . A reason for the lower stability scores in class 6 might be the increased psychosocial stress, reflected in the stated subjective caregiver burden. This relationship has already been described [69]. Despite positive ratings for both caregiver burden and their own general health status, the caregivers in class 1 perceive their care arrangement as relatively unstable. This seems somehow contradictory but cannot be explained by the variables investigated in this study. A current meta-study aims to explain the complex phenomenon of stability of home-based care arrangements for persons living with dementia and could help to better understand this contradictory finding [70]. --- Limitations Some factors limited the findings of this study. Although a large data set was used, it was not based on a random sample and therefore has some limitations with regard to the statistical generalizability. However, a comparison with other surveys in Germany shows that the DemNet-D study represents a valid sample of home-based dementia patients in Germany in terms of basic sociodemographic characteristics, but differs in terms of, for instance, subjective burden of the caregivers or use of health care services [35]. Since we conducted a secondary analysis, we had to deal with possible restrictions due to the original data, . Further research on informal dementia care dyads should also add relational indicators, such as relationship quality, relationship closeness or positive and negative dyadic interactions, as they are relevant for coping behaviors that in turn moderate health related outcomes [7,71]. As we used only crosssectional data, future research should add a longitudinal perspective; i.e. to analyze changes in clinical outcomes within different dyad subgroups [10]. This would be important to take into account the progressive quality of care processes [48]. It should also be noted that the data used for the analysis was collected in 2012 and thus is about 10 years old. --- Conclusion It is crucial for researchers in the field of dementia care to be able to identify subgroups of dyadic constellations within the overall population, which have similar patterns of life and care situations. This approach allows to further investigate specific interrelations, e.g. which subgroups of dementia dyads are commonly found in the general population, how prevalent they are, what characterizes them, what future bio-psycho-social health outcomes can be predicted, and whether those outcomes change over time. A latent class perspective can provide important insights about key underlying population subgroups, to subsequently tailor prevention and treatment as well as psychosocial support programs to their needs. The typology developed here allows a multidimensional description and thus a more comprehensive representation of dyadic subgroup constellations in home-based dementia care. The results indicate the need to approach dyad subgroups specifically in terms of counseling, service use promotion and easing of caregiver burden. Recent research findings on counselling services in the care and nursing context highlight the need for a comprehensive assessment of care situations and to derive support services needed [72]. Health and social care providers can also benefit from the results of this study. For instance, they can provide a basis for the development of a screening tool to assess specific living and care situations based on typical dyadic care constellations right after a diagnosis. With only a few indicators to be collected, it would be possible to carry out an early initial classification to identify specific dyadic constellations in a time-saving manner. This selection step could be implemented in existing support and counseling structures with little effort and could be carried out flexibly. The results might then be used to describe type-specific risk profiles regarding the health and social situation of dementia care dyads. Derived from this, preventive and relieving support services could be offered. There is no uniform group of dementia care dyads. Dyads rather differ significantly with regard to certain individual, relational and social aspects that lead to differences in relevant outcomes. These differences must be identified and target group-specific interventions to improve the quality of life and reduce the perceived burden of informal caregivers must be made available. A one-size-fits-all approach runs the risk of being too unspecific. --- --- Abbreviations DemNet-D: Evaluationsstudie von Demenznetzwerken in Deutschland; BIZA-D: Berlin Inventory of Caregivers' Burden with Dementia Patients; EQ-VAS: EuroQoL Group Visual Analogue Scale; QoL-AD: Quality of Life in Alzheimer's Disease; QoL: Quality of Life; SACA: Sense of Acceptance in Community Activities; BIC: Bayesian Information Criterion; LCA: Latent Class Analysis; DCN: Regional dementia networks; FAST: Functional Assessment --- --- Additional file 1. This file contains Table. It summarizes the best fitting 6-class model solution based on most likely class membership. Authors' contributions HW, KWO and FA designed the study. WB and FA performed the statistical analysis. HW, KWO, FA, LSW and LV analyzed the data, interpreted the results. HW took the lead in writing the manuscript. KWO aided in analyzing and interpreting the results and worked on the manuscript. JD and RT have read and commented on the full document. All authors discussed the results and commented on the manuscript. The authors read and approved the final manuscript. --- Funding The National Association of Statutory Health Insurance Funds supported this study within a funding scheme to further develop the long-term care insurance. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Open Access funding enabled and organized by Projekt DEAL. --- --- --- --- Competing interests The authors declare that they have no competing interests. ---

Background: Studies revealed the importance to assess dementia care dyads, composed of persons with dementia and their primary informal caregivers, in a differentiated way and to tailor support services to particular living and care circumstances. Therefore, this study aims first to identify classes of dementia care dyads that differ according to sociodemographic, care-related and dementia-specific characteristics and second, to compare these classes with regard to healthcare-related outcomes. Methods: We used data from the cross-sectional German DemNet-D study (n = 551) and conducted a latent class analysis to investigate different classes of dementia care dyads. In addition, we compared these classes with regard to the use of health care services, caregiver burden (BIZA-D), general health of the informal caregiver (EQ-VAS) as well as quality of life (QoL-AD) and social participation (SACA) of the person with dementia. Furthermore, we compared the stability of the home-based care arrangements. Results: Six different classes of dementia care dyads were identified, based on best Bayesian Information Criterion (BIC), significant likelihood ratio test (p < 0.001), high entropy (0.87) and substantive interpretability. Classes were labelled as "adult child parent relationship & younger informal caregiver", "adult child parent relationship & middle aged informal caregiver", "non family relationship & younger informal caregiver", "couple & male informal caregiver of older age", "couple & female informal caregiver of older age", "couple & younger informal caregiver". The classes showed significant differences regarding health care service use. Caregiver burden, quality of life of the person with dementia and stability of the care arrangement differed also significantly between the classes. Conclusion: Based on a latent class analysis this study indicates differences between classes of informal dementia care dyads. The findings may give direction for better tailoring of support services to particular circumstances to improve healthcare-related outcomes of persons with dementia and informal caregivers.

INTRODUCTION Abdulganiy, O., Moshood, I., Yusuf N., Muhammed, A.Y., & Abdul-Rasheed, S. note that a name is a word or a group of words by which an individual person, place or thing is identified. It is the nomenclature used in making reference to all things; tangible or intangible, living or non-living, concrete or abstract, liquid or solid. The study of personal names and naming system is referred to as anthroponomy, a field which is related to genealogy, sociology and anthropology. Under the umbrella of onomastics, a field dealing with the study of proper names, anthroponomy investigates names in their forms, meanings and use aspects . In human sociocultural contexts, naming is done in order to differentiate, to recognize and finally to know. The topic of names is a multidisciplinary one that has occupied the attention of philosophers of language, anthropologists, linguists and ordinary people. Personal names can best be analyzed by a combination of both philosophical and anthropological notions. Societies attach much importance to names and their naming practices. This goes without saying that the knowledge about names gives insight into culture, philosophy, thought, environment, religion and language of particular societies. In the same line of argument, symbolic nature of traditional names and their interpretation depicts religious beliefs of a given society and their interaction with foreign cultures. In this line of understanding, the objective of this paper was to investigate the semantic aspects, naming criteria, and typology in Mopke personal names. --- Linguistic Classification/ Geographical Location --- Linguistic Classification According to Grimes , updated in Maho , Gordon , Mokpe, Bamboko and Wovia are Bantu languages. They are Narrow Bantu and they form the languages of the Duala group. Following the numerical classification. However, contrary to this view and in line with the findings of Moka 1999, andMapeh 2001, this study considers Mokpe, Bamboko and Wovia as dialects of the Bakweri langauge. --- Geographical and Administrative Location Bakweri is spoken in all the four subdivisions in Fako Division. The Bakweris occupy the Western, Central and Eastern slopes of Mount Fako. The following linguistic communities make up the Bakweri language area. --- a) Wovia According to Eberhard et al. , the Wovia speakers are about 600 and are found a few kilometres West of Limbe. They occupy two villages, Wovia and Botaland , along the coast. Before, they used to occupy the pirated islands off the Limbe Coast. Synonyms of Wovia are, Ewota, Bota, Bubia and Bobea. --- b) Bomboko The Bomboko speakers are found in Limbe 1, Limbe 2, West Coast, Mbonge and Muyuka. They occupy 45 villages and number about 36000 speakers. The synonyms of Bamboko are Wumboko Bomboko and Mboko . --- c) Mopke With a total population of over 35000, the Mopkes occupy about 85 villages in Fako division . This figure is not far from what Lewis revealed as the total population of those who speak Mopke. --- LITERATURE REVIEW --- Theoretical Framework This paper takes the aspect of linguistic anthropology and sociolinguistics based on the theory that there is a strong interface between people's language and their cultural practices. In this presupposition, the premise is that language is not only used as a cultural resource and practice, but it is also envisioned as a powerful tool used to view and understand the worldview and philosophy of a particular society. Agyekum, stresses that language can be used as a microscopic lens to view and understand the social practices and dayto-day activities of a society. People's language, culture and thought are intimately interwoven. In this linguistic anthropological intimacy, language is considered as a social tool with the power to evoke realities beyond the literal linguistic content of what is uttered. According to Duranti language is a set of symbolic resources connecting the social fabric and the individual representation of actual or possible worlds; a cultural practice and verbal activities that link and fit verbal activities to the real world. Presumably, this is also true of Bakweri personal names. Since there is a very close connection between the world, which is reality and the word, which is language, linguistic signs are therefore representations of not only words but also to a greater extent the world. In the study, Mokpe names are the linguistic signs, while their sociocultural interpretations represent the real world. Since language and sociocultural practices are dynamic, the Mopke people and their naming system are eventually affected by this dynamism. As a manifestation and description of the complexity and diversity of way of life and practices, people's language is itself manifested in various ways. Such ways include their naming systems , marriage, family, kinship, political aspects, economic issues, occupational aspects, health systems, religious beliefs and practices, law, funeral activities, etc. Languages also depict issues of power, status, gender, policy, language contact and persuasion. Linguistic anthropology may also be used to explain how language allows for and creates differentiations between groups, individuals and identities . This is the case in sociocultural contexts specificity in the general theoretical frame of anthropology. Though naming is considered as a universal cultural practice where every society in the world gives names as tags to its people, it is also specific because how the names are given, the practices and rituals involved and the interpretations attached to the names differ from society to society and from one culture to another. This goes without saying that amid the fact that every society may have their unique naming system, societies with closely related cultures ought to have closely related naming practices. Thus being a Cameroonian national language, Mopke, may share some naming system practices with other Cameroonian linguistic communities though uniqueness among them may still prevail. This study was conducted within the premise of the principle of indexicality and specificity of sociocultural contexts. With the principle of indexicality, language is used as a tool through which people's socio-cultural world is constantly described, evaluated and reproduced . The Bakweri naming system is an aspect of cultural indexicality in which personal names are connected to some aspects of the sociocultural contexts of the society. Indexicality, therefore, is applicable in Bakweri names since they have sociocultural interpretation by referring to personal, temporal, spatial and social deixis. The names have personal deixis because there is often a person 'grandfather/ mother' whom a child is named after. In most African societies, one can always point to an elderly person, related biologically or distantly, in the society whether dead or alive whose name a younger child bears. Some Bakweri names carry temporal deixis in the sense that they may indicate the timing of the birth. For instance, the name "Efufa"for a child born during the rainy season. There are also spatial names referring to localities within the Bakweri community where a child was born. Social deixis on the other hand refers to the social entities of the referent. For instance, "Effange" for a child name after a worrior. Some names clearly depict that the bearer comes from the royal family or from a particular clan. For example, "Etuli" for a child born to a royal family. This paper therefore, seeks to document the social aspects of Mopke personal names, as a contribution to such initiatives like the Bakweri Language and Literacy committee which is committed to documenting the Bakweri language in every possible area and means in order to preserve this very important aspect of cultural heritage. --- Empirical Review Names refer to purely referential and specific elements of human experience. That is, individuals or collective entities, which they designate or denote . In some philosophical and linguistic viewpoints, names have been characterized logically in the absence of social contexts. Such perspectives consider names as arbitrary labels only that refer to certain signified entries, where the signifier and the signified may not share certain intrinsic qualities. This idea is true when we consider situations where people who bear the same name behave differently. In every culture, names have sociocultural contexts that identify the bearers. To backup this supposition, Agyekum quotes Akan maxim; nsεmmↄnentinayεkyεε din meaning 'it is because of criminal acts that names were shared'. This is to say that every person in this world has a name that solely identifies and marks him/her from all other peoples in the world. Algeo aptly points out that "People are almost invariably named. Indeed, a human being without a name would be socially and psychologically less than a fully man." In Saussure's notion, the name is the sign and the denotatum is the signified; simply meaning that, the name is a label that refers to a person. Surprisingly, in some Bakweri communities, by default, people who bear the same family name are supposed to be related genetically and so behave alike; this situation disputes the arbitrariness of names. Obviously proper names refer very specifically to the person who has that name. In this standpoint characterization of names is constant with Saussure's characterization of linguistic signs as arbitrarily connected to their referent . However, should this assertion be true, then names would have no functional correlation with culture . Inversely, this averment is not exclusively true and thus, not the focus of this paper because the focus here is on cultural significance of Bakweri names. The paper asserts that names of this community are not arbitrary but are based on socio-cultural and ethnopragmatic contexts.Various scholars in anthroponomy agree that names have attributes and are attached to referents. This is exactly what pertains in most African cultures where the social and cultural context analyses of personal names strongly reveal the power of names to emphasize social relationships. Personal names are iconic representations of complex social variables that indexicalise and relate to the name and the person. They include sex, hierarchy in birth, circumstances surrounding the birth, timing of the birth, the person's structure, power, status, etc. According to Rymes , the events involved in the naming ceremony and the choice of names given to children have traceable links to the referent, meaning that proper names are both "synchronically and diachronically motivated." In African norms, circumstances and social contexts during the birth of a child are fundamental for parents to give a name X instead of Y. The circumstantial context will be first and foremost gender, then the socioeconomic situation of the parents and the time of birth and their social links with other people down history, among other factors. These are all aspects of the synchronymy. The names then carries this name and since names have social meanings, people expect the bearer to live by it or make positive amends to that name. The varied meanings represent the diachronic aspects of the name. In this respect, it is gainsaying that the meaning of one's proper name evolves through a life history imbued with a lot of transformations and may be intimately linked with the "identity concerns" of an individual or society . It is also true that with time and evolving sociocultural standings, some individuals may be tempted to modify their names or even to take new names at the expense of the old ones so as to satisfy their self. According to Akinnaso , the fundamental reason for naming in human society is to provide a symbolic base for social interaction through a formal means of individual identification. Similarly, Mondende notes that the naming process inan African society is to pass a message or information from the name-giver to the entire family and the community at large. According to De Pina-Cabral, the 'naming processes carry with them social implications concerning what a person is and how he or she should be placed in the world'. In other words, naming in African society is to a great extent determined by sociocultural, environmental, circumstantial and ethno-pragmatic contexts . For example, among the Basotho of South Africa, a child is commonly named after a relative. Hence, names are used to immortalize the ancestors; it is also believed that naming children after ancestors would ensure the inheritance of the social virtues of their grandparents. Also, the Basotho believed that a child is a gift of badimo and thus, a child should be named after a known ancestor as a form of appreciation for a life well spent by the ancestor . Naming practices among the Basotho people of South Africa also serve religious, political and social functions .Accordingly, people expect the name to live the qualities denoted by a given name because of the inherent power of words ascertained in names to reflect the lives of people either positively or negatively. For example, the Asu expect a child named after a dignitary or a chief to behave properly so that nobody makes derogatory remarks about the name in attempt to denigrate it. Such children are also advised to behave well so as to avoid tarnishing their names. This may conclusively mean that names are meant to shape the children's upbringing, behaviour and socialization. Names in the Mokpe community often designate the characteristics of the named, at least at the time of birth. No wonder why people may occasionally modify their names or acquire new names and bynames based on their personal achievements. Names can therefore be clearly understood when placed in sociocultural context. Since names are not arbitrary as some may perceive, analysis of proper names should concentrate more on the functional theory bearing the society and culture in mind. Thus, names can be best understood and interpreted under the sociocultural context, meaning that people who know the language and culture of the people are able to interpret such peoples' names accordingly. This paper is a contribution to linguistic anthropology and to the study of African anthroponomy and the general theory of onomasiology championed by scholars like Asante , Suzman and Obeng among others. The paper is also written with an assumption that "most African societies have similar naming practices that correspond to each other in some ways . --- METHODOLOGY This section explains the methodology that guided the study. It presents such items like participants, instruments used to collect data, data collection procedures and data analysis. --- --- Instruments Data collection in this investigation focused not only on enlisting the names but essentially on the meanings of the names as used in the Mokpe community. Data collection exercise was done in Wovia, Bamboko and Mevio, all located in the South West Region of Cameroon. The selection of the villages based on the assumption that cultural practices in these areas are not contaminated by cultural contact impact. Participant observation, in-depth interview and self-intuition were used in the process of data gathering. The interview guide comprised of open-ended items focusing on identifying traditional names, their meanings and naming criteria. Participant observation was possible and was efficiently conducted since the researcher is a native speaker of the participants' language. Being a native speaker of Bakweri, the researcher used his intuition about names and their meanings in this community. It was also easy to associate with people in their daily activities, rituals, interactions and events to get the needed information in its naturalness. --- Data Collection Procedures In this investigation, interviews were conducted basing on open ended guiding questions so as to elicit naturalistic data from as much free conversation as possible. The interviews involved 20 informants arranged into groups basing on three villages mentioned before. On the other hand, participant observation was used to observe and note down sociocultural behaviour in context with the researcher behaving like a member of the studied community. The researcher easily and effectively participated in various socioeconomic, cultural and traditional activities of the people involved in this project as she is a member of the speech community, born and raised up in the area till secondary school level and has had constant interactions with the people since then. Observations were therefore done without spending much time to establish rapport with the respondents. --- Data Analysis In managing the data, thematic analysis technique flexible data analysis plan that qualitative researchers use to generate themes from interview data -was applied, which involved grouping the names into various categories basing on meanings and projected connotation as well as naming criteria. Accordingly, the findings are presented in subsequent sections in descriptions and explanations. The presentation and discussion based on the typology of names in the Mokpe community encompassing family names, circumstantial names, etc. --- Analysis and Discussion There is a consensus among anthropological writers that naming practice is generally a valued culture among African people , thus, given names are so relevant that some on their socio-cultural characteristics cannot be well understood except one peeps into the dynamic nature of their naming. This study has reveals that nature of naming in Africa is bore out in the fact that a name given to a child or an adult is often determined by different factors ranging from belief in individual spirit, belief in reincarnation, environment, social class, lineage system, ethnic affiliation, gender, day of the week, day of the month, time of the year, conditions of parents at birth, conditions of child at birth and circumstances surrounding the birth, historical events, family occupation, family deity, family name and economic situation, to mention but few. It is in this line of reasoning that the findings are presented. --- Typology of Mokpe Names This sub-section presents the typology of personal names in the Mokpe speech community. It is an empirical and sociocultural descriptive study of names that exist in this community. Considering both purely traditional and contemporary names, the description gives the translations and the ethnographic background of the names. This section also exemplifies the impact of western culture, in the umbrella of religion and modernity in the naming system and practices. A highlight is also made on the implications of the naming practices, especially in gender relations. --- Family Names In most African communities, family names are clan names given to children by their fathers, traditionally after discussion with the grand parents of the new born. However, nowadays, the mother of the child is also involved in the discussion. Among the Bakweri people today, a family name is a reference to the child's grandmother/ grandfather or any other important person. Example 1 illustrates family names specifically indicating a clan from which they name descends. --- Circumstantial Names There are certain names which are given based on the circumstances surrounding the birth of a child. These may relate to the places of birth, period of time, festivals or sacred days, manner of birth, etc. Sometimes the circumstances are not directly related to the newborn but rather to an ancestral figure; grandfather or grandmother. In the Mopke community, these names can be grouped into four for a better understanding: anthro-toponyms, temporonyms, manner of birth, and death prevention and survival names. --- CONCLUSION AND RECOMMENDATIONS Conclusion This study is an aspect of linguistic anthropology and is based on the assumption that there is a strong connection between people's language and their cultural practices. The emphasis is on how language is used as cultural resources and practices; and how it is viewed as a powerful tool used to view and understand the worldview and philosophy of a particular society. Thus, this goes without saying that language is used as a microscopic lens to view and understand the social practices and day-to-day activities of a given society. The Mopke personal naming system and practice, just like in other cultures, is a marker of the people's belief, ideology, religion, culture, philosophy and thought. The names are best understood and analysed when one has insight into the ethno-pragmatics, sociocultural norms, the language and culture of the Mopke people. The names are therefore not only an aspect of linguistic relativity but also refer to elements of the Mopke human experience and ways of life in a philosophical sense. Also, this paper lays emphasis on the fact that names are not mere arbitrary and meaningless labels but rather have indexical relationship to socio-cultural meanings and functions, places, time, people and events. Traditionally, the Mopke people believe that there is some inherent power and linkage in names, and expect the names to reflect and indexicalise the lives and behavior of people either positively or negatively. This is the case especially when the name refers back to an earlier ancestral figure. An individual's name is of concern not to themselves but to the whole society, for the individual performs and participates in the society. Accordingly, the Mopke typology of names indicates various contexts such as family, circumstantial, manner of birth, flora and fauna, markers of notable events and theophoros names, resulting from cultural contact. It is also worth noting that every potential Mopke person has at least two of these names: one's own name plus a family name. It is also possible for a person to accumulate some other names especially baptismal names as he or she grows up in the society. The Mopke naming practices have also revealed the patriarchy socio-cultural system emphasizing male dominance at the expense of female. Only Mokpe men for instance can acquire family names as personal names while women use family names only as surnames. --- Recommendations This study has demonstrated that Mokpe personal names are an important area of the Mopke culture that should not be ignored in any sociolinguistic and anthropological study. It is also clear that naming system is a multidisciplinary area of study for scholars in sociology, history, religion, anthropology, linguistics, ethnography and philosophy. It is therefore revealed to the respective scholars that a lot can still be investigated about personal names in the Mopke community. Areas such as the place of names in social --- Anthro-Toponyms In this study anthro-toponyms are personal names that pertain to the place of birth of the child. The place may even refer to the farm or along the way when the woman was travelling to another place. It is also possible to use the names of physical features or other geographical areas near the birth place as names for children. --- Temporonym The term temporonym is used for names that relate to the period of birth. In the Mopkecommunity, there are no traditional birthday names. Therefore, what is often denoted is the timing of the birth. In some instances, the names may also depict seasons during which the birth occurs. Just like in the case of anthrotoponyms, newborns may be given temporonyms not necessarily because of the timing of their birth but because it is a tribute to their grandparents. Some temporonyms are given to mark or remember a notable event in the family or clan. This kind of periodic names are very important not only to the name but imperatively to the entire community as they point back to crucial historical and cultural phenomena of the society. The events may denote happiness , sadness or even suffering. Example 3 exemplifies a few temporonyms in the Bakweri community. --- Example3: Mokpe Temporonyms --- Theophoros Names In the Mopke community, just like in many other cultures, there is a belief that what happens in the society is dictated and controlled by a personality outside the horizons of human beings. In many African societies this could be a traditional spiritual leader believed by all members of the society. A good example could be taken from Akan, a West African society in which there is a belief that if a mother suffers constant child mortality, it is the child's mother in the underworld that does not want the child to stay in the living world. To combat such an unfortunate situation, the parents give the child a weird name. They believe that if the name is unattractive, the other mother in the underworld will not like to accept the child over there and this would make the child stay. However, due to contact with western Christian culture, things are different in a handful of African culture where this personality is now referred to as God. In this case, newborns are given names which glorify, praise or give thanks to the said God. This is the case with the Mokpe community, as exemplified hereunder. American Journal of Arts, Social and Humanity Studies ISSN 2959-5827 Vol 3, Issue 1, pp 35 -47, 2023 www.ajpojournals.org

Purpose: Most of our Cameroonian national languages are demographically and socio-culturally pressured by the English language. The national languages which were previously used in elementary education and religious activities currently do not have any place in any official domain and thus are limited to home and other few immediate domains. Due to this unequal coexistence of the languages, many national languages are considered endangered, calling for efforts from stakeholders to prevent the death of such a precious cultural heritage. In this line of understanding, the objective of this study was to investigate the semantic aspects, naming criteria, and typology of Mopke personal names. It is a contribution to such initiatives like the Bakweri Language and Literacy committee which aims at documenting the Bakweri language in every possible area and means. Methodology: To achieve our objective, data was collected through participant observation, in-depth interview and self-intuition to investigate personal names of the Bakweri also called the Mokpe people of the South West Region of Cameroon, considering naming as an important aspect of the society.As regards the theoretical framework, we looked at Mopke names within the purview of linguistic anthropology considering names as not being arbitrary labels but socio-cultural tags that have socio-cultural functions and meanings. In order to analyse and discuss the patterns observed in the data, the Thematic Analysis technique used to present the typology of the names including family names, circumstantial names, theophorous names and flora and fauna names. The paper further examined the changing nature of Mokpe naming system and practice as dictated by cultural contact mainly with Christian/ western culture.The study wraps up with a recommendation for further investigation on issues surrounding naming practices and strategic measures to prevent this important African cultural resource.

Introduction Intimate partner violence against women is a serious but preventable public health problem. It typically includes physical, psychological, and sexual violence [1]. A study including women from 161 countries and areas indicated that 27% of ever-partnered women aged 15-49 years reported physical or sexual IPV in their lifetime [2]. IPV against pregnant women is of particular concern as it affects both women's and children's health [3]. Physical IPV during pregnancy showed a wide range , as did psychological IPV [4]. A recent meta-analysis found a worldwide prevalence of physical, psychological, and sexual IPV during pregnancy of 9.2%, 18.7%, and 5.5%, respectively [5]. These prevalence studies provide useful information about the occurrence of IPV at the population level, but it is unclear how and whether IPV changes across individuals over time [6]. In addition, the variations in IPV prevalence across studies may be due to different timings of IPV assessment [7]. Identifying categories of IPV across time is critical for developing effective and targeted preventions and interventions [8,9], and some research efforts have been made in this area. Jackson et al. revealed that IPV decreased from 13.1% before pregnancy to 11.3% during pregnancy [10]. Another study found that about 21% of women experienced some forms of IPV during pregnancy, and the figure increased to nearly 25% in the first 4 months postpartum [11], providing additional evidence to support the idea that women's exposure to IPV may change during the course of pregnancy and after childbirth. However, it is not likely that all women experience the same type of change in IPV over time. For instance, Johnson's typology of IPV includes several different IPV-related relationships: women in abusive relationships are controlled and abused over a long period, whereas other women may experience relatively short-term stress-related situational IPV [12]. Regarding pregnancy-related IPV, a cross-sectional study of 426 new mothers retrospectively reported their experience of IPV before, during, and 6 months after pregnancy, and it was found that 60.8% of the mothers reported continued exposure to IPV from pregnancy into the postpartum period [13]. Furthermore, a longitudinal study with 1083 pregnant women identified four categories of IPV based on the participants' experience of IPV over the study period: abusive relationship ; relationship with decreased violence over pregnancy ; relationship with stress-related violence ; and nonviolent relationship [14]. While these studies provide us with critical insight into changes in IPV, the follow-up periods of these studies were relatively short. A longer follow-up period is needed to improve our understanding of the trajectory of IPV. Another critical task is to explore the correlates of these groups. Dutton's ecological theory is a widely used model to interpret different systems related to IPV [15]. Previous reviews have indicated that exosystem , microsystem , and ontogenetic factors are associated with IPV victimization [16,17]. Among established findings, adverse childhood experiences are one of the most studied factors linked to IPV. Trauma-focused theories such as the self-trauma model point out that traumatic experiences could affect an individual's development via different mechanisms , which may further affect intimate relationships [18]. A meta-analysis found that ACEs were risk factors for IPV among Chinese women [19]. To our knowledge, very few studies have explored whether ACEs are specifically associated with different patterns of IPV. Such work is crucial to generating a more nuanced understanding of the impact of ACEs on IPV. In addition, we will examine exosystem and ontogenetic correlates across different IPV patterns. Previous work has indicated the association between IPV groups and these variables. For example, pregnant women in a long-term and violent relationship reported more severe depression, lower levels of partner involvement, and poorer social support [14]. However, few longitudinal studies have specifically examined how these factors predict different IPV categories from pregnancy to after childbirth. This study aimed to investigate the changing patterns and associated predictors of IPV across three years among pregnant women in Hong Kong. Specifically, the first objective was to identify categories of IPV based on women's IPV experiences from prepregnancy to three years after childbirth. We hypothesized that different categories of IPV would be identified. The second objective was to investigate risk and protective factors associated with the different categories. We hypothesized that women in a constantly violent relationship would have higher ACEs, poorer health conditions, lower family support, and lower partner involvement. --- Materials and Methods --- Study Design and Participants This was a longitudinal survey conducted between 2016 and 2020 in Hong Kong. Pregnant women were recruited from the antenatal clinic at Kwong Wah Hospital, a public hospital managed by the Hospital Authority in Hong Kong. The inclusion criteria were: all women attended these clinics for their first antenatal visit; they were 18 years old or older; they could understand written Chinese; and they provided informed consent and were willing to provide contact methods for our follow-up survey. --- Procedures Numerous safety and security issues had to be considered. First, we provided informed consent to all participants before participating in the study and told them that they could withdraw from the survey at any time. Second, trained research staff participated in the study's whole procedure. We first approached women in the clinical center, a safe space for participating in our survey. During the follow-ups, we asked the participants whether it was safe and appropriate to maintain privacy and confidentiality while assessing. Third, we provided the necessary information for referral and encouraged women to seek help when identified as abuse victims. The hospital was available to provide professional assistance and consultation to the identified abuse victims if needed. Fourth, data confidentiality was strictly protected. All study data are pseudonymized and stored on protected servers. Only members of the research team can access the data. Lastly, the research protocol was approved by the Institutional Review Board of the Hospital Authority Kowloon West Cluster Research Ethics Committee and the Clinical Research Ethics Review of the Hong Kong Polytechnic University and CRESC201905). In 2016 , 758 participants were recruited and completed a structured questionnaire while waiting for their first antenatal checkup. They also provided their expected delivery date and preferred method to contact them for the follow-up. In 2017 , we contacted the participants again for the follow-up survey. In 2020 , we successfully contacted 340 of them, representing 44.9% retention. The main drop-out reason was that some participants refused to join because of the surge in coronavirus cases in Hong Kong that was due to the pandemic at the time of the follow-up. --- Variables and Measures --- IPV The Chinese version of the abuse assessment screen [20] was used to assess IPV experiences. The AAS has four items, namely participants' experiences of physical violence, psychological violence, sexual violence, and fear because of IPV. Each item was rated either yes or no. Participants were asked to retrospectively report exposure to IPV before and during pregnancy at T1, during pregnancy and up to 4 weeks after childbirth at T2, and three years after childbirth at T3. Three items were used in the current analysis, which is in line with previous work [14]. Participants who reported any positive response to any item were classified as having IPV exposure. --- Mental Health Conditions Depressive symptoms were evaluated using the 10-item Chinese Edinburgh postnatal depression scale [21] . Anxiety and stress levels were measured using the 7-item anxiety subscale and the 7-item stress subscale , respectively, of the Chinese version of the depression anxiety stress scale [22] . All items were rated from 0 to 3, with a higher score indicating higher levels of depression, anxiety, or stress. --- Health-Related Quality of Life Participants' health-related quality of life throughout the three periods was assessed using the 12-item Short-Form Health Survey [23] . A total score ranges from 0 to 100, with a higher score representing better mental or physical HRQoL. --- Partner Involvement Participants reported their partner involvement at both 4 weeks and 3 years after childbirth. Seven items were used, including partner involvement in assisting with daily housework and giving emotional support to the pregnant partner. These items were rated on a 4-point Likert scale. All items were summed, and a higher score indicated more partner involvement. Previous work also used these items to measure partner involvement [14]. The α was 0.87 at 4 weeks after childbirth and 0.90 at 3 years after delivery. --- Family Support Participants' perceived family support at both 4 weeks and 3 years after childbirth was assessed by the Chinese version of the multidimensional scale of perceived social support [24], a 7-point Likert scale from 1 to 7. Four items were used to evaluate family support. We summed all four items, and a higher score indicated a higher level of family support. The α was 0.92 at 4 weeks after childbirth and 0.92 at 3 years after childbirth. --- Adverse Childhood Experiences Exposure to childhood adversities was measured using an adaptation of the Adverse Childhood Experiences-International Questionnaire [25]. The initial validation of the ACE-IQ in the Hong Kong context has been reported elsewhere [26]. Fourteen items were used to assess different domains of ACEs, such as physical abuse, sexual abuse, emotional abuse, physical neglect, emotional neglect, and domestic violence. Each item was dichotomized into "1" = "exposed" and "0" = "not exposed". Items were summed to obtain an ACE score, and a higher score indicated a higher exposure to childhood adversities. The α was 0.61 in this study. --- Demographic Characteristics Demographic information was collected, including age, education attainment, marital status, employment status, whether they were receiving any social security assistance from the government, monthly household income, chronic health conditions, and risk behaviors . --- Statistical Analysis Percentages of IPV experiences from pre-pregnancy to 4 weeks after childbirth and 3 years after childbirth were calculated. As shown in Table 1, a violent relationship was defined as one with exposure to any IPV throughout the period from pre-pregnancy to 4 weeks after childbirth and 3 years after childbirth. A decreased violent relationship was defined as one with IPV reported from pre-pregnancy to 4 weeks after childbirth but terminated at 3 years after childbirth. A nonviolent relationship was defined as one with no IPV experiences throughout all the periods. The Chi-square test or F-test was used to compare distributions or scores of interested variables across the three categories. A three-phase ordinal logistic regression analysis was used to explore the correlates of IPV changing categories. The dependent variable was assigned in the order of "NVR < DVR < VR", as the severity of violence increased in this order [14,27]. In phase 1, we conducted a forward stepwise ordinal logistic regression on all demographic characteristics. In phase 2, we entered the following variables individually after controlling for demographics: partner involvement at T2 and T3, and family support at T2 and T3. Finally, we added the following variables individually in phase 3 after adjusting for demographic characteristics: ACEs, risk behaviors, chronic illness, and health variables . All analyses were conducted with SPSS 26.0. Note. IPV+ = reported IPV. IPV-= no IPV. VR was defined as having experienced any violence from prepregnancy until 4 weeks after childbirth and also at the follow-up . DVR was defined as reporting IPV from pre-pregnancy until 4 weeks after childbirth, but not at the follow-up. NVR was defined as having no IPV experiences throughout all the periods. a We grouped those who did not report IPV experiences from pre-pregnancy to 4 weeks after childbirth but had IPV experiences at 3 years after childbirth into the VR category, as the number was small . --- Results In total, 340 pregnant women participated in all three periods . We compared those who took part in all assessments and those who did not. The results showed that those who completed all assessments had higher IPV percentages and higher exposure to ACEs than those who dropped out. In addition, those who dropped out had a higher level of education. Of the 340 participants , 22.9% reported IPV before pregnancy, 13.5% reported IPV during pregnancy, 14.7% reported IPV 4 weeks after childbirth, and 11.8% reported IPV 3 years after childbirth. As for IPV categories from pre-pregnancy until 4 weeks after childbirth and 3 years after childbirth , 11.8% of women experienced persistent IPV across time , 20.6% of women reported decreased IPV , and 67.6% of women did not have any IPV experiences over these periods . Table 2 shows differences between IPV groups in the study variables, including demographic characteristics, risk behaviors, chronic illness, ACEs, partner involvement, family support, and health outcomes. There were no statistically significant differences in demographic characteristics, risk behaviors, and chronic illness across the different groups except for whether they received social security assistance. Specifically, women in the VR group were more likely to receive social security assistance . Furthermore, there were significant between-group differences in ACEs, mental health conditions, and quality of life . Women in the VR group reported more ACEs and poorer health consequences. In addition, they reported lower family support and lower partner involvement than those in the NVR group. Table 3 presents the three-phase ordinal logistic regressions examining the associations between the study variables and IPV. In phase 1, receiving social security assistance was associated with an increased chance of experiencing more severe IPV . In phase 2, higher partner involvement and higher family support were associated with a lower chance of experiencing severe IPV. In phase 3, after controlling for demographics, pregnant women reporting ACEs was associated with a higher possibility of them having severe IPV experiences . Furthermore, depression at all three periods , stress at all three periods , and anxiety at 4 weeks after childbirth were associated with increased likelihood of the dependent variable, indicating that poorer mental health was related to an increased chance of experiencing more severe IPV. Better HRQoL, particularly in the mental dimension, was associated with lower odds of suffering more severe IPV . Note. T1 = during pregnancy; T2 = 4 weeks after childbirth; T3 = 3 years after childbirth. NVR = nonviolent relationship; DVR = decreased violent relationship; VR = violent relationship. The dependent variable was the type of relationship. It was ordinal: the threshold estimate "1" was the cut-off value between "nonviolent relationship" and "decreased violent relationship" . The threshold estimate "2" was the cut-off value between "decreased violent relationship" and "violent relationship" . PCS = physical composite score; MCS = mental composite score; CEPDS = Chinese Edinburgh postnatal depression scale; DASS = depression, anxiety, and stress scale; ACEs = adverse childhood experiences. * p < 0.05. ** p < 0.01. *** p < 0.001. --- Discussion The overall IPV prevalence decreased from pre-pregnancy to 3 years after childbirth. Yet, when we explored IPV categories across different periods, about 12% of women reported persistent IPV from pregnancy to 3 years after childbirth. These women were more likely to have had more ACEs and to have poorer health than women who did not experience IPV at all periods. Higher partner involvement and family support could decrease the chance of IPV. Our results showed reductions in IPV prevalence from pre-pregnancy to 3 years after childbirth ; this should be cautiously interpreted because of our relatively low retention rate. The literature has inconsistent findings on IPV prevalence from pregnancy to postpartum. A follow-up study in Mexico City found that IPV slightly increased from pregnancy to postpartum [28]. However, another study showed decreased IPV from 32.2% during pregnancy to 25.3% after childbirth [29]. Different cultural contexts may explain these mixed findings. It is possible that women's status differences and the degree of gender inequality within a specific culture differently impact the rates of IPV against pregnant women [30]. Future studies should explore IPV in other contexts. In addition, different methodologies may result in a wide variation in IPV prevalence [13]. As for individual changes in IPV, a substantial number of women continuously experienced IPV from pregnancy to 3 years after childbirth. Our successful identification of different groups resonates with the literature suggesting that exploring individual variabilities in IPV is crucial [31]. Indeed, more recent work has supported that different women may experience different types of IPV [13,14]. Our evidence of continued IPV from pregnancy to 3 years after childbirth can alert healthcare providers that violence is likely to continue for years. Therefore, screening for violence during pregnancy and immediately after childbirth can help prevent future IPV, but it is insufficient to limit screening to only these periods. Our current study suggests that long-lasting screening till years after childbirth is critically important to protect the woman and her baby. Furthermore, we found that women in the VR group were at higher risk of poorer mental health. This finding is in line with existing reviews showing strong associations between IPV and mental disorders in pregnant women [32,33]. In total, 20.6% of women reported IPV in the earlier stages but did not report IPV at 3 years after childbirth. Higher family support and partner involvement could buffer against IPV, which echoes the buffering model of social support. The model points out that stressful events may negatively influence those with little or no support. In contrast, stronger support systems could protect individuals, helping them to recover from stressful events [34]. Specifically, individuals receiving higher social support may be more likely to gain purpose in life and build a sense of control. These individuals could create a more adaptive narrative of stressful events and thus alleviate their negative impacts [35]. It is noteworthy that we did not identify a sufficient number of new cases because of our limited sample. Future studies with more available data are recommended to screen for new cases because they may negatively affect the mental outcomes and behavioral problems of mothers and their children [36]. ACEs are potential risk factors for long-term IPV. Our findings indicated that women who experienced IPV were more likely to have more ACEs than women in the NVR group. The result held even after controlling for demographic covariates. A previous review has evidenced the negative impacts of ACEs on IPV [19]. Notably, this risk might be cumulative: the more ACEs, the higher the possibility of experiencing IPV might be in the future [37]. Moreover, IPV against pregnant women can increase the risk of postnatal child abuse. A longitudinal population-based study in Hong Kong found that IPV during pregnancy was associated with greater odds of both lifetime and preceding-year child physical abuse [38]. Another recent follow-up study in Japan supports this finding [39]. Therefore, prevention and intervention programs for treating past traumatic events are urgently needed to break the cycle of violence. --- Implications Individual differences in IPV across years after childbirth have been identified. Women in the VR group had poorer health conditions than those in the NVR group. This echoes the notion that IPV prevention and intervention programs assuming all women are at similar risk of victimization may be insufficient to provide individualized help [40]. Given that there are different categories of IPV from pregnancy to the postpartum years, we suggest future studies explore specific interventions and services for each category. Continuous clinical screening for violence during prenatal and postnatal care is needed. More attention should be paid to those who report IPV during pregnancy, as it is at high risk of recurring years after childbirth. When IPV is detected, timely and effective programs should be applied. A review of the effectiveness of interventions for IPV around the time of pregnancy found a significant decrease in IPV when applying home visitation programs and some multifaceted counseling interventions [41]. More family support and partner involvement could protect women from IPV experiences. Family-based counseling has great strengths in reducing various types of violence against women by increasing couples' awareness and improving their relationships during pregnancy. A family-based counseling intervention in Iran based on the principles of GATHER is a successful example that illustrates the promising effects of family-based counseling [42]. Additionally, the For Baby's Sake whole-family approach is another promising program to work with parents from pregnancy to two years postpartum to break cycles of domestic violence and improve children's outcomes. Specifically, this approach supports the entire family to end domestic violence, addresses the cycles of domestic violence and abuse , and seeks to improve parents' mental health, infants' emotional and social development, and parent-child attachment outcomes [43]. Experienced childhood adversity could distinguish IPV groups from those in the NVR group. This indicates that treating past traumatic events is critical to preventing future victimization. Trauma-informed care is a comprehensive, multilevel approach that helps service providers and clients understand the impact of traumatic events on health indicators and behaviors [44]. For instance, perinatal care providers are well-positioned to provide trauma-informed perinatal care, which could prevent or reduce the negative impact of ACEs [45]. --- Strengths and Limitations This study has several strengths, including a relatively long-term follow-up to identify different categories of IPV experiences throughout the years after childbirth and to test correlates of these categories. However, limitations should be acknowledged when interpreting our results. First, the use of self-report questionnaires may have led to reporting bias. Second, as a result of the low retention rate, the sample size of the study was relatively small, which prevented us from capturing a sufficient number of participants who only experienced IPV three years after childbirth but not before pregnancy, during pregnancy, or 4 weeks after childbirth. Future studies are suggested to explore the experience of this group of women, as partners may generally be less physically violent because it can harm the baby [46], while emotional IPV may not decrease [47]. Lastly, our results may not be generalizable to other contexts, as various cultural and ethnic factors may lead to different attitudes toward IPV. Future work may consider conducting a multicountry design to confirm and compare our current findings. Despite the limited generalizability of the results, our categories of IPV support findings on pregnant women from different countries [13] and other samples [40,48,49], suggesting that individual differences in IPV exist and even last for years after childbirth. --- Conclusions This 3-year study on pregnant women extends previous work by identifying different IPV changing patterns from pre-pregnancy until 4 weeks after childbirth and 3 years after childbirth. Three IPV categories were identified: VR, DVR, and NVR. Women in the VR group were more likely to have poor mental health problems and more ACEs. Higher partner involvement and family support were protective against IPV. --- Data Availability Statement: Data are available on request from the authors. --- --- Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Intimate partner violence (IPV) against pregnant women adversely impacts women's and infants' health. This study aims to provide longitudinal evidence regarding how pregnant women's exposure to IPV changes over time. Additionally, we examine the risk and protective factors associated with these changes. In total, 340 pregnant women were recruited from an antenatal clinic in Hong Kong. IPV experiences and health conditions were assessed at pregnancy and at both 4 weeks and 3 years after childbirth. The women also reported adverse childhood experiences (ACEs), their family support, and perceived partner involvement. We found IPV prevalence among the study sample decreased from 22.9% before pregnancy to 13.5% during pregnancy, 14.7% at 4 weeks after childbirth, and 11.8% at 3 years after childbirth. We further found three types of IPV: 11.8% of women had a violent relationship (VR) persistently over time from pregnancy to 3 years after childbirth, 20.6% experienced decreased IPV (DVR), and 67.6% reported a nonviolent relationship (NVR) throughout the study period. VRs were associated with more severe mental health problems and higher ACEs. Family support and partner involvement may be protective factors for decreased IPV. Our present findings highlight the importance of identifying different IPV types over time to provide targeted intervention to the most vulnerable groups.

Introduction Under section 31 of the Children Act 1989, care proceedings can be issued if there is concern a child is subject to, or at risk of significant harm, which can lead to removal of a child from their parents. Previous analysis of infants and newborn babies subject to care proceedings in Wales revealed the scale and rising number of families involved and recommended the need for preventative action [1]. Characteristics of mothers of infants involved in such proceedings have also been examined, including mental health needs [2][3][4]. We aim to extend this work by examining a broader range of parental vulnerabilities for both mothers and fathers of children of any age involved in care proceedings. Appropriate and effective health and social support are required to potentially prevent some of the need for care proceedings [5,6]. However, a joined-up health and children's social care response to parents requires far greater knowledge about parents' healthcare needs and their interaction with health, and social care services. This study aims to advance the evidence base regarding interaction with health services by focusing on parents in care proceedings and providing completely new evidence, which will enable services to be more effectively tailored. Combinations of domestic violence, parental mental health issues and/or learning disability, and parental alcohol and/or drug misuse have received considerable attention in relation to risk of child abuse and neglect [7,8]. Skinner and colleagues [9] have recently called for a better understanding of wider factors impacting on families involved with child protection services. More comprehensive evidence of health needs and vulnerabilities, including more in-depth exploration of specific health conditions of parents entering care proceedings and their use of different types of healthcare provision , will also enable enhanced response from the family courts and other services. This study sought to address such evidence gaps with a view to aiding assessment of current policy and its future development. Population-level data collected routinely by Cafcass Cymru for mothers and fathers was linked to electronic health records, to examine demographic characteristics of parents, overall health service use, and health profiles. --- Methods --- Study design and data sources A population-level cohort study with a matched comparison group, with the group of interest being parents involved in public law care proceedings. --- Data sources and linkage Data were accessed via the SAIL Databank [10][11][12], a trusted research environment that hosts extensive individual-level anonymised health and administrative data for the population of Wales. During the anonymisation process of data sources within the SAIL Databank, individuals are assigned unique identifier fields -Anonymous Linking Field and Residential Anonymous Linking Field [13,14] -to link data at individual and residential levels respectively. The primary source of family justice data was electronic case management data routinely produced and maintained by Cafcass Cymru. All instances of s.31 care proceedings initiated between January 2011 and December 2019 were included. Further detail on Cafcass data are available elsewhere [15,16]. Demographic information for parents was obtained using the Welsh Demographic Service Dataset , which provides demographic characteristics of people registered with a general practice . Health records from the Patient Episode Database for Wales , Emergency Department Data Set , Outpatient Data Set Wales and Welsh Longitudinal General Practice were analysed for two years pre care proceedings. These records contain attendance and diagnosis data on inpatient activity, emergency admissions, outpatient appointments, and GP appointments respectively. --- Study population Parents of children involved in s.31 care proceedings in Wales between January 2011 and December 2019 were included in the study . Of these, 9,269 were successfully matched and assigned an ALF. Only parents with valid demographic information of sex, age and deprivation were included. The final cohort consisted of 8,821 parents . An existing method was used [17] to create a list of all parents with children in Wales at a fixed date of 1 st July 2015 and who were not involved in care proceedings. A comparison group of parents was selected from this list using frequency matching at index date). The final matched comparison group consisted of 32,006 parents. Index dates for the cohort were set at the earliest court date, and study mid-point for the comparisons. The baseline period for health data coverage was set as two years proceeding an individual's index date. --- Measures --- Demographic characteristics Demographic characteristics of sex, parental age and youngest child age were derived from Cafcass Cymru and WDSD at index date. The income domain from the 2014 version of the Welsh Index of Multiple Deprivation was used as area level deprivation, taken at or within two years of index date and grouped into quintiles. --- Overall healthcare use Healthcare interactions within the baseline period were analysed for any hospital admission, new emergency department attendances , new outpatient appointment attendance and any GP record. Hospital admissions were categorised into emergency, elective or --- General health conditions We provide a broad categorisation of health conditions grouped according to the chapter level of the International Statistical Classification of Diseases [18]. For emergency hospital admissions, all diagnostic codes in primary or secondary diagnostic code positions were included. As a high proportion of mothers would have had routine pregnancy and birth related admissions we excluded ICD-10 chapter 15 . Any primary care diagnoses codes within the GP Read classification system were included with codes mapped to approximations of ICD-10 chapters . --- Mental health and substance use conditions Parents' primary care and hospital records were examined for the presence of clinical codes indicating mental health contacts or admissions. If an individual had one or more mental health-related contact or admission code recorded during the baseline period, they were categorised as having a mental health outcome. Code lists developed and provided by the Adolescent Mental Health Data Platform [19] were used and included common mental disorders e.g. depression and anxiety; severe mental illness; eating disorders; neurodevelopmental disorders e.g. attention deficit hyperactivity disorder, autistic spectrum disorder; and conduct disorders [20]. Health records were also analysed for clinical codes indicating substance use indicative of problem, harmful or hazardous use of alcohol and/or illicit drugs [20]. If an individual had any such code recorded during the baseline period, they were classified as having a substance use contact or admission. --- Injuries Emergency department attendances were analysed for the presence of injury-related clinical codes during the baseline period, using the attendance group variable for accidents, assault, and self-harm [21]. --- Data analysis Descriptive analyses were conducted to characterise the cohort and comparison groups. Proportions of parents with the measures of interest were calculated during the two-year period prior to the index date; outcomes were not required to be mutually exclusive. One-way analysis of variance tests were computed to compare means between cohort and comparison groups for continuous variables. Chi-squared analyses was used to investigate differences between groups for categorical variables. Data processing and analyses were carried out using SQL and R [22]. --- Results --- Demographic characteristics Mothers accounted for 57.4% of the cohort. Nearly three quarters of cohort parents lived within the two most Age in years 1 3 <0.001 * Given the matched comparison design there were no significant differences between cohort and comparison for sex and deprivation. For all other variables shown in this table p-values were >0.001. deprived quintiles . The mean age of cohort mothers was around three years younger than fathers . There was a notable difference in the proportions of younger parents between the groups, with 8.4% of cohort parents aged 15-19 years compared to 1.3% of the comparison group. The mean age of the youngest child was 1.5 years younger in the cohort, compared to the comparison group. --- Health measures --- Overall healthcare use Both cohort mothers and fathers experienced higher healthcare use across all measured healthcare settings apart from elective hospital admissions in the two years prior to care proceedings . Differences between the groups were generally more pronounced for mothers than fathers. The largest differences between groups were 'reactive' type health services, such as emergency admissions and attendances. Within the cohort, a third of mothers and nearly a fifth of fathers had at least one emergency hospital admission compared to 15.3% and 7.8% in the comparison respectively. Cohort mothers and fathers had higher emergency attendances than comparisons . Cohort parents were also more likely to have higher severity emergency attendances , based on attendances triaged as 'immediate', 'very urgent', or 'urgent'. Since a greater proportion of cohort mothers had infants at the index date compared to the comparisons, this may have influenced the maternity hospital admissions. --- Health conditions There were also higher levels of emergency admissions in the cohort than comparison group for both parents , with reasons for these admissions shown in Figure 3 and Supplementary Table 3. The most common conditions in the cohort also had the largest variation compared to the comparison group, which included mental and behavioural disorders , external causes of morbidity and mortality , and injury, poisoning and other consequences of external causes . Significant differences were found between groups for both mothers and fathers except for neoplasms, eye diseases, ear diseases, congenital conditions, and genitourinary system diseases for fathers. The conditions with large relative differences between the study groups in primary care diagnosis records were mental disorders, 'injury and poisoning', causes of injury and poisoning and causes of morbidity and mortality . These were in common with emergency hospital admission conditions. The conditions with the largest relative differences in emergency hospital admissions and GP records are assessed in more detail in Supplementary Tables 3,4. --- Mental health and substance use conditions Mental disorders were by far the most common recorded health condition for individuals in the cohort , over 2.5 and 3 times higher than for comparison mothers and fathers respectively . The most common type of mental health condition was depression. The relative differences of severe mental illnesses were 11 times higher for mothers in the cohort group and 7 times higher for fathers. Conditions such as developmental disorders, attention deficit hyperactivity disorders, eating disorders and autism were all also considerably more prevalent in the cohort group. Substance use was recorded for around one in five parents. Parents in the cohort were around 14 and 10 times more likely to have drug and alcohol related substance use conditions recorded respectively. --- Injuries The cohort group had increased levels of accident and emergency attendances for overall injury and all injury subcategories . Accident-related injuries were 1.5 times more likely in the cohort. Cohort mothers were nearly 10 times more likely to have an assault related attendance; cohort fathers were 5 times as likely. Cohort mothers and fathers were 14 and 10 times more likely respectively to have an attendance for self-harm than the comparison group. --- Discussion --- Summary of main findings The most pronounced difference between the cohort parents and the comparison group was found in emergency type health services. Differences between the study groups were particularly pronounced with regards the use of services for mental health need, substance use and injuries/injury and poisoning. Although overall healthcare use across healthcare settings was higher for mothers in the study cohort than fathers, the differences between the cohort parents and the comparison groups were similar. Common mental health conditions were around three times more likely in cohort parents. Although overall only a small proportion of parents had severe mental illness diagnoses , the levels were far greater for cohort parents, than comparison parents . It is evident that for a proportion of parents, vulnerabilities include both mental health need and problems of substance use. The elevated level of assault or self-harm for the cohort parents is also notable, with a stark difference between parents involved in care proceedings and our comparison group. --- Study strengths and limitations To our knowledge, this is the first study to examine parental vulnerabilities of both mothers and fathers involved in care proceedings in the UK, and as such allows comparisons between parent type within the cohort, but also against a comparison group of parents matched on age and deprivation. This paper builds on previous work by the same lead authors of the Family Justice Data Partnership [23] first published as a descriptive funder report [24]. For this publication we have added further academic rigour through addition of statistical testing for all health care use outcomes. Better understanding of the needs and vulnerabilities of these parents may provide opportunities to improve a range of support and preventative intervention for these families. This work covers wide range of measures, providing a broad picture of health service use and underlying conditions, and by linking health and family justice data at population level for fathers, builds on the evidence base [25,26] for this group as well -a group often excluded from such research and policy work [27,28]. Studies based on administrative data are necessarily limited by the scope and quality of available data and are collected primarily for non-research purposes. Specific strengths and limitations of Cafcass Cymru data are reported elsewhere [15,16]. Cohort parents had more children aged under 1 year at the index date compared to comparisons, which may have influenced levels of healthcare use for mothers -for example, for pre-and post-natal appointments. The earliest application date within the study period for each parent was also used, as a proxy measure to represent the first occurrence within care proceedings. We recommend future work aims to account for any bias resulting from recurrent care proceedings [29]. The SAIL Databank contains data from around 80% of general practitioner practices in Wales; as such, data for GP-based measures was available for the majority, but not all individuals; GP measures were not adjusted for the reduced coverage, which we recognise is a study limitation. We compared cohort findings against a matched comparison group ; this study design choice was made to allow more meaningful comparisons to be drawn between study groups. Factors such as deprivation are known to adversely affect health outcomes [30,31] and as we matched on deprivation readers should be aware that results are likely to underestimate effects in comparison to the general population. As our comparison group selection method used category matching using age bands this resulted in imbalances in ages between groups as noted in Table 1; this choice was made to increase the size of the comparison group, however, further studies should consider matching on more closely aligned ages. It should be noted that match rates in the cohort were 82% , whilst this is in line with previous work [16] we have no further information to understand if there are differences between the matched and non-matched parents. It could be hypothesised that the non-matched parents are more vulnerable and as such could result in under reporting of heath care use. It would be worthwhile for agencies such as Cafcass Cymru to aim to improve data quality to improve future match rates and research design. --- Comparison of research findings with previous literature Uniquely, this study examined both mothers and fathers interaction with healthcare services prior to court proceedings. However, there is an important body of related international research on the mental health needs and co-morbid substance use, for parents involved with child welfare services [32][33][34][35][36][37], and children in care [38][39][40]. Although the research we report is specific to parents who are involved in formal family court proceedings, our findings are consistent with the broader published research in reporting elevated rates of mental health need often co-occurring with substance use. Notable in the published literature, is the work of Wall-Wiehler and colleagues in Canada who reported elevated rates of mental health diagnoses, treatment use and social factors for mothers, both 2 years prior to and 2 years after children were taken into care. In adding to the extant knowledge, the findings we report draw fathers clearly into view, a group whose needs are often marginal to discussions about the family justice system [41]. By differentiating health service utilisation, we have also uncovered the higher use of accident and emergency health services among parents with problems of mental health and substance use in this particular population [42,43]. A key finding in the international literature is that parents with problems of mental health and substance use are more likely to require emergency healthcare on account of accidents, injury, or self-harm, or because they have not sought help with health conditions at a timely point from primary care providers. However, this is the first-time emergency health care use has been evidenced for mothers and fathers in care proceedings. Looking ahead it will be important to understand causal factors implicated in elevated use of highcost emergency health care. Drawing on the broader literature, a range of explanations have been proffered, which include that same-day GP appointments can be difficult to obtain [44], that there are significant waiting lists for mental health and drug and alcohol services, and that these gaps in provision result in parents' turning to emergency healthcare [45]. The same can be said, where community-based crisis services are unavailable [46]. Further research to probe reasons behind high rates of emergency care use are important, given problems of access to health care have been exacerbated by the COVID-19 pandemic [47]. --- Recommendations for policy and practice The findings presented highlight the elevated health needs of both mothers and fathers prior to involvement in care proceedings in Wales. Higher levels of mental health needs, substance use and injury related conditions, compared to a comparison group are particularly noteworthy. The study suggests considerable healthcare costs for parents involved in public law care proceedings, however, this would require further substantiation through separate analysis of health and social care utilisation over a longer period. High use of emergency healthcare services strongly suggests the potential failure in provision of-or access to-support services at an earlier point to prevent or manage crisis. Elevated rates of self-harm are very concerning, for example. Given pressures on emergency healthcare provision, the evidence is that emergency departments are unable to offer treatment over and above attending to immediate physical healthcare needs [48]. However, this work indicates that proactively connecting parents with relevant support services, such as for mental health, is an important factor for those providing emergency healthcare services, which may help reduce demand in the longer term. This point is not new, and there is substantial literature that calls for better management of patient journeys through healthcare services, and far greater integration of health and social care provision . This conclusion, which calls for improved and more tailored mental health care provision [49] is particularly relevant for parents in care proceedings, where services need to be attuned to parents histories of adversity and trauma [50]. --- Further work International literature suggests such vulnerable populations experience higher rates of repeat emergency hospital use; a hypothesis that warrants testing through further research, as a particular service response is required in relation to frequent users of emergency services [51,52]. Further work is required to provide more detailed findings to understand how healthcare use varies depending on a multitude of factors including: protected characteristics and heritage; household-based factors ; and factors related to family court . In the context of established awareness of the relationship between inequality and health need [30] it is critical that parents in care proceedings are not simply treated as a simple homogenous group. For example, future research should consider the intersectionality of characteristics such as gender or race with healthcare needs [53,54]. In this work we concentrated on the period preceding care proceedings. Significant life events, such as having a child removed, can lead to immediate psychosocial crisis prompting a deterioration in health conditions, especially mental healthrelated issues including suicidal ideation, along with worsened socioeconomic conditions [55]. It is therefore important to also consider further work to understand health conditions, and patterns of healthcare use over the lifetime of involvement in care proceedings and beyond. This may indicate periods of highest health service demand and highlight when services are most required to support parents and families. Such work should also consider other significant life events, such as incarceration. Linkage of datasets from across the justice system via the Data First programme [56] will provide future ability to investigate levels of incarceration for this population using SAIL. Finally, the potential multiple and long-term effects of such experiences for the children and young people involved in the family justice system should be examined. This may further enforce the need for increased advocacy services within health and social care to support vulnerable children and families as laid out in the Well-being of Future Generations Act , and the Welsh Government Programme for Government [57]. --- Conclusion Both mothers and fathers in care proceedings in Wales experienced greater levels of health vulnerabilities during the two-year period prior to court proceedings compared to a comparison group of parents matched on deprivation, sex, and age. The higher use of emergency healthcare is particularly noteworthy and indicates considerable crisis health need among parents. Elevated mental health, substance use, and injury-related conditions are coupled with higher use of emergency services. Better understanding of the needs and vulnerabilities of this population, including the reasons why parents are making greater use of emergency healthcare may provide opportunities to improve a range of support and preventative interventions that respond to crises in the community. --- Statement on conflicts of interest None to declare. --- Funding Nuffield Family Justice Observatory aims to support the best possible decisions for children by improving the use of data and research evidence in the family justice system in England and Wales. Covering both public and private law, Nuffield FJO provides accessible analysis and research for professionals working in the family courts. Nuffield FJO was established by the Nuffield Foundation, an independent charitable trust with a mission to advance social well-being. The Foundation funds research that informs social policy, primarily in education, welfare, and justice. It also funds student programmes for young people to develop skills and confidence in quantitative and scientific methods. The Nuffield Foundation is the founder and co-funder of the Ada Lovelace Institute and the Nuffield Council on Bioethics. Nuffield FJO has funded this project , but the views expressed are those of the authors and not necessarily those of Nuffield FJO or the Foundation. --- --- Ethics statement The project proposal was reviewed by an independent Information Governance Review Panel at Swansea University. This panel ensures that work complies with information governance principles and represents an appropriate use of data in the public interest. The IGRP includes representatives of professional and regulatory bodies, data providers and the general public. Approval for the project was granted by the IGRP under SAIL project 0990. Cafcass Cymru also approved use of the data for this project. The agency considered the public interest value of the study, benefits to the agency itself, as well as general standards for safe use of administrative data. --- Abbreviations

Under section 31 of the Children Act 1989, public law care proceedings can be issued if there is concern a child is subject to, or at risk of significant harm, which can lead to removal of a child from parents. Appropriate and effective health and social support are required to potentially prevent some of the need for these proceedings. More comprehensive evidence of the health needs and vulnerabilities of parents will enable enhanced response from family courts and integrated other services.To examine health vulnerabilities of parents involved in care proceedings in the two-year period prior to involvement.Family court data provided by Cafcass Cymru were linked to population-based health records held within the Secure Anonymised Information Linkage Databank. Linked data were available for 8,821 parents of children involved in care proceedings between 2011 and 2019. Findings were benchmarked with reference to a comparison group of parents matched on sex, age, and deprivation (n = 32,006), not subject to care proceedings. Demographic characteristics, overall health service use, and health profiles of parents were examined. Descriptive and statistical tests of independence were used.Nearly half of cohort parents (47.6%) resided in the most deprived quintile. They had higher levels of healthcare use compared to the comparison group across multiple healthcare settings, with the most pronounced differences for emergency department attendances (59.3% vs 37.0%). Health conditions with the largest variation between groups were related to mental health (43.6% vs 16.0%), substance use (19.4% vs 1.6%) and injuries (41.5% vs 23.6%).This study highlights the heightened socioeconomic and health vulnerabilities of parents who experience care proceedings concerning a child. Better understanding of the needs and vulnerabilities of this population may provide opportunities to improve a range of support and preventative interventions that respond to crises in the community.

Background Physical activity improves quality of life and reduces the risk of non-communicable disease [1,2]. WHO guidelines recommend 150-300 minutes of moderate to vigorous intensity of physical activity per week to reduce the risk of NCDs as well as improve health and well-being [3]. Globally, 1 in 4 adults do not meet the global recommended levels of physical activity [4]. In Sub-Saharan Africa , where there is a high burden of human immunodeficiency virus [5], physical activity may improve the quality of life of people living with HIV [6][7][8]. However, physical activity appears to be poor among PLWH with few studies reporting that PLWH spend ~ 75% of their time sedentary [9,10]. Contributors to physical inactivity among PLWH include chronic inflammation, side effects of antiretroviral therapy , poor nutrition, and other HIV-related co-morbidities [9,11]. Besides biological drivers of reduced physical activity, studies have reported a number of behavioural and sociocultural attributes which contribute to low level of physical activity among PLWH. These include lack of time, knowledge, motivation, concerns about physical appearance, tiredness from work, lack of financial resources to pay for physical activity facilities and lack of support from the family and community members [12][13][14]. The majority of studies assessing underlying factors contributing to low physical activity among PLWH adults have been conducted in older populations of high-income countries [15]. Studies included evaluations of participation and motivation in individuals or community-based physical activity programs [16][17][18]. Studies focusing on perceptions and facilitators have highlighted that PLWH perceive physical activity to be beneficial for their health but also find it challenging to be active [19]. In addition, these studies have reported that the facilitators and barriers of physical activity were related to psychological, physical, and social-environmental dimensions [19]. However, studies conducted in high-income countries cannot be extrapolated in SSA due to differences in societies, cultures, and health systems. Studies on HIV, social cultural issues, and physical activity in SSA are scarce [20], and this has been stipulated as a gap in a recent systematic review on physical activity and HIV in SSA [21]. Many theoretical and behavioural models have been used to understand and conceptualise factors influencing initiations, practices, and maintenance of physical activity [22]. Previous studies have used various theoretical models of behavioural change including the health belief model, trans-theoretical model and the social-cognitive model [23][24][25]. However, critics of these health-behavioural models have argued that changes focusing on individual's lifestyle or behaviour including physical activity needs more than psychological-behaviour approach, since individuals make decisions and choices within a wider social-ecological context [26]. The theoretical social ecological model shows the interplay of all levels including; interpersonal, intrapersonal, organizational and policy, illustrating how factors at one level influence factors at another level in behaviour [27]. The strength of the SEM model is that it provides a comprehensive understanding of the determinants of health behaviours and mechanism of change for each level [27,28]. Previous studies have shown that cultural-sensitive behavioural theories such as the SEM model are useful in attempting to understand the factors which influence physical activity behaviour in vulnerable populations [15,29]. In addition, other studies have used SEM to create and maintain interventions which facilitate communities to actively engage in physical activity [27][28][29]. Using qualitative study, we aimed to explore the perceptions, facilitators and barriers of physical activity among PLWH to better understand multifactorial influences of physical activity at individual, social and family, community and environmental levels. --- Method --- Study setting and design This was a qualitative study conducted in Mwanza region in the North-Western Tanzania from August-November 2019 among PLWH. The study was nested within the cohort study, the "Chronic Infection, Co-morbidities and Diabetes in Africa " registered at http:// clinicaltrials.gov as NCT03106480 [30]. Mwanza has a population of ~ 2.7 million and HIV prevalence of ~ 9.2% [31]. The common ethnic groups found in the region are Sukuma, Haya, Jita and Kerewe and the main incomegenerating activities in this region are petty trading, fishing, livestock keeping, farming and mining [31]. This study deployed the SEM adopted from McElroy and colleagues [27] to understand the underlying perceptions, facilitators and barriers of physical activity in PLWH. The original SEM incorporates intrapersonal , interpersonal , community, organizational and public policy levels. For this study, we did not assess organizational and public policy level as it was out of our scope. The main focus was to understand factors influencing physical activity in the every day's lives of PLWH. We analysed community level which also included environmental conditions together to describe the overlapping of influences on physical activity including existing physical activity promotional programs and the availability of public institutions which support or provide access to physical activity [32]. --- Recruitment of participants The participants were recruited from the CICADA cohort study, purposively selected to provide rich information on perceptions, opinions, social-cultural values and practices of physical activity in the community and at the individual level. The purposive selection process considered representation of sex, age group, occupation, marital status, living area and antiretroviral therapy status [33,34]. The CICADA cohort study recruited PLWH from HIV clinics located in the public health facilities in the Nyamagana and Ilemela districts. Potential study participants were telephoned and invited by the first author to come at the National Institute for Medical Research clinic for more information about the study. Those interested were given a detailed written information sheet to read and an informed consent form to sign before being enrolled in the study. A total of fifty participants were invited, forty-three participants agreed to participate in the study while seven participants did not want to take part due to time constraints or being relocated from Mwanza. --- Data collection, management and coding procedures Data collection, management and coding were conducted in an iterative process informed by relevant guidelines to report qualitative research [35]. For credibility, we used two qualitative data collection methods. Sixteen in-depth interviews were conducted to capture in-depth personal distinct, awareness, opinions and experiences of physical activity, whereas three focus-group discussions were used to capture shared common opinions, values and experiences within the communities on perceptions, facilitators and barriers of physical activity [36,37]. We used different data collection methods to address specific dimensions of each objective. The final number of interviews and focus groups was based on when data saturation was reached [38]. The topic guides for the interviews and focus groups were developed in English and consisted of open-ended questions on perceptions, facilitators and barriers influencing physical activity [19,28,29,32]. The topic guides were translated into Kiswahili and included questions about participants' perceptions of the terms used to express "physical activity" and "exercise" . These terms describe different concepts in Kiswahili. By definition, physical activity is defined as any bodily movement produced by skeletal muscles that result in energy expenditure in daily living and can be categorised from four domains: occupation, recreational sports, domestic, and transportation [39]. Exercise is defined as a subset of physical activity that is planned, structured, and repetitive with an objective of either improving or maintaining physical fitness [39]. However, in practice, the words are often used interchangeably [3], and this has previously created some challenges and misunderstandings in the interpretation of physical activity and exercise when used in other languages, cultures or contexts [4]. In our study, we used the two terms interchangeably throughout in interviews and focus groups. Questions about existing supportive or challenging circumstances and environment for physical activity practices, awareness, motivations, practices of physical activity, and whether these had changed after being diagnosed with HIV were included. The guides were pilot tested in three interviews to assess language, concept and content understanding of the questions and were then revised accordingly before they were used in the study. BK and ES conducted all the interviews and focus groups in Kiswahili. BK and ES are female researchers at the National Institute for Medical Research with 10 years of experience and training in conducting qualitative studies. BK and ES worked together interchangeably as interviewer and note taker during the data collection process. The interviews and focus groups were conducted at the National Institute for Medical Research clinic where the participants attended their routine visits for the CICADA study. We ensured a conducive environment for interviews and focus groups to allow freedom of expression. Each interview and focus group began with informal conversations with the intent of building rapport with participants. Throughout the interviews and discussions, we also conducted member checking by recapping of questions/responses with the respondents and this allowed validation of the responses. We also validated their responses during the data analysis and interpretation of results by making follow up with the participants through phone calls as recommended by Birt et al. [40] for trustworthiness. For focus groups, the first group was for men, second group was for women and third group included both. The two focus groups were conducted by sex category to avoid gender and power relations between men and women when responding to the questions. Interviews lasted about 30 min while focus groups lasted at least 45 min. Interviews and focus groups were audio recorded using Sony tape recorder and fully transcribed and translated to English by an accredited research assistant independent of the researchers who collected the data. Data were coded manually and analysed according to the principles of deductive content analysis [41]. Specifically, content analysis allowed us to condense and broaden the description of the physical activity phenomenon in PLWH, and the outcome of the analysis was described either in concepts or categories of factors influencing physical activity among PLWH [41]. First, the transcripts were thoroughly read. Based on the readings, the empirical material was categorised in the following major themes: 1) perceptions of physical activity, 2) facilitators of physical activity and 3) barriers of physical activity. Secondly, following this initial categorisation of major themes, we identified sub-themes and sub-categories within each major theme which was then coded as belonging to the individual level, the social and family level, or the community and environmental level as described in the SEM . The coding of perceptions, facilitators and barriers of physical activity at the individual and social and family level mainly used data from interviews, while the coding of themes at the community level used data from focus groups. In addition, participants' quotes were used to illustrate the themes of perceptions, facilitators and barriers of physical activity at the individual, social and family, and community and environmental level. Data coding and interpretation was conducted by BK and ES separately and then compared. If there were any disagreements on the data coding and interpretation of themes, the two researchers would go through the transcripts and discuss the differences until consensus was reached. --- Results --- Participant's characteristics A total of forty-three participants above 18 years and on ART services for ≥1 year, participated in the sixteen interviews and three focus groups with nine participants per focus group . --- Results overview We found differing interpretations of physical activity and exercise which were based on specific physical activity practices rooted within the existing gender stereotypes and gender norms in the communities. Despite of the variation in the interpretation of physical activity, most PLWH perceived physical activity as beneficial for their health and were motivated to do it; if their family members and friends provided physical and emotional support that included doing physical activity together. Barriers of physical activity mentioned were time constraints, lack of social groups, lack of access to financial resources and facilities to be able to do physical activity. HIV-infection was not mentioned as a main barrier of physical activity but rather as a health concern from family members and friends thinking that physical activity could deteriorate their health. PLWH also reported that they hardly heard any public health messages regarding physical activity from health care professionals in HIV clinics and do not find any public health physical activity promotional messages displayed in the health facilities or community. --- Perceptions of physical activity --- Individual level Awareness and interpretation of the term 'physical activity' Participants revealed differing interpretation of the terms "physical activity" and "exercise". The term "physical activity" was translated to which was perceived as an abstract concept; while the term "exercise" was perceived as a more commonly used term. Participants would generally describe physical activity practices with examples using the word "exercise". Participants' clearly identified physical activity with certain practices. Household chores and income-generating activities involving walking long distances were described as "physical activity", In contrast "exercise" was perceived as recreational activities with the intention of improving physical strength or endurance. As reported by a participant: Experiences and practices of physical activity Participants also perceived that income-generating activities such as petty trading; walking long distances selling fruits were sufficient physical activity for them. The commonly mentioned physical activity practices were walking, running, rope skipping and playing football which were reported to be easy to perform and suitable for many age groups. For example, as reported by a participant: " "For me, the work I do requires a lot of physical activity and I also used to walk from town to Igoma as part of my exercise" [Interview-41 year old man]. Some men perceived sexual intercourse was part of physical activity. This was reported in individual interviews as well as in participants' focus groups on community perceptions of physical activity. For example, as explained by a participant: "Physical sexual intercourse with a woman is an exercise" [Interview-47 year old man]. Perceived benefits of physical activity PLWH perceived physical activity was important to their health, they believed by doing physical activity they will be able to improve their physical strength and health in general but also improve their personal appearance. As mentioned in an interview: "Physical exercise helps to keep you alive and keep the body in a good condition" [Interview-39 year old woman]. --- Social and family level Family members and friends' opinions and experiences of physical activity PLWH reported that family members and friends had low level of awareness, and poor physical activity practices. Despite this, some PLWH reported getting encouragement or company from friends who like exercise like running or walking. However, some respondents were unable to respond to this question because they lived alone and do not have many people around them. As reported by participant: "I don't have groups or friends I like to live on my own.,I like listening to music if I have nothing to do" [Interview-61 year old man]. --- Community and environmental level Perceptions of physical activity related to gender stereotypes and norms Apart from the 'individual' and 'social and family' perceptions of physical activity, shared sociocultural values in the community can also contribute to individual choices and behaviour of physical activity. In the focus groups, participants expressed perceptions of physical activity practices which were rooted within gender stereotypes and roles or norms in the community. The focus groups revealed that men perceived themselves and were seen by women as being more active than them. Out-door recreational physical activities were reported as more appropriate for men while household chores were identified as physical activity for women. As mentioned in a discussion: " the society that surrounds me, men do more exercise like running though it's just few of them. --- Women are only doing physical activity like household chores and take it as exercise" [Focus groups-34 year old woman]. Women reported that household chores were sufficient physical activity for them, and that they did not need to do any more. As reported by PLWH in one of the focus group discussions: " --- Facilitators of physical activity Individual level Motivation and reasons to do physical activity Participants felt motivated to do physical activity because it restores their physical abilities and improves their health. Self-efficacy and self-esteem would also improve due to the ability to appear healthy and not be recognised as someone with HIV, as they reported: --- "What inspires me is, I want my body to be healthy and fit. Even if I am sitting among the people living with HIV, I cannot be recognised . I do it for my health, I want to stay well" [Interview-32 year old woman]. --- Social and family level Emotional and physiological social support and interaction Family members and friends' encouragements and participation in physical activity practices were reported as one of the main motivations for PLWH to do physical activity. Thus, the encouragement and participation of family members and friends are important facilitators of physical activity at the social level. As reported by a participant in an interview: "We play with the children in my area, like rope jumping. Sometimes we play netball. I even sometimes play football. I play with the youth there in our street" [Interview-32 year old woman]. --- Community and environmental level Availability of facilities The presence and access to football pitches and open spaces in the community for physical activity was mentioned as one of the facilitators of physical activity. As reported by a participant: "In the community there is no gym, but we have football pitches and many people use it to run" [Focus groups-45 year old men]. --- Barriers of physical activity Individual level Time management Participants commented that lack of time was one of the main barriers of physical activity. Participants did not have enough time to do physical activity because of work-related activities. As a participant explained in an interview: --- "I wish to walk more … I don't have enough time … sometimes I find that I have a lot of work and there is no time to go walking" [Interview-23 year old woman]. Health concerns related to HIV One participant feared physical activity would worsen their HIV disease leading to more deterioration of their heath rather than development of physical strength. --- "After diagnosis, I thought maybe doing hard work will make me weak, I thought like I will die and leave my children, but after some time I was okay" [Interview-45 year old woman]. Lack of financial resources Lack of resources or money to pay for access to fitness centres was also reported as a barrier of physical activity. As highlighted by a participant: --- "I won't agree if they tell me to pay ten thousand while I earn five hundred thou-sands. It will be difficult … my income has to be in line with the gym costs" [Interview-61 year old man]. --- Social and family level Fear of disease to worsen from family members Participants also reported that family members perceived physical activity practices to potentially worsen their HIV disease. Family members sometimes discouraged them from doing physical activity. This was mentioned only by few participants. As reported by one participant: --- "My mother advises me not . She fears my condition" [Interview-30 year old woman]. Lack of encouragement from family members Family members discouraged physical activity due to lack of their time to participate in these physical activities. Immediate family members did not discuss or advise on any issues pertaining physical activity. As reported by the participant: --- "Neither my husband nor my children tell me about doing any exercises" [Interview-36 year old woman]. --- Community and environmental level Lack of social programs to support physical activity PLWH reported that lack of social groups participating in physical activity in their communities was a barrier to physical activity. In addition, both public health offices and HIV clinics failed to provide health messages to promote physical activity and that includes displays in the community on physical activity. Little support and advice on physical activity from health care professionals According to participants, health care professionals placed little emphasis on physical activity to PLWH at the health care centres. Hence, this was mentioned as a barrier since PLWH believed physical activity messages from health care professionals on physical activity would be like a reminder for them to do physical activity. The health care professionals mainly emphasise on maintaining a healthy diet, good hygiene and ART treatment adherence. As reported by a participant in a group discussion: "There is no place or groups for doing exercises. --- They never talk about exercise for sure" [Focus group-43 year old man]. --- Discussion This study sought to explore perceptions, facilitators and barriers to physical activity among PLWH using the social ecological model. Our main findings of this study were as follows; physical activity was described and interpreted with language and cultural perspective that distinguishes physical activity from exercise and in practice physical activity was interpreted based on gender stereotypes and roles such as that household chores are for women while recreational exercise are for men. PLWH reported family members and friends' advice and support motivates them to do physical activity apart from physical activity improving their health and wellbeing. Barriers of physical activity as reported by PLWH were lack of time, infrastructure and social groups, lack of financial resources for physical activity, and poor information sharing on physical activity from health professionals in the HIV clinics and community. Although HIV infection was not seen as the main barrier of physical activity among PLWH, family members were reluctant to encourage physical activity to their relatives because of fear that the HIV disease would worsen. PLWH perceived and referred to physical activity as household chores activities, income-generating activities and other day-to-day activities as a sufficient level of physical activity while exercise was regarded as recreational or geared towards building physical strength. However, the majority of PLWH perceived that physical activity improves their physical strength, self-efficacy and health in general. Such perceptions on type, amount and practices of physical activity informs us about PLWH understanding of physical activity which may contribute to low levels of physical activity. Lack of awareness and misperceptions on the guidelines pertaining to physical activity inhibits physical activity [42,43]. Physical activity programs raising awareness of physical activity in PLWH and their communities may lead to improvement of physical activity practices in this population and this has been shown in a study using health belief model that supervised exercise could change individual's perceptions and knowledge of physical activity and the risks of NCDs [44] among the few studies in PLWH. Research suggests that existing gender stereotypes and gender roles influence physical activity participation and choices across cultures and communities [45]. The present study also found that PLWH reported gender differences in their perceptions of physical activity. Certain physical activities were perceived as suitable for either men or women such as football for men and household chores for women. Also, it was perceived that men do more physical activity than women. Women perceived that they were doing less physical activity than men although they might have done quite a bit of physical activity through household chores. Women also expressed that household chores and income-generating activities were enough physical activity but they could also be enabled to engage in more leisure physical activity like men. Similar findings were reported in other studies on social construction of gender stereotypes and physical activity. The studies highlighted certain physical activities were perceived predominantly for boys and "less cool" for girls. Boys were perceived as more active while girls spend more time socialising [45][46][47]. However, other study have shown that participation in physical activity at an earlier age contributes to a more flexible attitude later on toward gender norms in physical activity [48]. Further, a systematic review conducted in SSA reported that the traditional gender stereotypes of women not engaging in out-door activities influenced the level of physical activity participation because out-door physical activity was perceived as "unfeminine" [49]. Our findings highlight the need for contextual gender-based interventions appropriate for PLWH addressing positively the existing gender stereotypes and roles in relation to physical activity, providing opportunity for women to be able to do sufficient physical activity. In this study, we also observed that physiological social support and interactions from family members and friends were facilitators of physical activity in PLWH. Friends and family members play a major role in enhancing physical activity. This social support in physical activity has been widely studied and has been noted to be a strong facilitator of physical activity among individuals. In other context, studies have gone further by indicating clearly the directions and differences of social support from friends and family members that could influence the motivations and maintenance of physical activity practices [50][51][52]. Even though these studies have been conducted elsewhere and not in SSA, their results and our results show some potential of transferability theoretically to understand how social support from family members and friends may enhance physical activity practices. PLWH reported benefits resulting from doing physical activity such as increased physical strength, selfefficacy and improved health which motivated them to initiate and maintain physical activity [1,53]. However, other studies have argued that the concerns of PLWH for health improvements and improved physical appearance are among the fundamental coping strategies to reduce HIV related stigma [28]. Overall, our findings provide evidence for the value of social support to PLWH to increase physical activity in their day to day living. Interventions directly emphasizing on improving social support to PLWH from family members and friends will aid in promoting physical activity in this vulnerable population. Some of the main barriers to physical activity reported among PLWH were time availability, lack of facilities, infrastructure and social groups, poor support from family members, high costs for recreational activities at fitness centres, and health concerns due to HIV-infection. Similar results were reported in other studies conducted in high-income countries and most were conducted in women with HIV and older populations and have used SEM and a comparison group of HIV-uninfected individuals to explore barriers of physical activity [54,55]. These studies described biological factors resulted from the HIV infection such as physical exertion, opportunistic infections, presence of bodily pain, and depression as barriers to physical activity. Social and cultural factors including family discouragements, time constraints and unfamiliarity with physical activity facilities and machines in fitness centres have also been reported as significant barriers to physical activity [13,[55][56][57]. On the contrary, cultural issues relating to religion, cultural identity, body image and physical appearance were not identified as barriers to physical activity in our study, unlike in other studies [54,57]. Nevertheless, there some other studies which reported illness stereotypes where people with chronic diseases share concerns about physical activity as a risk for their health [58].. Thus, this was also reported in our study by family members and PLWH. Findings observed in this study on barriers, may inform development of appropriate interventions for PLWH to improve their physical activity practices in this context and reduce long-term effects of non-communicable diseases and few studies on health promotion interventions in other age populations have highlighted promising results when it comes to improvement of physical activity [59]. Lack of support and advice on physical activity from health care services was also found to be among the barriers in the promotion of physical activity. PLWH reported lack of education on physical activity at the HIV clinics from health care professionals. The Tanzanian government's management and care of HIV & AIDS guideline has clearly stipulated that in PLWH who are stable on ART, physical activity should be recommended to patients [60]. On the other hand, the recommendations for physical activity available in this guidelines are not detailed and are merely mentioned as a sub-topic in relation to diet [60]. Therefore, in practice, it can easily be overlooked or not stressed enough by health care professionals. The Tanzanian guideline has been criticised due its lack of alignment with the local concerns for ART service delivery [61] and for physical activity, lack of detailed information on physical activity recommendations [62]. The findings of our study support the importance of incorporating physical activity promotional messages in the management of HIV. --- The strengths and limitations of the study The strengths of this study includes the use of SEM to explore in-depth the concepts and context of physical activity including social and cultural factors that influence physical activity among PLWH in their day to day living and their communities [53,63]. By understanding these factors, public health specialists and policy makers would be able to design appropriate contextual based interventions which will promote physical activity among PLWH. In addition, to our knowledge this is the first study to explore social and cultural factors that influence physical activity among PLWH in Tanzania, and one of few studies in SSA. Our study employed qualitative research guideline known as COREQ [64] during its implementation, data collection and reporting of the results. Throughout, the study assessed in detail the components of rigour and trustworthiness . The study used two different data collection methods, interviews and focus groups to address all the dimensions of the specific objective and the findings could be transferable to other settings, hence, this weights the strength of this study. In addition, the study was implemented by experienced qualitative researchers strictly following appropriate guidelines in all the procedures . One of the limitations of this study is that it has been conducted in only one city of Tanzania and thus, the results may not be transferrable to all settings or communities in Tanzania at large. In addition, some degree of interviewer bias and participant selection bias was unavoidable as the participants who agreed to be interviewed, were involved because they had time and interest to participate. We have chosen to structure our findings at the levels of the SEM model, but we acknowledge that several of the facilitators and barriers of physical activity are cross-cutting and at times do not present at a specific level in the everyday lives of PLWH. Finally, our sampling framework was limited to PLWH since our objective was to investigate the factors that influence PLWH and their physical activity practices. For future studies, we propose health workers and other members from non-government officials who work with PLWH to be involved in these studies to enrich additional information on perceptions and physical activity among PLWH, at the community and for policy implications level. --- Conclusions We found there are differing interpretations of physical activity and exercise and that physical activity practices are rooted within existing gender stereotypes and roles in the communities. Physical activity related to household chores was for women and exercise or outdoor activities were for men. PLWH felt motivated to do physical activity because of the encouragements from family members and friends and because of perceived health benefits which included physical strength, improved health, and not looking like a person living with HIV. However, reported barriers to physical activity included lack of time, limited awareness, health concerns related to HIVinfection, lack of support from family and friends, lack of resources to pay for access to fitness facilities, lack of social groups for physical activity and lack of facilities. Further, existing gaps in implementing the management and care of PLWH guidelines in relation to physical activity was also observed as a barrier to physical activity as reported by PLWH. HIV-infection and other healthrelated outcomes were not mentioned as the main barriers by the PLWH but rather as a health concern from family members and friends for PLWH if they do physical activity. The main barriers were social-cultural issues which were the mirror of the facilitators of physical activity. The use of SEM model has allowed to us to identify and understand different factors at different levels that influence physical activity among PLWH. --- Implication for policy and practice This study was undertaken partly to inform future interventions or strategies to understand the individual, socio-cultural issues related to physical activity practices among PLWH and their communities. Development of new strategies to promote physical activity must address issues related to awareness and emphasise on commonly known practices of physical activity in relation to WHO recommendations. Further, involvement of family and friends to provide support emotionally and physiologically to PLWH on physical activity would help to improve physical activity among PLWH. Interventions are needed to address the elements of gender stereotypes and roles, attitudes and beliefs related to physical activity that exist in the community and specifically to promote physical activity in women. In addition, these results will also support development of new physical activity public health promotional messages, and to be displayed in healthcare facilities, public institutions and for health care professionals to share adequate information on physical activity to PLWH. Building of supportive environments including facilities or infrastructures for physical activity is needed in the communities to promote physical activity in PLWH. --- --- --- --- Consent for publication N/A. --- Competing interests The authors declare no conflict of interest. • 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 People living with HIV (PLWH) have low levels of physical activity. Using the social ecological model to understand perceptions, facilitators and barriers of physical activity in this population is of importance for developing contextualised interventions to improve physical activity in PLWH.This was a qualitative sub-study conducted between august and November 2019 as part of a cohort study on diabetes and associated complications in HIV infected in Mwanza, Tanzania. Sixteen in-depth interviews and three focus groups with nine participants in each were conducted. The interviews and focus groups were audio recorded, transcribed and translated into English. The social ecological model was considered during the coding and interpretation of the results. Transcripts were discussed, coded and analyzed using deductive content analysis.Forty-three PLWH aged 23-61 years participated in this study. The findings showed that most PLWH perceived physical activity as beneficial to their health. However, their perceptions of physical activity were rooted within existing gender stereotypes and roles in the community. Running and playing football were perceived as activities for men while household chores activities were for women. Further, men were perceived to do more physical activity than women. For women, household chores and income-generating activities were perceived as sufficient physical activity. Social support and engagement of family members and friends in physical activity were reported as facilitators of physical activity. Reported barriers of physical activity were lack of time, money, availability of physical activity facilities and social support groups, and poor information on physical activity from health care providers in HIV clinics. Human-immunodeficiency virus (HIV) HIV infection was not perceived by PLWH as a barrier for doing physical activity but most family members did not support them to do physical activity, fearing that it might worsen their condition.The findings demonstrated differing perceptions, facilitators and barriers of physical activity among PLWH. Interventions addressing awareness, gender stereotypes and roles related to physical activity from individual to community level are needed. Supportive environment and infrastructures are needed to improve physical activity levels in PLWH in Tanzania.

Introduction: The World Health Organization declared COVID-19 as a pandemic in March 2020, and this was followed by a series of preventive measures that included social distancing, travel restrictions and lockdowns in India. Pregnancy is a vulnerable time, with several physical and psychological changes associated with it. The added burden of pandemic could lead to significant stress, and it would be helpful to understand the mitigating factors of stress in this population. Objectives: The objective of this study was to examine the mitigating factors associated with COVID-19 pandemic among pregnant women. Methods: The study was conducted in an Urban Primary Health Center in Bangalore that provides maternity care to lowmiddle income population. Antenatal check-ups are conducted here daily. Pregnant women visiting the clinic for routine antenatal care were approached and informed consent was sought for an interview. 295 women consented to participate in the study. The qualitative interview was conducted in a quiet room. Open ended questions were used to understand the participants' personal, familial, occupational, and social factors related to COVID-19. The transcripts of the interviews were manually coded for recurring themes by two research assistants. These were examined and similar or identical themes were grouped together. These were further analyzed, and themes were summarized. Results: The mean age of the participants was 24.9 years, approximately half of the participants were in their first trimester and primiparous. The majority were high school educated and selfemployed. 25% of the participants reported mild to moderate depressive symptoms assessed using PHQ-9. The most predominant theme among personal factors was negative emotions that included fear and anxiety. The uncertainty about the transmission and the lack of clarity about the causes during the pregnancy were the reasons for these fears. They reported that they found news and media more stressful. They reported that even though the lockdown restricted their movement, they enjoyed the time they spent with the family, especially their husbands. Most participants reported job loss, theirs and their husbands' and had to encounter severe economic difficulties. However, the predominant theme was the social support that they received from family, friends, neighbors, and local governing bodies. Conclusions: Social support was reported to be the most predominant factor that helped the pregnant women to cope with the problems presented during the pandemic. Social support, both instrumental and emotional were important mitigating factors for stress during the pandemic. Strengthening the social support system by support groups and community networking should be a crucial component in government-led initiatives as a factor that may promote resilience in difficult situations like the pandemic. --- Disclosure of Interest: None Declared --- EPP0981 Mental health challenges in health care workers during COVID pandemic S. Bilichodu Rangappa1,1 * and S. Avula2 Introduction: Mental health issues of the health care workers are often overlooked. It's often presumed that situations like COVID are handled well by this group of population and hence their own mental wellbeing is ignored and neglected. Objectives: The objective of this study was to evaluate the depression and anxiety levels in healthcare workers who were on COVID duty. Methods: This study was performed over telephonic interview of all the healthcare workers who were performing COVID duty from March 2021 to December 2021. Hospital Anxiety and Depression Scale was administered. Various factors influencing the presentation were then analyzed. Results: Over 534 healthcare workers were screened for psychiatric symptoms. About 76 fulfilled HADS criteria. 7.86% scored above the anxiety cut off point and 6.36% scored above the depression cut off point. About 373 expressed concerns about their safety and security as they were staying away from their families and about 469 expressed concerns about uncertainties about duty patterns. Conclusions: Health care workers should be screened for psychiatric illness if they are in constant stress. They should be well trained to carry out COVID duties which will reduce the anxiety about the duty patterns. Better awareness about COVID 19 may lead to decreased levels of anxiety and depression. --- Disclosure of Interest: None Declared

fewer non-locals and to less frequent non-severe presentations. Despite the type of patients (for underlying diagnoses) remained unmodified, an interesting reduction of anxiety symptoms and suicidal behavior was noticed. Literature from ED studies during the first wave are consistent with our finding regarding the number of visits; suicide attempts seemed unmodified or decreased elsewhere (Giner et al. Curr Psychiatry Rep 2022;24(1):1-10). Limitations of our study include peculiarities of the Venetian territory, limited sample and time of observation. Future directions encompass the integration with data from the community setting and later developments.

Introduction Being concerned about one's own health serves an important adaptive functionit increases survival. Dealing with symptoms of illness in a timely manner is beneficial. However, some individuals become overly health anxious and develop worries that range from mild to extreme concerns . This is called health anxiety . Severe HA has been found to increase health care use and the risk of long-term sick leave and is a persistent condition if left untreated . HA is often associated with other anxiety disorders, such as panic disorder and generalized anxiety disorder . Despite growing evidence of the consequences of HA, we know little about the distribution of HA in the general population. One reason is that most research on HA has been performed with patient populations . --- Measurement of HA The three most frequently used self-report measures to assess HA, all originally developed for screening purposes for the diagnosis of hypochondriasis, are the Short Health Anxiety Inventory , the Illness Attitude Scale and the Whiteley Index . The WI is a self-rated instrument developed in 1967 as a diagnostic tool for hypochondriasis but later began to be used to screen for HA. The use of diverse self-report measures makes comparisons between screening studies difficult. --- Distribution of HA There is no consensus on the appropriate cut-off points to define HA with different versions of WI, although most studies have chosen to set a cut-off point. Hedman et al. have recommended a cut-off point of 5 with a WI with 14 questions and dichotomous response options. However, no other studies have reported this threshold in population studies. The lack of criteria to define HA through the different self-report measures is a challenge within this field of research. This affects the reliability and validity of assessments of HA prevalence, which has been reported to be between 2.7% and 13%. A study based on the IAS estimated a point prevalence of 6% in a German population , but the representativeness of the sample was dubious because most participants in the sample had low education levels. Martin and Jacobi , however, found an overall 12-month prevalence of 2.75%, using diagnostic interviews for screening purposes. In New Zealand, a convenience sample of elderly individuals over 65 years found a prevalence of 7% with the SHAI . Sunderland et al. observed a point prevalence of 3.4% and a 12-month prevalence of 4.2% based on an Australian national survey. However, their assessment was restricted by the use of only one single question: 'Have you ever worried a lot about serious illness despite reassurance from a doctor?' Furthermore, two articles from Canada and the USA reported a prevalence of illness worry of 6% and 13%, respectively. These authors assessed illness worry based on the same question but asked the respondents to report based on different time periods: 'In the past 12 months, have you had a period of 6 months or 1 month or more when you worried about having a serious physical illness most of the time?' In line with Ferguson , we assume that HA is more accurately represented as a dimensional rather than a categorical construct. To the best of our knowledge, there have been no studies exploring the distribution of HA in a general population, despite the belief that HA is experienced on a continuum ranging from mild concerns to severe anxiety . The Tromsø study: Tromsø 7, a health survey that included all adult inhabitants in the municipality of Tromsø, Norway, has given us a rare opportunity to describe the different degrees of HA in a general population. The aim of this article is to study the distribution of a self-report HA measure in a general adult population and the association of HA with sociodemographic variables. In the available data set, HA was measured with the six-item WI . --- Method --- Study design and population This study used cross-sectional, self-reported data from the Tromsø study: Tromsø 7. The Tromsø study is a large Norwegian population-based health survey that includes the collection of self-reported data, interviews, physical examinations and the collection of biological material. Seven surveys have been conducted since 1974 with different birth cohorts. The Tromsø study is described in detail elsewhere . The current Tromsø study, Tromsø 7, was conducted in 2015-2016. All inhabitants in the municipality of Tromsø aged 40 years or older were invited, for a total of 32 591 men and women. A questionnaire was included in the invitation, which the participants brought with them in a filled-out form to the clinical examination. At the examination, participants provided additional self-reported information on a wide range of topics, including sociodemographic and health-related information. The invited participants received one reminder if they did not attend their examination. By the end of 2016, 21 083 participants had taken part in Tromsø 7, representing an attendance rate of 65%. A total of 67% of the invited women and 62.5% of the invited men participated in the study. However, the participation rates were lower among participants 80 years or older, with an attendance rate of 34% for women and 48% for men. --- Variables --- Dependent variable We measured HA using the one-factor, WI-6. The WI originally consisted of 14 questions, each to be answered with a false/true response. Today, versions of the WI are available with 14, 13, 11, 10, 8, 7 and 6 items with both true/false and 5-point Likert scale response options . The 5-point Likert scale version is considered beneficial for use in a general population as it is easier for respondents to use and captures the continuity of a phenomenon . Table 1 provides an overview of the WI-6 questions. All respondents answered each question with one of the following response options: 'not at all', 'to some extent', 'moderately', 'to a considerable extent' or 'to a great extent'. Item scores were transformed into values from 0 to 4 , and the item scores were summed to a total score ranging from 0 to 24, where Y = 0 represented no HA and Y = 24 represented the highest possible measurement of HA. In the questionnaire, the introduction was omitted, which limited our knowledge of which time frame the participants used as a reference. --- Demographic and social variables Since the existing literature has reported inconsistent findings on the background demographic characteristics associated with HA, we chose to include both demographic and social variables. Age on 31 December 2015 was reported and was included both as a continuous variable and in 10-year age groups. As only 16 participants were over 90 years of age, the two oldest age groups were merged into '80 years or older'. Education was reported as 'the highest level of education you have completed', with four categories: primary education up to 10 years of schooling, vocational/ upper secondary education , college/university or college/university . The wording of the household income question was as follows: 'What was the household's total taxable income last year?', with eight categories ranging from 'less than 150 000 Norwegian kroner ' to 'more than 1 million NOK' . The household income categories were merged into four categories: low , lower middle , upper middle or high . The national average value was set as the reference value in the regression analysis for both the education and household income variables. Participants were asked two questions concerning their family life: 'Do you live with a spouse/partner?' and 'How many children do you have?' Participants reported their biological children, adopted children, stepchildren and foster children. We merged these alternatives into one dichotomized variable named 'Children', where 0 indicated 'no' responses to all of the alternatives and 1 indicated a 'yes' response to one or more of the alternatives. We also included two questions about the quality of friendship: 'Do you have enough friends who can give you help and support when you need it?' and 'Do you have enough friends you can talk confidentially with?' These two variables were highly correlated and were merged into one variable named 'Close friends'. This new variable included three categories: 'No', for those who answered 'no' to both original questions; 'To some extent', for those who answered 'yes' to only one original question; and 'Yes', for those who answered 'yes' to both original questions. Finally, the participants rated their participation in the organized activity with the following options: 'never or just a few times a year', '1-2 times a month', 'approximately once a week' or 'more than once a week'. --- Statistical analysis All analyses were performed with STATA version 15.1 . Participants with missing values for the dependent variable were excluded prior to the analyses. In the descriptive analyses, the means were calculated for continuous variables, and the frequency distributions were calculated for the categorical variables. Due to the non-normal distribution of the dependent variable , we considered both a negative binomial distribution and a decreasing exponential distribution to model any association between HA and the relevant covariables . Exponential regression gave the best fit when tested with the Akaike information criterion and Bayes information criterion . We, therefore, used a multivariate exponential regression model in both the bivariate and multivariate analyses: Y = a • e - i = 1,….,n; n = number of covariables The covariables age and gender were included in all analyses. Only statistically significant covariables were included in the final multivariate model. The level of statistical significance was set at 0.05. We also conducted survey weighting analyses to adjust for non-response, which gave identical results as the unweighted analyses. The presented results are from the unweighted analyses. The dependent variable in the estimated model is presented with the exponential beta [exp], where exp describes the percentage difference in the WI-6 score relative to the reference category for the different covariables with all else held constant. We found a significant interaction between age and household income and performed an additional analysis with age-stratified groups: we categorized those under 67 as 'workingage participants' and those 67 years or older as 'retirement-age participants'. --- Ethics The study was approved by the Regional Committee for Medical and Health Research Ethics . All participants gave written informed consent before admission. --- Results --- Participant characteristics and the distribution of HA in the study population A total of 21 083 persons between 40 and 99 years old participated in this study; 52.5% were women, and the mean age was 56 years. A total of 817 participants had one or more missing items on the WI-6 and were excluded from the analysis. The distribution of HA was highly skewed with exponential distribution . The mean and median scores of the WI-6 were 3.3 and 2 points, respectively. In total, 75% of the participants had a total score of 5 points or less, 5% had a score above 10 points and 1% had a total score of more than 14 points. The study population's demographic and social characteristics, including the mean WI-6 scores in the participant subgroups, are listed in Table 2. The WI-6 scores ranged from 0 to 20-24 in all subgroups categorized according to the demographic and social variables. --- Associations between HA and sociodemographic factors In the bivariate exponential regression analyses, all variables except gender were significantly associated with HA . In the multivariate analysis, the variables concerning family life were non-significant and were excluded from the final model. Table 3 shows the final model including the demographic and statistically significant social factors. Respondents who did not have friends to get support from and talk confidentially with had significantly higher HA than those who did have such friends . Answering 'no' to the 'Close friends' questions was associated with a 61% higher WI-6 score while answering 'to some extent' was associated with a 41% higher WI-6 score compared to that of the reference category . Similarly, regarding the frequency of participation in organized activity, participating in an organized activity once a week or more was associated with significantly lower HA than the other categories. Household income showed a negative association with HA. Gender was not significantly associated with HA. We found age to be negatively associated with WI-6 scores. In addition, there was a significant interaction between age and household income. A stratified analysis for age was therefore performed, with the participants divided into working-age participants and retirement-age participants, as shown in Table 4. The mean HA value was similar in both groups. The stratified analysis showed that HA decreased with increasing age among working-age participants but showed no association with age among retirement-age participants. The quality of friendship remained strongly negatively associated with HA in both strata. Participation in organized activity was significantly negatively associated with HA among retirement-age participants who reported participating weekly, but this relationship was nonsignificant among working-age participants. In addition, the associations of HA with both education and household income were non-significant in the retirement-age participant group. --- Discussion --- Distribution of HA In the present study, HA in the population is explored as a continuous rather than a dichotomous characteristic. We found an exponential distribution of HA on the WI-6 scale ranging from 0 to 24. There was no evident cut-off point to distinguish participants with severe HA based on their WI-6 scores, indicating the importance of analyzing HA as a complex, continuous construct. To the best of our knowledge, this is the first study to report HA as a continuous phenomenon in a general population. Although the mean HA level was low, there was great variation within all participant subpopulations. This finding illustrates the continuity of HA. We found no difference in the level or presence of HA related to gender, which is in concordance with other epidemiological studies on HA . Although women are reported to have a higher prevalence of other anxiety disorders than men , this does not appear to be the case with HA. In light of our results, we think it is relevant to also explore the consequences of HA-related health care use, disability and comorbidity for slight and moderate as well as severe HA. Because we used a representative sample rather than a convenience sample, our results show a valid distribution of HA in a general adult population. In addition, by using a validated measurement tool rather than one question, we were able to better capture the complex nature of HA. Finally, most of the research on the prevalence of HA was published before 2010 . Therefore, our research is important to provide updated knowledge of the occurrence of HA in a general population. --- Psychological Medicine --- The quality of friendship and participation in organized activity To the best of our knowledge, this is the first report showing that the quality of friendship is highly associated with HA. Our results show that 17% of the population indicated that they had no or little perceived support and confidentiality from friends. Interestingly, our model estimated that compared to those who answered yes to both 'Close friends' questions, participants who did not report such high quality of friendship had 42-61% higher HA. As there is some overlap between the different anxiety disorders , it is relevant to examine the association of friendship with other anxiety disorders. One prospective study found that the quality and quantity of social networks were non-significant for the development of anxiety disorders, whereas perceived loneliness was a significant predictor of anxiety . The questions in our study related to the quality of friendship might be interpreted similarly to questions about perceived loneliness. In our study, reduced perceived quality of friendship was the single most important factor associated with the level of HA. Perceived loneliness may also play a part in the negative association between HA and participation in organized activity. In our study, we did not differentiate participation in physical or other social activities, as both types of activities are thought to be overall beneficial for health . Participation in organized activity may increase survival , and it is associated with higher life satisfaction and lower anxiety and depression . In a review paper on anxiety disorders, exercise was found to be a protective factor for the development of anxiety disorders . This might be in concordance with our cross-sectional results, which showed that participation in an organized activity once a week or more was significantly associated with low levels of HA. --- Interaction between age and household income on HA In the stratified analyses, we found that while household income was highly associated with the level of HA in the working-age group, this factor was not significantly associated with HA for the retirement-age group. This finding may suggest that factors other than household income are associated with HA in older participants. The difference in significant associations may also indicate differences between the two groups. Most of the participants aged 67 years or older were retired, which would have reduced the differences in their income, as most retired participants would have a percentage-wise reduction in income. We also identified descriptive differences between the two groups related to education, with higher education being more common among younger participants. However, in the retirement-age group, we also found a large variation in maximum HA, ranging from 14 to 22 points out of 24 points. Interestingly, we found age to be negatively associated with HA up to the age of 67, indicating a decline in HA in the retirement-age groups. This association is in accordance with the results of Sunderland et al. on a population aged 15-85 years. They found that the prevalence of HA was lowest in the youngest and oldest participants, with a peak in middle age. A similar trend was observed in our study among participants 40 years and older. Interestingly, the negative association of age with HA was non-significant among retirement-age participants. The findings of Boston and Merrick , who found no association between HA and age in a population over 65 years, are also in accordance with our results. We cannot conclude whether an age or cohort effect can explain this finding. If it is an age effect, we would expect similar findings in future research. If our HA research illustrates a cohort effect, we can expect an increasing number of people with HA in the years to come. --- Strength and limitations A strength of our study is the large study sample and high participation rate compared to other population-based studies . In addition, we chose to use the recommended 5-point Likert scale, which has been found to have better psychometric properties than dichotomous options . Previous studies using different versions of the WI have mostly used dichotomous response options. The WI is recommended for use as a screening tool for HA and is beneficial for use in population health surveys because of its relatively limited number of questions compared to the SHAI or the IAS . We hope that the use of a recommended measurement tool for a general population will motivate other studies to include this relatively short measurement of HA in future health surveys. More studies are needed to further explore the usefulness of this procedure. Despite the representative sample in our study, a Norwegian study found that people with mental illness were less likely than those without mental illness to participate in population surveys . However, the Tromsø study is marketed as a health check in addition to a health survey. In contrast to other anxiety disorders, HA is characterized by the seeking of reassurance for bodily stress and fears . We therefore do not believe that HA-related avoidance has affected participation in the Tromsø study. In the questionnaire, the introduction was omitted. Unfortunately, this limited our knowledge of which time frame the participants used as a reference. There is a lack of knowledge of the variation in low and moderate HA over time. However, severe HA is persistently present, with little variation over time periods . When the participants gave self-reported information regarding HA, they answered the seventh question: 'Do you have recurring thoughts about having a disease that is difficult to be rid of?' We decided to not include this question in our study in line with the recommended use of the WI-6 . This seventh question was not validated for use in a general population. All analyses with both versions of the questionnaire gave identical results. Therefore, we do not believe that excluding the seventh question influenced our results. --- Conclusion In conclusion, our study demonstrates the continuity that is believed to characterize HA. Future studies should explore the impacts of this continuum. The findings indicate that social factors of friendship and participation in an organized activity may be decisive for HA levels. --- Conflict of interest. None declared.

Background. Health anxiety (HA) is associated with increased risk of disability, increased health care utilization and reduced quality of life. However, there is no consensus on which factors are important for the level of HA. The aim of this study was to explore the distribution of HA in a general adult population and to investigate whether demographic and social factors were associated with HA. Methods. This study used cross-sectional data from the seventh Tromsø study. A total of 18 064 participants aged 40 years or older were included in the analysis. The six-item Whiteley Index (WI-6) with a 5-point Likert scale was used to measure HA. Sociodemographic factors included age, sex, education, household income, quality of friendship and participation in an organized activity. Results. HA showed an exponential distribution among the participants with a median score of 2 points out of 24 points. In total, 75% had a total score of 5 points or less, whereas 1% had a score >14 points. Education, household income, quality of friendship and participation in organized activity were significantly associated with HA. The variable quality of friendship demonstrated the strongest association with HA. Conclusion. Our study showed an exponential distribution of HA in a general adult population. There was no evident cut-off point to distinguish participants with severe HA based on their WI-6 score, indicating the importance of analysing HA as a complex, continuous construct. HA demonstrated strong associations with quality of friendship and participation in an organized activity.

2 farming could be readily explained if early farming were more productive than foraging, estimates suggest that this may not have been the case 2 and alternative hypotheses based on environmental, social, and demographic parameters have been proposed 13,14 . It has also been suggested that the transition from foraging to farming results in people working harder, having less leisure time, and being less productive per hour worked 2 . Based on data from contemporary hunter-gatherer societies, Sahlins 1 argued that hunter-gatherers represent the "original affluent society" who, despite a lack of material wealth, have a livelihood that allows them to work only 2-4 hours per day. Although this claim challenged the assumption that the foraging-to-farming transition represented an escape from an arduous foraging lifestyle, subsequent studies have found that there is substantial variation among foraging and farming populations in how much they work [3][4][5][6]15 , that many hunter-gatherers face substantial annual fluctuations in food availability 6 , and that many foraged foods require a substantial amount of time to process once brought back to camp 16 . Given this diversity, comparisons between populations are limited in their ability to isolate the effect of adopting agriculture on time allocation. Here, we examine variation in time budgets within a single population -the Agta, a community of small-scale politically egalitarian hunter-gatherers from the northern Philippines who are increasingly engaged in agriculture and other non-foraging work 9,[17][18][19] . The Agta live in small camps of fluid membership, within which individuals cooperate extensively in foraging and food sharing 20 and where ~50% of adults are distantly related or unrelated by kinship 21 . We conducted quantitative ethnographic fieldwork with the Agta in 2013 and 2014, collecting data on the time allocation of 359 people across ten camps . Time allocation data were collected through observational scans. We conducted four scans each day during daylight hours, with the first scan between 06:30 and 09:00 in the morning and three more at three-hour intervals. In each scan, we recorded the activity of all members of the community, grouping activities into four main categories: childcare, domestic chores, leisure, and out-of-camp work . This resulted in a total of 10,706 personobservations. Out-of-camp work was divided into two categories: foraging and non-foraging work. Foraging work consisted of fishing, gathering, honey collecting and hunting. Although the majority of out-of-camp nonforaging work consisted of agricultural labour, this category also included activities such as the collecting of non-food items to sell . Leisure time included socialising, resting, playing, and sleeping. Of adult leisure time , 71.9% was spent in close proximity to at least one other adult. Of this time, adults were in close proximity to an average of 2.20 other adults . There was no sex difference in the mean number of these social interactions and 49% of interactions between adults were with individuals unrelated through either genetic or affinal kinship. By comparing across Agta camps that vary in their relative engagement in foraging versus non-foraging out-ofcamp work, we are able to explore the association between changing livelihoods and time allocation. We show that across Agta camps, increased engagement in non-foraging out-of-camp work is associated with increased total out-of-camp work and reduced leisure time, and that there is a significant sex difference, with women significantly increasing their out-of-camp work as camps move away from foraging. Across all camps, adults spent an average of 29.2% of daylight time engaged in out-ofcamp work , 24.0% engaged in domestic chores, 12.2% engaged in direct childcare, and the remaining 34.7% of time at leisure. However, there were significant sex differences in time allocation . First, although adult men spent significantly more time engaged in out-of-camp work and significantly less time engaged in domestic tasks and childcare than adult women , female out-of-camp mean = 17.3% , P <0.001; male domestic chores mean = 20.2% , female domestic chores mean = 27.7% , P <0.001; male childcare mean = 4.86% , female childcare mean = 19.5% , P < 0.001; two-tailed permutations tests; Across the ten study camps, there was significant variation in engagement in non-foraging out-of-camp work, with non-foraging as a proportion of all out-of-camp work varying from 0% to 80%. Across the ten study camps, we found that greater involvement in non-foraging out-of-camp work as a proportion of all out-of-camp work was negatively associated with leisure time , we used Bayesian multilevel multinomial modelling 22 to predict adult leisure and work time across camps while controlling for the individual-level fixed effects of age, sex, and whether an individual had a child under the age of 2 years. This method also allowed us to take into account the multinomial nature of time-allocation data 23 . Confirming the previous linear regression results, the models suggested that for women but not for men, there was a negative association across camps between engagement in non-foraging and predicted rest time . The results of this model provide a good fit to the data, confirm the age and sex related changes described above, and suggest little relationship between time of day and engagement in out-of-camp work . Our results provide evidence from within a single population of hunter-gatherers that greater engagement in farming and other non-foraging work is associated with increased out-of-camp work time and decreased leisure time. Although we cannot necessarily equate leisure time with affluence 6,24 , given that previous studies have shown that more sedentary and agricultural Agta camps have worse health and increased child mortality 9,25 , the claim that the transition away from foraging among the Agta is associated with a deteriorated standard of living is broadly supported. Although no activity was exclusively the domain of one sex, we found a general sexual division of labour among the Agta, with men doing more out-of-camp work than women and with women doing more childcare and domestic chores. In line with previous findings 6,20,26 , this appears to be driven by the time constraints imposed on mothers by caring for young children. We also found pronounced age differences in time allocation, with adult leisure time increasing with age, and with out-of-camp work and childcare both peaking at ~30 years. The negative relationship between leisure time and engagement in non-foraging out-of-camp work is driven largely by women, who spent much more time engaged in out-of-camp work and less time at leisure in those camps more heavily engaged in agriculture and other non-foraging work. Why is this the case? In part, it may be that agricultural work requires a greater total labour investment. Although this would be consistent with previous economic analyses of small-scale farming 2 , it does not explain why the additional burden falls disproportionately on women. One possibility is that male and female agricultural work is more substitutable than foraging work, as may be the case where productive hunting or fishing requires many years of experience 27 . Alternatively, cultural norms relating to the sexual division of labour may apply differently to nonforaging work. Finally, men may be unable to spend additional time engaged in out-of-camp work without cutting into a minimum amount of required rest/leisure time. This would be consistent with analyses of time budgets in non-human primates 28 , and with our finding that men and women had a similar amount of leisure time despite differing substantially in the amount of time devoted to other activities. Although the differences observed in the relative engagement of Agta communities in foraging and farming provide a useful natural experiment for exploring the effect of economy on time allocation, extrapolation from our results to foraging-to-farming transitions in pre-history should be made with caution and particular attention should be given to the factors that may alter the relative productivity of foraging and farming among the Agta. For example, while the wet rice agriculture practiced in Palanan remains labour intensive and non-mechanised, it is likely to be much more economically productive than early farming 2 . Also, the Agta are able to increase the returns on foraging by trading with their non-Agta neighbours; they trade or sell approximately a quarter of all foraged food for rice and other agricultural products. This kind of proteinfor-carbohydrate exchange is common between contemporary foragers and their farming neighbours 6,29 and may also reduce reliance on wild carbohydrates 30 . Finally, the livelihood of the Agta is influenced not only by their interactions with non-foraging neighbours but also by national policies relating to the status of indigenous people, land rights, and the environment 18 . Comparisons with farming aside, the amount of leisure time available to the Agta and other hunter-gatherers is testament to the success of the human foraging niche, made possible by our ability to share, process, and cook food, to make and use sophisticated tools, and to accumulate foraging skills and knowledge both within individual lifetimes and across generations 31,32 . These traits may themselves be promoted by having the leisure time to interact and exchange cultural knowledge with large numbers of people 32,33 . --- Methods --- Ethnographic context The Agta communities included in this study live in the coastal municipality of Palanan, which sits within the boundaries of the Northern Sierra Madre Natural Park, northern Luzon, Philippines. Within Palanan, as in the wider region, the Agta are a minority, accounting for ~5% of the population 18 . The Agta are politically egalitarian small-scale hunter-gatherers who cooperate extensively in childcare, foraging and food sharing 20,34 and have a flexible system of residence, with households moving frequently between camps. Many Agta families have long-standing trading relationships with non-Agta farmers and regularly trade foraged foods for rice and other agricultural goods as well as cigarettes, tools, alcohol, and household items. Across the four Agta camps for which data were available, the proportion of all foraged foods by weight that were sold or traded with non-Agta was 27.9% . Our time allocation categories were modified from those developed for the Agta in a previous study 18 and are given in Supplementary Table 1. In some observations, individuals were engaged in more than one activity concurrently . In these cases, in order to preserve the multinomial structure of the data, we randomly selected one of the concurrent activities. Our analyses focus on differences in time spent at leisure and at work between camps. This allows us to overcome the high degree of interdependence in the time budgets of individuals and the relatively modest number of observations per individual . Among the Agta, we rarely encountered individuals who knew their own age. To estimate age, we took the mean values from posterior probability distributions of age produced using a Gibbs sampling MCMC algorithm based on age ranking order data provided by the Agta and a plausible a priori age range for each individual provided by the ethnographers 36 . --- Statistics All analyses were conducted in R.3.5.1. We used an alpha level of .05 for all statistical tests. Variables used in the regression analyses met the assumption of normality. To evaluate the statistical significance of pairwise differences in time allocation between age and sex categories, we compared the observed differences between categories with expected distributions generated by resampling from the original data 10,000 times. To explore the influence of engagement in non-foraging activities on leisure time, we fitted Bayesian multilevel multinomial models in the brms package 22 . These models capture the multinomial nature of time budget data whilst allowing for the investigation or control of both fixed and random effects. We fitted separate multinomial models for adult males and females where the response variable considered five activities: childcare, domestic chores, foraging work, non-foraging work, and leisure. 'Domestic chores' was set as the reference category, so that each model predicts the log-odds ratio of the proportion of time engaged in each activity relative to the proportion of time engaged in domestic chores. In each model, we included a random effect of individual to control for non-independence of data collected repeatedly from the same individuals, and fixed effects of age, age 2 , age 3 , in addition to a term for the time of day and its quadratic. Importantly, we also included a categorical variable for camp, so that, having controlled for variation in age and time of day, we could then compare the predicted time budgets of individuals across camps. Each model was fitted with three chains of 3000 iterations, of which 600 were used for the warm-up. Population-level effects were scaled before model fitting. We chose normal priors for all population-level effects . Model diagnostics highlighted adequate mixing of chains, and the correspondence between posterior predictive distributions and the observed data was high . --- Code Availability The code used to analyse the relevant data is provided as Supplementary Software. --- Data Availability The individual-level data that support the findings of this study are available from the corresponding author upon reasonable request. Any further work on the data depends on community approval. --- camp = 22.1%, based on the returns of 114, 49, 31, and 60 foraging trips respectively). This figure was greater for meat and fish, of which 50.7% by weight was traded or sold . On average, the Agta received ~1.4kg of rice for 1kg of meat and ~1kg of rice for 1kg of fish. Assuming that meat, fish, and rice contain ~200, 100 and 350kcal per 100g respectively trading of meat for rice and fish for rice yields a ~2.5 fold and ~3.5 fold increase in calories, respectively. These benefits, however, will be somewhat diminished by the travel costs associated with trade. The ten Agta communities included in this study have a mean size of 35.7 people . Based on data derived from genealogical interviews we estimate that mean within group relatedness was r = 0.12 and mean within-group shared reproductive interest, a measure that captures relatedness through marriage 35 was s = 0.15 . --- Northeastern Luzon has a tropical climate with high humidity and heavy rains concentrated roughly between August and November and with several typhoons or tropical storms during this period each year. December and January are the coolest months and April-June are the hottest. Our time allocation data were collected between March and August 2014. During this time the weather was generally dry and hot, with no significant storms. At this time of the year, the Agta favour fishing over hunting because rivers are often calm and clear. The main agricultural activity in the region is wet-rice agriculture, a labour-intensive method of farming which can produce up to three crops each year. Most farms required labour throughout our March-August study period, harvesting at least one rice crop and planting another. Although few Agta own such rice fields themselves, they are involved in planting, tending and harvesting rice on farms owned by non-Agta. The observed differences in the engagement in agricultural work between Agta camps are largely a consequence of their proximity to farms and interest in engaging in this work, rather than temporal differences in opportunities for labour. --- Data Collection All data were collected in 2013 and 2014 as part of ethnographic fieldwork approved by the UCL Ethics Committee . We explained our methods and data anonymity through presentations and posters in the local language, and received informed consent for all participants. Data collection and --- Competing interests The authors declare no competing interests.

A long-standing hypothesis suggests that the transition from hunting-and-gathering to agriculture results in people working harder, spending more time engaged in subsistence activities and having less leisure time 1,2 . Tests of this hypothesis are, however, obscured by comparing between populations that vary in ecology and social organisation as well as subsistence [3][4][5][6] . Here, we test this hypothesis by examining adult time allocation among the Agta, a population of small-scale hunter-gatherers from the northern Philippines who are increasingly engaged in agriculture and other non-foraging work. We find that individuals in camps engaging more in non-foraging work spend more time involved in out-of-camp work and have substantially less leisure time. This difference is largely driven by changes in the time allocation of women, who spend substantially more time engaged in out-of-camp work in more agricultural camps. Our results support the hypothesis that hunting-and-gathering allows a significant amount of leisure time and that this is lost as communities adopt small-scale agriculture. Agriculture emerged independently in multiple locations worldwide from around 12,500 years BP and by 5,000 years BP had replaced hunting-and-gathering as the dominant mode of human subsistence 7,8 . The transition from foraging to farming is associated with population growth, sedentism, and the emergence of increasingly hierarchical political structures 6,7 . For individuals, the adoption of farming has been associated with an increase in fertility 9,10 and a decline in dietary breadth and overall health 11,12 . Although the transition from foraging to

INTRODUCTION Racism is part of the fabric of life for African Americans and is among the causes of enduring negative health outcomes. There is really nothing new or startling in the assertion that social circumstances encountered as part of day-to-day experience influence physical health. At the turn of the last century, W. E. B. Du Bois and Kelly Miller --- CONCEPTUALIZING RACISM Nearly forty years ago, James Jones in Prejudice and Racism challenged definitions of racism that minimized the important issue of power differentials by focusing primarily on attitudes, distorted ideologies, and unfair treatment. Jones argued that racism involves assertions of power that are expressed on individual, institutional, or cultural levels. Recent conceptual advances have underscored the significant limitations of cognitive and behavioral views of racism and have included critical discussions of its systemic foundations. Two particularly comprehensive conceptualizations of racism were developed by Paradies and Harrell . They described racism as a system where power is unevenly distributed along racial lines resulting in the oppression and exclusion of non-White groups. These conceptual advances are critical to a comprehensive understanding of the breadth and persistence of racism and to the success of ameliorative policies and therapeutic interventions. Just as multidimensional and complex forms of racism can be incorporated into the individual's cognitive structures, they are woven into the fabric of the society. Indeed, racism emanates from individual, cultural, institutional, and structural domains and has psychological and physiological effects. Thus, according to Paradies , racism transcends the incorrect or irrational behavior of individuals because "non-human actors such as laws and institutions can perpetuate systemic racism" . Through values and practices, social institutions support the efforts of dominant groups to maintain power and privilege . Bonilla-Silva's structural approach to racism provided a critical challenge to the assumption that racism is a "free-floating" ideology. He proposed that racism is a central organizing principle that governs the arrangement of the society. Structural racism encompasses social mechanisms that have made the privileges associated with being White and the disadvantages associated with being a person of color relatively permanent characteristics of Western societies. It fostered the "color-blind" policy racism that permeated public policy in the last two decades of the twentieth century and constitutes the foundation upon which all other forms of racism are built. Essentially, structural racism energizes and drives the ways in which racism is sustained via interpersonal, cultural, and institutional actions . --- MEDIATORS OF PHYSIOLOGICAL ACTIVATION AND ALTERATIONS --- Racism as Stressful In considering racism's impact on health, the most straightforward case is that racism makes the lives of the disadvantaged more stressful and thus leads to negative mental and physical health consequences . Clark et al. propose that perceiving an experience as racist initiates a series of processes that results in negative emotional reactions and physiological changes. Physiological arousal includes activation of the hypothalamic pituitary adrenal cortical system, which involves the release of the stress hormone cortisol, and also entails changes in immune and cardiovascular functioning. Cardiovascular changes and a myriad of responses including respiratory functions and pupil size come under increased control of the sympathetic branch of the autonomic nervous system during emergencies. These physiological adjustments constitute the sympathetic adrenal medullary axis of the stress response. Investigators have marshaled laboratory and survey evidence linking experiences with individual and institutional racism to physiological changes. One irony that early psychophysiological studies revealed is that striving to actively control stressors actually increases activation within the cardiac sympathetic nervous system and disrupts autonomic balance . Thus, the findings cautioned that those who recognize racism and attempt to control it may be particularly vulnerable to its negative physiological effects. Similarly, increased physiological activation is a possible mediator of poorer performance on tasks that occurs when individuals believe their efforts could potentially confirm racist stereotypes . Though striving to come to grips with racism or to disconfirm its mythology may result in increased physiological activation, passivity and acquiescence are not healthy alternatives. Rather, those who actively confront racism should be armed always with behavioral strategies, discussed below, that will allow them to control levels of physiological arousal. Though currently much of the research on racism and health is grounded in the stress model, it applies most appropriately where institutional and individual racism generate discrete and perceivable toxic situations and social encounters. However, racism encompasses more than encounters with strident, unenlightened people and oppressive, constraining institutions. Conceptualizing racism within the stress and coping paradigm leads to the conclusion that if one does not perceive circumstances as racist or appraise events as stressful, the physiological reactions to them will not occur. There is evidence that the conscious perception of threatening events is but one of several pathways leading to the disruption of physiological regulation. Limitations of the traditional stress model encourage the search for other mechanisms that mediate racism's impact. --- Prenatal Effects of Racism In a provocative article, Kuzawa and Sweet show how racism potentially affects the uterine environment and epigenetic activity, and thus contributes to negative health. The study of epigenetic processes considers an array of factors that influence how genes are expressed . A growing consensus holds that environmental and genetic events constantly interact and are best conceptualized as interdependent . Thus, the mapping of the human genome was simply one step toward understanding the complex interplay between biological and environmental events that govern the unfolding genetic potential. In addition, mounting evidence from animal and human research suggests that during the prenatal period there are epochs where health outcomes are singularly vulnerable to nutritional and psychosocial factors ). Remarkably, the model Kuzawa and Sweet developed proposed the prenatal effects of racism are potentially transgenerational. The effects of stressful environments on the fetus can potentially cause lingering metabolic alterations during adulthood, which in females will impact the uterine environment of offspring. Racism troubles the entire fabric of life in oppressed communities creating encounters with environmental toxins and poor nutrition, as well as noxious and damaging social and psychological exchanges and circumstances. The uterine environment is adapted to maintain conditions that lead to optimal fetal growth and development. However, chronic psychosocial stressors experienced by the mother during various periods of gestation can threaten the well-being of the fetus. In many respects then, racism is encountered long before the brain of its victim is developed enough to appraise any particular manifestation of it. Low birth weight would be among the more ostensible effects of these insults , but the harmful impact may be subtle and less apparent at birth. Programmed effects of racism on health, which may include altered neural development, may not become evident until much later in life. --- Racism's Unconscious Pathway Two sets of studies highlight another route through which racism influences physiological activity and health outcomes. Experiments using animals have revealed that stress responses can activate a pair of distinct neural circuits in the brain ). Normally, information from the environment moves from the senses to a structure at the base of the brain called the thalamus and from there to the cerebral cortex. In the cortex the sensory input is integrated with previous input, appraised, and evaluated. Ultimately, as the stress and coping model details, the cerebral cortex is where information is represented as the conscious emotional and cognitive events that are experienced when the organism is threatened or placed in danger. LeDoux has shown that information about stress in the environment can take an alternate neural path. The thalamus not only receives and transmits sensory input to the cortex, it also connects with a second set of neural circuits that reside within the amygdala, a neighboring structure at the base of the brain. The amygdala initiates a cascade of neural events that activate lower brain centers and cause stress responses in the HPA and SAM systems. For example, the central nucleus of the amygdala activates pathways to the lateral hypothalamus and brainstem to stimulate HPA and SAM responses. Connections between the hippocampus and lateral amygdala mediate contextual information that modulates fear responses. Thus, subcortical structures provide a rapid route to physiological activation, but this activation is not dependent on conscious perception of the stimulus. Conscious perception occurs through the separate stream of neural events involving higher cortical activity, but the unconscious subcortical pathways of stress are worthy of further study. A second line of research, human classical conditioning studies, provides additional evidence that physiological systems can be activated unconsciously. Experimental studies have demonstrated that conditioned autonomic responses take place even when the conditioned stimuli are made unrecognizable by masking techniques . Additionally, clinical neuropsychological research has shown that conditioned activation occurs in patients whose severe brain damage renders them incapable of generating new episodic memories . These patients are not able to remember when an unpleasant event-for example, a shock or loud noise-was associated with a particular color. Still, if given a choice, they will subsequently avoid that particular color. Clearly, humans do not have to recognize a conditioned stimulus or consciously process its link to a negative event for the stimulus to elicit physiological arousal. Thus, conscious awareness of elements of racism is not a necessary condition for physiological activation to take place. The brain initiates early rapid physiological arousal unconsciously that prepares the individual for action, sometimes before the stressor is fully perceived. Accordingly, it is likely that repeated encounters with individual or institutional racism and its context can elicit physiological reactions of which the individual is unaware. The setting or social environs where reactions have occurred may also result in physiological activation. For example, negative interpersonal encounters need not happen on the job or in schools on a daily basis. The work site or the classroom alone might be quite capable of eliciting physiological arousal, unconsciously, because of their connection to aversive events. --- Emotion Regulation and Racism Another avenue by which racism may trigger strong physiological reactions is activated when individuals attempt to manage the anticipatory responses and subsequent reactions to socially racist stressors. This process is called emotion regulation. Emotion regulation refers to the ability to increase, maintain, or inhibit the expression and experience of an emotional response. The alteration in emotional responding occurs in milliseconds and can be voluntary or automatic . Suppression, the inhibition of one's emotional response, takes place when the emotion is denied admission into consciousness. Successful suppression of an emotion reduces the reported or observed experience of subjective experience, though the physiological arousal may persist . It was noted earlier that active coping can enhance responses to stress. Ironically, several studies report that those who deny or suppress the existence of racism experience higher systolic blood pressure than individuals who confront and actively cope with racially noxious situations . There is strong evidence that individual differences in prefrontal activation play an essential role in emotion regulation. In particular, individuals who report persistent negative affect and who show increased physiological reactivity to stressful events are more likely to have a difficult time regulating the intensity and duration of a negative emotion once it has been activated . Indeed, the emotional response to social rejection or social pain can be very intense, especially in the case of individual or institutional racism. Social pain may activate brain regions that are responsive to physical pain. Eisenberger et al. provide evidence that the prefrontal cortex, in particular the anterior cingulate cortex and right ventral prefrontal cortex, are active when social pain is experienced. These noxious events are likely to engage and often tax emotion regulatory and restorative processes. Important progress in the study of the effects of racism on health will result from studies of brain mechanisms involved in regulating reactions to intensely distressing social experiences. --- Racism, Worry, and Autonomic Balance Some social neuroscientists who study human adaptation to complex and demanding environments have made the activity of the vagus nerve their central focus . The vagus nerve is part of the parasympathetic branch of the autonomic nervous system and in many instances reduces activation while assisting in building resources in the body. For example, one division of the vagus nerve reduces the energy demands by slowing heart rate, while other branches increase digestive activities that allow us to store energy that may be needed later . Neural circuitry coursing from the frontal cortical regions connects with the vagal circuits in the brain stem and facilitates flexible cardiac adaptation to a changing environment . While increased vagal activation slows the heart, the vagus can speed heart rate by withdrawing its effects. Thayer and Friedman note that changes in vagal activity take place on a second-by-second basis, and facilitate rapid behavioral and cognitive adjustments as environmental demands require them. Further, dysregulation of vagal functions signals the presence of chronic diseases and is potentially lethal . Research suggests that the frontal lobes, which contain brain centers responsible for some of the most complex human cognitive functions, also house neural pathways that mediate the calming influences of the vagus on the heart. Failures in several higher-order "executive" cognitive processes that reside in the frontal lobes interfere with vagal functioning. For example, when well-ordered and coherent cognitive processing is replaced by perseverative thinking that includes worry and rumination, vagal control of the heart is reduced and a variety of other stress responses occur . Perseverative thinking, which can be conscious as well as unconscious, extends the time course of stress reactions . Normally, the frontal lobes are the source neural input that constantly inhibits the excitatory activity in lower brain centers that are responsible for peripheral stress responses. Thayer and Friedman argued that chronic threats and aversive events stemming from racism can ultimately interrupt the calming impact of the vagus nerve on cardiac functions by inducing worry and rumination. Therefore, vagal dysregulation resulting from disturbances in executive cognitive functions is an alternate mechanism through which several forms of racism might disrupt physiological activity. --- Racism and Cognitive Schema Cognitive structures that racism generates are the source of the final alternative path along which physiological effects may be mediated. Various forms of racism can be represented in the mind as unpleasant memories as well as cognitive schemata that deprecate non-Western ethos, history, and practices and, by extension, people of color themselves. These mental dispositions threaten psychological well-being and, in the case of cultural racism, block the positive emotional states that normally arise when one embraces one's cultural heritage and lineage . Research suggests that interrupting processes that bring about positive emotions may be particularly threatening to physiological regulation. Fredrickson has shown that pleasant emotions undo the cardiovascular activation that accompanies negative emotional states ). Research using functional magnetic resonance imaging has started to identify the neural underpinnings of racist schema, though these studies are in their infancy. In both Black and White participants, perceiving Black faces, but not verbal race labels, caused greater amygdala activation than the perceiving White faces . As noted earlier, the amygdala is sensitive to the presence of a threat; African Americans showing activation in this region in the presence of Black faces reveals signs of self-deprecation. It is not unreasonable to posit that for people of color, isolation and alienation resulting from racial schemata that disparage Black people and culture lead to increased levels of negative emotions and contribute to a general sense of defeat. The presence of these unpleasant mental states is likely to be associated with elevated blood pressure in Black populations and HPA activation in general . --- Allostasis: From Physiological Changes to Disease Ultimately, investigators who propose that the social environment influences health must explain how temporary fluctuations in physiological activity lead to more permanent alterations associated with disease. Traditionally, homeostasis has been recognized as the process by which physiological systems regulate their levels of activity around fixed optimal set points . More recently, the focus has shifted to allostasis , a concept that encompasses the capacity of the body to achieve "other" levels of activity in response to environmental changes and demands. The ability to maintain stability within systems conserves energy and thus has had an evolutionary advantage, as long as the organism did not confront extraordinary circumstances. However, through allostasis, the adaptable organism was able to make physiological adjustments when they were required. Sometimes, by the process of allodynamism appropriate alterations were achieved through changes in the underlying mechanisms controlling physiological activity. However, the adjustments in physiological levels may shade toward pathology when they are extreme, sustained, or both. Researchers use measures of allostatic load as an index of cumulative demands placed on several physiological systems that may result in enduring unhealthy physiological changes. The concepts of allostasis and allodynamism constitute a final common pathway for models that link racism to poor health outcomes. These models discuss racism's health effects in terms of environments that fail to nurture psychological and physical well-being , weathering effects on organs and physiological systems , and the embodiment of inequitable social systems ). Racism threatens health when it produces physiological changes that through any of the paths we have enumerated are extensive, occur repeatedly, or take place during certain critical developmental periods. For example, mounting evidence suggests that the nocturnal blood-pressure levels of African Americans who report encounters with discrimination fail to dip, that is, reach a normal nadir during sleep . This may constitute an allostatic adjustment that portends the onset of hypertension. Important research on allostasis and allodynamism is specifying the brain centers and peripheral physiological mechanisms that make permanent and pathological the fluctuations in physiological activity that are intended to help one cope with environmental demands . In so doing, these studies contribute important details to maps that chart the physiological paths leading from racism to poor health. --- FORMS OF RACISM AND PSYCHOPHYSIOLOGICAL CHANNELS Various manifestations of racism will show distinct affinities for one or more of the pathways through which environmental circumstances influence physiological outcomes. Table 1 posits how four forms of racism are mediated psychologically and translated into physiological outcomes. Each kind of racism generates internal representations, specified in the third column, that give rise to psychological mediators of physiological changes. The psychological mediators are presented in the fourth column. Table 1 argues that the indirect effects of structural racism are experienced primarily in terms of racialized outcomes. Often the disparities it produces are processed outside of awareness, and are represented cognitively in the form of classically conditioned associations between race and negative circumstances or by more complex racist cognitive structures. Structural racism contributes strongly to implicit associations between Black people and crime, poverty, and destitution. In contrast, institutional and individual forms of racism are more likely to be encoded as episodic memory, sometimes consciously. Here racism often takes the form of stressful events that elicit appraisal and coping processes. As the table states, in some instances racism operates through classically conditioned processes which drive many of the emotional components of responses. Figure 1 conceptualizes an approach to the study of racism and health outcomes. We placed the forms from Table 1 at the first level, with the next row specifying the conceptual model that fits the mechanisms through which each type of racism operates. Brain mechanisms critical to the effects of each manifestation of racism are then described. The next level shows two resulting psychological processes that are critical to the impact of racism, transient changes in emotional and cognitive states and relatively permanent alterations in well-being and cognitive schema. The physiological outcomes on the last level represent the changes racism causes in peripheral physiological and brain activity, as well as the early markers of disease. This illustration is a map of human suffering, with many of the paths representing conduits through which racism impacts social, psychological and biological domains. --- Interventions Evidence suggests that several interventions may reduce the physiological impact of racism. While we will describe a number of these, more extensive reviews of racism-related treatment strategies are available . Therapeutic strategies are any corrective actions that interrupt one or more of the paths between the events and processes depicted in Figure 1. We will identify the pathway of focus for each of the interventions under discussion. Cutler argued that interventions focusing on improving public health emanate from the individual, community, and national levels. His tone was decidedly hopeful, as he noted that several national programs have already achieved remarkable success in changing smoking and drunk-driving behaviors. Thus, there is no reason to shrink from developing a therapeutic agenda for mitigating the effects racism has on biological systems. Individualized approaches to intervening offer promising strategies for reducing physiological activation represented at the bottom of Figure 1. Mindfulness meditation and progressive relaxation are self-controlling techniques that influence autonomic, neuroendocrine, and immune function . It would be wise for clinical professionals to make training in these techniques readily available to populations vulnerable to racism's impact. They should be provided routinely for women during pregnancy. The procedures may mitigate physiological activation traceable to acute and chronic stressors or conditioned to contextual stimuli of which the individual may or may not be aware. Stress management procedures that include cognitive behavioral therapy also influence physiological reactions . These therapeutic techniques have the added advantage of restructuring nonproductive and self-deprecating racist cognitive schemata. Cognitive therapies also help individuals confront their emotional and behavioral responses to racism including tendencies to suppress, worry, or ruminate. Community and national strategies have the potential to launch a proactive attack on cultural racism and the racist cognitive schemata it nurtures. In the United States, African Americans, First Nations People, and Hispanic groups continue to be stereotyped or play restricted roles in film, television, and Internet sources of entertainment. Misrepresentations are a major source of cognitive schemata that associate these groups with crime, vagrancy, and incompetence. Internationally, researchers have expressed concern about the impact of such portrayals on the health of people of color and women . Properly funded by local and national agencies, indigenous writers and artists are capable of developing a rich cultural milieu that counters the negative media images plaguing people of color today. Positive emotional states accrue when racially oppressed groups are brought into contact with authentic images, activities, and cultural products rooted in their traditional culture. These positive emotional states improve one's sense of well-being and usher in healing physiological effects. Finally, the structural aspects of racism must be eliminated and only an extensive, long-term national commitment will do this. Unemployment, crime, high rates of incarceration, inadequate schools and housing, and rampant poor health outcomes stem from neglected infrastructure in communities of color. Worry and rumination in the face of persistent neglect -though predictable responses-pose physiological hazards. Structural and policy-based forms of racism affect the lives of people of color silently and deeply, perhaps because they function subtly in a color blind fashion . Erasing racial disparities in health outcomes is wedded to providing the racially oppressed with prospects for improved employment, access to quality health care with minimal discrimination, a positive sense of well-being, and knowledge of the world and skills to navigate it successfully. These salutary conditions can be achieved through a national commitment to revitalize the economic, medical, and educational structures in communities of color. --- CONCLUSION The strides the United States has made against interpersonal racism should invigorate efforts to "mitigate, resist, and undo" the effects of other more insidious manifestations of racism on health. The required interventions will range from multimodal individual psychotherapy to neighborhood and national movements that demand change in cultural institutions and social structures. Ultimately, the marker of progress against the dire forces racism has unleashed will be the reduction of the rates of virulent diseases in historically oppressed communities. Proposed models, psychological processes, and physiological pathways that link four forms of racism to health-related physiological changes

This commentary discusses advances in the conceptual understanding of racism and selected research findings in the social neurosciences. The traditional stress and coping model holds that racism constitutes a source of aversive experiences that, when perceived by the individual, eventually lead to poor health outcomes. Current evidence points to additional psychophysiological pathways linking facets of racist environments with physiological reactions that contribute to disease. The alternative pathways emphasize prenatal experiences, subcortical emotional neural circuits, conscious and preconscious emotion regulation, perseverative cognitions, and negative affective states stemming from racist cognitive schemata. Recognition of these pathways challenges change agents to use an array of cognitive and self-controlling interventions in mitigating racism's impact. Additionally, it charges policy makers to develop strategies that eliminate deep-seated structural aspects of racism in society.

This study will be guided by Sandra Bem's Gender Schema Theory . It holds that young ones acquire and assimilate knowledge gradually depending on gender types understanding. This theory believes that young ones develop cognitive schemata of gender which they get from the rules of their society. The theory holds that young ones' learn roles of female and male from the Society they grow in. Children, according to the theory, rectify their conduct to conform to the rules set by their culture since they are still very young. Bem suggested that the cognitive development of a child cumulated with influences of the environment heavily affects the way a child thinks which determines female and male characteristics. When children interact with the people around them, they observe how male and female genders behave. These interactions cognitively influence them on how males and females behave and they implement this behavior. They classify themselves as either male or female and form behavior along these categories. The sex roles that the young ones implement are formed from these categories. --- METHODOLOGY The researcher used contextual design to analyze the representation of female and male characters in three books of children by Barbara Kimenye. --- Copyright © 2022 IJMSSSR All rights reserved --- FINDINGS The research findings indicate there was unequal and imbalanced representation of gender in the books analyzed. Male characters were far more than female characters. --- GENDER REPRESENTATION Every book a child reads influences either positively or negatively the kind of thought a child gets. Books also have a huge inspiration on the child's views and outlook. The information and illustrations represented in these books may affirm good and bad news about individuals, gender, job, and future aspirations . Children's literature acts as a mirror, reflecting the happenings of a particular society. For this reason, writers of children's books ought to be careful on the kind of message they convey. In this study, the researcher showed how boys' and girls' characters in books of children are represented. A study by McCabe found that male individuals are the main characters and female characters are very few. These differences in gender representation in books of children point to the erasing of girls and women, indicating to children that those females are of little significance compared to the male counterparts. The disparity in numbers of men and boys as protagonists may make children to accept the latency of girls and women. Hamilton et al, conducted a study on children's illustration books issued in the late 90's. It was established that the male characters featured two times as female characters. They pointed out that the underrepresentation of female characters would lead to a negative impact to children's growth and this will restrict their career ambitions, mold their assertiveness towards the future as mothers and fathers and also affect their character features. Children's books ought to be realistic with the message that they pass across, so as to prevent the young ones from getting wrong and outdated knowledge of the surrounding Society . --- CONCLUSION The study analyzed gender representation in three selected children's books; Moses in a Mess, Moses on the Move and Moses and the Raffle by Barbara Kimenye. The study found out that there was significant under representation of female characters in the books analyzed. --- RECOMMENDATIONS While addressing the issue of gender representation, it is expected that the researcher will inform the Ministry of Education in recommending children's books that do not portray biasness in gender. Basing generalizations on the findings of this study, the researcher recommends that there is need to educate parents and teachers to use books that have well-rounded male and female characters. Secondly, teachers should look actively for books portraying girls and women in a positive light with active and dynamic and numerous roles. Also, the authors and publishers need to be very sensitive to avoid gender stereotyping in children's books that they write and publish. They should publish children's books that have equal number of female and male characters and illustrations as well.

The study sought to critically analyze the representation of gender in three selected children books by Barbara Kimenye; Moses in a Mess, Moses on the move and Moses and the Raffle. Children's books act as socialization agents. They transmit social customs, and replicate the morals of a community. However, the representation of gender in numerous books of childrenis imbalanced. This is most likely to negatively affect children's character and perception of character that is acceptable for their gender in the community. Objective of this study examined how female and male characters have been represented in children's books. The study analyzed children's books for ages 10 and 16. The study sought to create awareness to parents and teachers on the importance of identifying books that are not gender biased and that are not gender imbalanced. This research study is thus a useful material for reference to other researchers and readers. The study was guided by Sandra Bem's Gender Schema theory: The concept of Gender-Typed Knowledge. The study was conducted through descriptive research design. The three children's books were selected from the target population using purposive sampling technique. The researcher used contextual analysis as a method of qualitative analysis in order to analyze the selected sample. The study provided enough and viable information on improving the quality of children's books which have a positive impact in children's holistic development.

Introduction Pandemics, such as COVID-19, impact the mental health of healthcare workers and are specifically associated with increased levels of stress, burnout, and moral distress [1][2][3]. HCW treating infected patients are found to have increased fears of disease transmission to family and self-infection [4]. The COVID-19 pandemic exposed ill-equipped health systems and resource scarcity, resulting in increased rates of moral injury for HCW tasked with challenging ethical decision-making [5]. Coupled with the impact of pandemicrelated stress and increased rates of fear, anger, depression, and anxiety in the general population, HCW are at increased risk for mental health crises [6,7]. Ubiquitous disruption, continuous exposure to high-intensity stress, and a staggering death toll during the COVID-19 pandemic highlight an urgent need to prioritize the emotional and physical 2 of 14 needs of HCW [8,9]. Additionally, systems and organizational pressures seem to mostly contribute to the burnout of healthcare providers rather than provide support to staff [10]. Research into the burden of the COVID-19 pandemic on HCW is sometimes limited to single perspective/site narratives, and fails to examine the emotional, mental, and financial burden of HCW who are locked-out of the healthcare system due to policy, location, or profession, despite a desire to help. Significant research explores the experience of nurses and physicians, especially those working in critical care units [11]. A growing body of literature examines and compares the COVID-19 experiences of professional groups such as clinicians and academic professionals [12] and social workers [13], but these are again professions which were forced to adapt their workflows and client interactions, not halt them [14]. The purpose of this study was to systematically gather data over the course of the COVID-19 pandemic from HCW through an interprofessional lens to examine provider experiences from two perspectives: HCW who are locked-in the healthcare system with little control to make changes , and HCW lockedout of healthcare systems . Three research questions drive this study: How do locked-in and locked-out HCW feel prepared and supported during the COVID-19 pandemic? To what extent do locked-in and locked-out HCW report feelings of well-being, burnout, or moral distress? Finally, How do locked-in and locked-out HCW describe their experiences of the COVID-19 pandemic? With a potential impact on public health policy and practice, this study presents preliminary data from Project COPE to inform the broadest patterns of HCW experiences to direct health system practices and support of HCW. --- Materials and Methods Project COPE is an IRB-approved mixed-methods study launched in April 2020 and led by an interdisciplinary team of researchers. HCW from all specialties were recruited to participate using social media platforms, including Facebook™, Instagram™, and LinkedIn™, and other community outreach . Interested individuals were directed to a secured data collection site to share their experiences during the pandemic via an anonymous survey. Quantitative and qualitative data were captured in an initial survey and invited to participate in weekly follow-ups. No geographical restrictions were placed on recruitment. The initial survey screened participants for inclusion criteria . Demographics, profession, practice setting, years of experience, COVID-19 impacts on practice, and a validated measure of moral distress [15] were included in the initial survey. Participants were invited to submit a 5-min video blog in response to one or more of the following prompts: What types of experiences led you to feel moral stress this week? How are you feeling about the next time you go to work? or How are you coping with the pandemic while you are at home? Finally, participants were invited to opt into weekly follow-up surveys by providing an email address. Participants electing to participate in the weekly surveys received a personalized survey invitation link via email, allowing the researchers to track measures longitudinally. This survey repeated the initial survey moral distress measure and included the Well-Being Index [16] and a single-item burnout measure, the "Mini-Z" [17]. Additional questions included whether respondents saw patients that week, if any family members or friends had been infected with COVID-19, potential stressors , and respondents' coping strategies. Specifically, coping was assessed by a single item from the prompt "Which of the following have you found helpful in the past week?" Responses included a mix of 20 positive, evidence-based strategies, such as seeking support from coworkers, exercise, or recognizing meaning in their work, and maladaptive strategies, such as alcohol or tobacco use [19][20][21][22]. All survey responses were analyzed following a convergent parallel mixed-methods design. Researchers concurrently implemented the quantitative and qualitative portions of the study. The qualitative data was prioritized over the quantitative, yet kept separate until each portion of the analysis was completed [23][24][25]. While merging qualitative and quantitative data, researchers considered how each data stream contextualized the other, and what meaning might be gleaned from observing confirmation, disconfirmation, or expansion of one data source with and from the other [23]. This process is depicted in Figure 1. or friends had been infected with COVID-19, potential stressors , and respondents' coping strategies. Specifically, coping was assessed by a single item from the prompt "Which of the following have you found helpful in the past week?" Responses included a mix of 20 positive, evidence-based strategies, such as seeking support from coworkers, exercise, or recognizing meaning in their work, and maladaptive strategies, such as alcohol or tobacco use [19][20][21][22]. All survey responses were analyzed following a convergent parallel mixed-methods design. Researchers concurrently implemented the quantitative and qualitative portions of the study. The qualitative data was prioritized over the quantitative, yet kept separate until each portion of the analysis was completed [23][24][25]. While merging qualitative and quantitative data, researchers considered how each data stream contextualized the other, and what meaning might be gleaned from observing confirmation, disconfirmation, or expansion of one data source with and from the other [23]. This process is depicted in Figure 1. Qualitative analysis of 257 video blogs followed the Rapid Identification of Themes from Audio Recordings methodology where recordings are analyzed in one-minute segments [26]. Following a deductive/inductive thematic analysis, two evaluators created a detailed codebook of themes, including themes deduced from the literature, and induced from vlogs collected in Project COPE. Inductive theme development was conducted collaboratively, with evaluators viewing video segments together and comparing notes and coding in real-time. Quantitative analysis was limited to the baseline and first weekly survey collected from each respondent due to a skew in long-term participation towards complementary and integrative HCW . Moreover, to correct a significant sampling bias towards individuals identifying as white women, we collapsed race and gender into binary categories of "other races and ethnicities versus white non-Hispanic/non-Latinx" and "other gender identities versus women." Burnout, moral distress, and wellbeing scores were modeled as ordinal scales and collapsed into binary "high" and "low" categories. Four or more items on the WBI, a score of 4 or 5 on moral distress [16], and a score of 4 or 5 on the Mini-Z were considered "high" [17]. Professional categories were created to aid in analysis. These categories included allied health , medicine , mental health , and other . Due to the large numbers in each group, nurses and massage therapists were left in their own categories. Quantitative analysis included a series of logistic regression models. Model specification was informed by extant literature but was limited by overrepresentation of some professions . Burnout, moral distress, and well-being scores in Qualitative analysis of 257 video blogs followed the Rapid Identification of Themes from Audio Recordings methodology where recordings are analyzed in one-minute segments [26]. Following a deductive/inductive thematic analysis, two evaluators created a detailed codebook of themes, including themes deduced from the literature, and induced from vlogs collected in Project COPE. Inductive theme development was conducted collaboratively, with evaluators viewing video segments together and comparing notes and coding in real-time. Quantitative analysis was limited to the baseline and first weekly survey collected from each respondent due to a skew in long-term participation towards complementary and integrative HCW . Moreover, to correct a significant sampling bias towards individuals identifying as white women, we collapsed race and gender into binary categories of "other races and ethnicities versus white non-Hispanic/non-Latinx" and "other gender identities versus women". Burnout, moral distress, and well-being scores were modeled as ordinal scales and collapsed into binary "high" and "low" categories. Four or more items on the WBI, a score of 4 or 5 on moral distress [16], and a score of 4 or 5 on the Mini-Z were considered "high" [17]. Professional categories were created to aid in analysis. These categories included allied health , medicine , mental health , and other . Due to the large numbers in each group, nurses and massage therapists were left in their own categories. Quantitative analysis included a series of logistic regression models. Model specification was informed by extant literature but was limited by overrepresentation of some professions . Burnout, moral distress, and well-being scores in HCW were assessed in three separate cumulative logistic regression models, controlling for demographics and years of experience, and stratified by thematic clustering of professions. For both variables, analyses considered the "other" category as the reference group. In addition to modeling the full ordinal scales, measures were also collapsed into binary "high" and "low" categories within the logistic regression models. To further compare professionals across thematic clustering, differences in coping strategies and work concerns were explored between those seeing patients and those not seeing patients using single-tailed Fisher's exact tests. Statistical analyses were performed using Microsoft Excel™ and SAS™ software version 9.4 . Alpha was set at 0.05 for all analyses. --- Results This study included 1299 baseline participant responses from 15 April 2020 to 16 March 2021. Project COPE respondents overwhelmingly identified as women and white . Of the baseline survey respondents, 70.7% enrolled in weekly follow-up surveys and provided a valid email address. These participants submitted a total of 2411 weekly responses and 257 vlogs over the study period. Demographics on the full study sample at baseline are given in Table 1. Results are presented under four themes and six subthemes. Themes include: professional identity , intrinsic stressors , extrinsic factors , and coping strategies. Quantitative results are nested under themes to demonstrate how mixed-method results informed interpretations and theme construction. Qualitative notes on appearance, demeanor, and context are included with quotes to ascribe additional depth. --- Theme 1: Professional Identity At baseline, this sample included, among others, 592 massage therapy professionals and 273 nursing professionals. A small contingent of 81 respondents identified as dual-role professionals . Respondents almost universally framed their response and experience of the pandemic in their professional identity. Two subthemes emerged: a desire to help and a feeling of being stuck or trapped . Multiple respondents who identified as massage therapists discussed cross-training or holding additional certifications during vlog submissions, but only indicated massage therapy as their profession on intake surveys. This included one massage therapist who described her experience as a new registered nurse working on a COVID unit. She sits on the floor in front of a pair of folding doors. Her hair is wet and her face red. As she speaks, she looks away from the camera, and she begins to slouch as she speaks: I realized I am the one that is in contact every single day that I work with COVID patients, so . . . I had an epiphany that I will not be able to do massage for a really long time. Like other respondents, she reflected on the realities of holding dual professions. 3.1.1. Subtheme 1.1: Locked-Out Many participants, particularly massage therapists, described feeling cut off from their work and sought to find meaning in other ways. One respondent exemplified this experience and explained her decision to stop seeing clients, citing the risk of spreading COVID-19. The video is well lit, and she sits in front of an overflowing bookcase. Her hair is fixed with stiff curls, and her eyes are bright, but the camera shakes as she speaks: I don't know what it's going to feel like the next time I have my hands on someone. I think it's going to feel really great, but I also think it's going to feel really scary. In the baseline survey, a total of 768 respondents reported that they stopped seeing patients. Participants described this interruption in patient care as required by government action , employer , professional association , educational institution , recommended by employer or professional association . Eighty-one respondents reported being furloughed or laid off, and 362 described the decision to stop seeing patients as a personal choice. --- Subtheme 1.2: Locked-In Other HCW, particularly those working in acute care settings, described a sense of feeling trapped. In one video blog, there are dark circles under the respondent's eyes and a faint redness on the bridge of their nose and down their cheeks, presumably from wearing a respirator at work. Their hair is pulled into a messy ponytail, and they are wearing a wrinkled, grey t-shirt. I've even thought about calling in sick for shifts because I just don't want to go in and have to deal with all of this stuff . . . I don't want to go to work. At baseline, 521 respondents reported they were continuing to see patients. Changes in patient interactions included incorporating telehealth , decreased patient load , increased patient load , assisting in critical care , and assisting in other departments . Many respondents described feeling unsupported and ill-equipped to provide care to COVID-19 patients. One advanced practice provider said in a video blog: Yes, I chose to do training in medicine, and yes I wanted to be there, but I didn't want to feel like the resources weren't there or feel like I couldn't provide the right answers. --- Theme 2: Intrinsic Stressors The second theme centers on stress and anxiety related to the pandemic and other social factors. Participants expressed a sense of responsibility to help control these stressors. Two subthemes were identified: Fear of COVID-19 and concern about spreading Covid, and "Allostatic Load". --- Subtheme 2.1: Fear of COVID-19 and Concern about Spread Fear of Covid included general concerns about death and suffering, but, more often, many respondents specifically highlighted aspects of the disease process they feared experiencing or watching others experience. Concerns over the spread of COVID-19 took two distinct forms: concern about spread in the clinical environment, and a fear of spreading COVID-19 to family members and friends. This led to feelings of isolation as respondents struggled to balance anxieties and reasonable precautions. In an extreme example, a medical assistant and massage therapist reported testing positive. She is visibly short of breath: Unfortunately, because I felt fine on Saturday, I saw three of my massage clients, and so they were all exposed. I was wearing a mask and a face shield . . . [but] one of the three have tested positive. Her face falls, and she sits in silence for a few seconds. Concern about spreading the virus was even prevalent among students. A first-year medical student described her concerns visiting her parents: There was quite a bit of stress in that having parents above the age of 65 and have comorbidities and then coming to a family's house and having to deal with potentially spreading the virus to them unknowingly. It's the unknowing part of COVID that's the most stressful. These concerns were also represented in our quantitative results. Of the respondents, 71.4% endorsed fears of loved ones getting sick or dying, 52.1% endorsed concern for their own health, and 35.9% indicated concerns about their colleagues falling ill. A higher proportion of those seeing patients reported feeling isolated or lonely and concerns about care of family members . Respondents who reported not seeing patients endorsed concerns about personal resources, such as financial concerns, advanced directives, and home supplies . Full comparison of personal concerns between locked-in and locked-out HCW is included in Table 2. Some respondents felt the burden of additional social, political, and cultural stress added to concerns about COVID-19, including the contention surrounding the 2020 US presidential election and racial disparities highlighted by the death of George Floyd. Participants pointed out the pandemic was not occurring in a vacuum. Comments about politics, the Black Lives Matter movement, and the environment were made exclusively by those with "non-essential" roles, but some did hold dual certification. For example, a massage therapist shared her excitement about finishing a medical assistant course and getting to start work in an urgent care unit; then she shifted to describing struggling with the political views of her massage clients: I'm worried that if I make [wearing masks] a mandatory thing, I'm going to lose that client . . . It's a very conservative town, and a lot of people don't like to wear masks. She is pacing through her house, presumably recording on her phone which is held low in her hand. One respondent, who did not disclose their profession, stated in a video blog: I have a lot of people that are stressed by the politics and the election . . . There's different stress stimulus that wasn't present a few months ago. She sits close to the camera, and the light from the screen highlights heavy circles under her eyes. --- Theme 3: Extrinsic Factors Respondents frequently discussed economic impacts of the pandemic including being furloughed or laid off, as well as making decisions to open or close their businesses or to furlough or lay off team members. This was frequently seen among locked-out HCW, but one nurse described her experience: Our hospitals have a huge decrease of patients, which means starting next week we start furloughs . . . which is a very strange feeling, thinking that you're very important and you're going to do all this good, [then] you are not needed. One respondent described frustrations of her practice and work as a massage therapist being labeled as "non-essential" by her state. She sits on the floor in her bathroom, leaning against the cabinets wearing a bathrobe. She describes her decision to comply with the state orders despite hearing of other practitioners who have found loopholes to continue practicing, such as being part of a medical office. I'd be putting my livelihood, my business, and my professional recognition at stake if I were to justify to myself that medically necessary massage was ok with the mandates that our governor has made. Descriptive statistics of work-related experiences of locked-in HCW and a comparison of work-related concerns in locked-in versus locked-out HCW captured by Project COPE are included in Appendix A. --- Theme 4: Coping Strategies Respondents often reported activities intended to help them cope with the realities of the pandemic. Common activities included walking, gardening, reading, and meditation. Several participants described their coping strategies as "unhealthy," but often presented activities without judgment. One massage therapist said: I am sleeping a lot. I don't get out of bed very often. I have been looking for something to fill my time, so I've taken up some gardening and getting some ideas ready for yard work. Further representing this theme, one nurse shared: I've been coping by eating a lot of comfort food . . . I'm gaining weight like crazy, like 20 pounds since quarantine started. She is standing in front of a bathtub. Her hair is fixed with large curls, and she wears a fleece jacket. Respondents often reported difficulty with feelings of isolation. I've noticed with the downtime the social isolation has gotten to me. Some of my less healthy coping patterns have resurfaced like, um, tobacco smoking and comfort foods, um, and staying up way too late. Her speech is pressured, and she looks down while speaking, rarely making eye contact with the camera. In some cases, respondents described activities we coded as "hypervigilance". This included obsessive news and media consumption, attempts to calculate their own exposure risk, and robust decontamination procedures. Below is an example of this kind of response: We have a quarantine station we set up in the front door . . . We wipe everything down as soon as we get home, strip our clothes, they go right into the wash, we go right to the shower before we do anything else when we get home. The respondent is sitting on her couch. Her hair is wet and her face red. Lines on her face trace the impression of a face mask or respirator. In another example, one participant submitted an audio-only recording. She speaks slowly, stumbling over her words. I feel more vulnerable to COVID . . . I've been paying a lot of attention to the number of people I've massaged and their occupation. Once I reached a hundred and fifty since we've been reopened, I looked at my data and . . . 11.3% of those 150 are nurses. A proportion of respondents reported at least one maladaptive behavior including using alcohol , sleeping pills , illicit substances , or tobacco . No significant differences between those who had and had not stopped seeing patients were found. Additional information on coping strategies is included in Table A4 in Appendix B. --- Moral Distress and Burnout A total of 393 participants completed the Well-Being Index in the first weekly survey, with 344 reporting at least one symptom of burnout. Forty-nine reported no signs of burnout, while 86 reported four or more. The mean composite score for the WBI was 2.40 , indicating low average burnout among participants. The Mini-Z captured 381 responses in the first weekly survey. The majority of respondents ) scored at a level 3 or lower and 248 scored a 2 or lower. Only four respondents reported level 5 . Most respondents reported experiencing moral distress less than once a day, if at all . Those who stopped seeing patients were less likely to report higher levels of moral distress than those who continued seeing patients after controlling for years of practice, level of education, and race . Having stopped seeing patients was also associated with lower odds of reporting higher moral distress in medicine professions . In the baseline survey, those who stopped seeing patients were less likely to report higher levels of moral distress than those who stopped seeing patients . Having stopped seeing patients was also significantly associated with lower odds of reporting higher moral distress in medicine professions . See Table 3 for additional results. In the first weekly survey, those who reported they had stopped seeing patients were less likely to report higher levels of moral distress . Similar trends were observed when outcome measures were treated as binary variables. Those who stopped seeing patients were less likely to report moral distress . These results are displayed in Table 4. --- Table 4. Odds ratio estimates of locked-out versus locked-in HCW on moral distress, burnout, and well-being, controlling for baseline demographics and years of experience. --- Scale --- Discussion Through this mixed-methods study, the experiences and perspectives of healthcare providers were explored and stratified across professionals locked-in and locked-out of delivering healthcare services. Several key findings can be drawn from this study: "Locked-in" and locked-out healthcare professionals experienced feelings of burnout, moral distress, and struggled with overall well-being; experiences contributing to these feelings are sometimes varied across professional category; and differences in distinct experiences do not appear to be associated with differences in reported feelings of burnout or moral distress. Locked-in and locked-out HCW experienced similar concerns about the lack of planning and resources, frustrations with communication and misinformation, and the conflict between professional and personal responsibilities. Differences in concerns related to documentation and organizational policies may be largely due to differences in work settings or that locked-out HCW, such as massage therapists, were more likely to be independent practitioners. Further research is needed to illuminate the causes of these patterns. One example of differences in distinct experiences not being associated with differences in feelings of stress was financial and personal resource concern. Financial and personal resource concern was reported in vlogs of both groups, but locked-out healthcare workers reported concerns more often through vlogs and through Fisher's Tests of survey responses, stratified by professional category. At least one locked-in professional described concerns about the financial impact of impending furloughs, and 43.2% of locked-in survey respondents reported personal resources as a concern. This study also found contradicting results in measures of burnout and moral distress, which may be explained by the decisions associated with having a choice on whether to see patients. "Locked-out" professionals were almost half as likely to report higher levels of moral distress, but one and a half times more likely to report higher levels of burnout compared with those who felt "locked-in". Respondents who indicated they made a personal choice to stop seeing patients were half as likely to report higher levels of burnout on the WBI compared with those who stopped for other reasons. Those who made a choice to stop seeing patients may have seen the decision as a way to contribute to the public health effort to limit disease transmission, and though they may have perceived themselves to be worn down and ranked higher on the Mini-Z, they did not experience the symptoms captured in the WBI. Conversely, those who continued to see patients reported feeling lonely and isolated as well as concerns about dependent care at higher rates than those who had stopped seeing patients. Based on qualitative findings, it may be that those continuing to see patients are more likely to self-isolate out of fear of spreading COVID-19. However, further investigation is recommended. Both groups endorsed a sense of uncertainty about how long the pandemic would last, fears of social instability, and frustrations about pandemic misinformation. Among locked-out professionals, this was expressed as general anxiety and concern for public health. Project COPE participants who continued to see patients expressed frustrations and fear for their personal well-being and family health. Regardless of profession, COVID-19 was shown to have a prominent impact on healthcare professionals, and responses across both participant groups highlighted the numerous areas of impact both personally and professionally experienced during the pandemic. --- Conclusions Several limitations were inherent to the study design and were mitigated as much as possible. Through our sampling methods, there was an imbalance in the professional groups included, which limited the rigor of statistical methodologies and our ability to explore the mechanisms of our findings. In addition to professional category distribution, sampling was limited to predominately white women, which may influence the generalizability. Comparison across professional categories was also limited by the applicability of standard measures across healthcare settings. For example, differences in work-related stressors may be due to differences in practice settings. Similarly, phrasing of measures may have influenced respondents' understanding of the questions or ability to apply it to their perspective. Questions pertaining to organizational support are potentially skewed due to the oversampling of massage therapists, many of whom own their own practices [27]. Project COPE added new voices to the rapidly growing literature on healthcare provider burnout, moral distress, and well-being through longitudinal video recordings with providers both locked-in and locked-out of healthcare services. Similar to other studies, this manuscript highlights the impacts of isolation and fear on healthcare workers but added new evidence around the experiences of professions largely absent from the literature. This study found increased levels of burnout in both categories of providers, but data suggest differing challenges to stress, practice constraints, and disease-related fears. Public health efforts and future research may benefit from studies exploring the distinct mechanisms related to burnout in those locked-in and locked-out of health care services to provide better support to a group that desperately needs it. --- Data Availability Statement: Data from this project is not publicly available due to the potential of identifiability of video blog recordings. --- --- Appendix A

During the COVID-19 pandemic, healthcare workers (HCW) were categorized as "essential" and "non-essential", creating a division where some were "locked-in" a system with little ability to prepare for or control the oncoming crisis. Others were "locked-out" regardless of whether their skills might be useful. The purpose of this study was to systematically gather data over the course of the COVID-19 pandemic from HCW through an interprofessional lens to examine experiences of locked-out HCW. This convergent parallel mixed-methods study captured perspectives representing nearly two dozen professions through a survey, administered via social media, and video blogs. Analysis included logistic regression models of differences in outcome measures by professional category and Rapid Identification of Themes from Audio recordings (RITA) of video blogs. We collected 1299 baseline responses from 15 April 2020 to 16 March 2021. Of those responses, 12.1% reported no signs of burnout, while 21.9% reported four or more signs. Qualitative analysis identified four themes: (1) professional identity, (2) intrinsic stressors, (3) extrinsic factors, and (4) coping strategies. There are some differences in the experiences of locked-in and locked-out HCW. This did not always lead to differing reports of moral distress and burnout, and both groups struggled to cope with the realities of the pandemic.

INTRODUCTION Cataract is the leading cause of blindness globally. [1] Reducing cataract blindness is a priority in the Universal Eye Health: a Global Action Plan 2014-2019 [2] endorsed at the 66 th World Health Assembly in 2013. The UEH Action Plan aligns with the broader Universal Health Coverage initiative and recognises the importance of promoting equity to achieve universal eye health . Assessing progress towards health equity necessitates monitoring inequality . [3] Cataract blindness and services are unequally distributed across countries and regions, with higher rates of cataract blindness and worse cataract service indicators in poorer and less developed settings. [1,4] Within countries, women generally [5,6]-but not always [7]-are more likely to be cataract blind than men, with lower access to and use of cataract surgical services. Data to monitor inequalities beyond gender are not routinely collected and reported. A small number of surveys have analysed cataract outcomes for associations with social variables such as place of residence and education level and/or literacy, usually as independent risk factors in multivariable logistic regression models. [8][9][10] However, because individuals are not unidimensional , the multivariable approach typically fails to capture the experience of specific subgroups. Little insight is typically available for the groups 'in the middle' such as illiterate men or literate women, and removes the opportunity to understand how social factors interact with one another to increase vulnerability in the most disadvantaged. [11] This study aimed to assess inequality in cataract blindness and cataract services at the intersection of three social variables using data from national surveys in Pakistan and Nigeria. --- METHODS --- Data sources Data from nationwide representative surveys in Pakistan and Nigeria were selected for analysis as they are among the largest blindness surveys undertaken, collected data on a wide range of social variables, and used comparable methods. [12,13] Participant characteristics and previously reported cataract outcomes from these surveys are summarised in Table 1. Ethical approval for the surveys was granted prior to original collection by the relevant entities in each country. For the analyses presented here, approval was obtained from the London School of Hygiene & Tropical Medicine Ethics Committee . --- Cataract results Crude prevalence of cataract blindness --- Outcome variables The three outcome variables examined were cataract blindness, cataract surgical coverage , and effective cataract surgical coverage . Cataract blindness was defined as presenting visual acuity of worse than 3/60 in the better eye where the principal cause was cataract. [12,13] Cataract surgical coverage measures the number of people in a defined population with operated cataract as a proportion of those having operable plus operated cataract. In this study 'operable cataract' was defined as cataract causing blindness . CSC was calculated using the formula [/] *100 where x is individuals with unilateral pseudo/aphakia and operable cataract in the other eye; y is individuals with bilateral pseudo/aphakia; and z is individuals with bilateral operable cataract. [8,15] Effective cataract surgical coverage measures the number of people in a defined population with operated cataract and a good outcome as a proportion of those having operable plus operated cataract. eCSC was calculated using the formula [/] *100 where a represents individuals with unilateral pseudo/aphakia achieving presenting VA of 6/18 or better in the operated eye and operable cataract in the other eye; b describes individuals with bilateral pseudo/aphakia achieving presenting VA of 6/18 or better in at least one eye; and x, y and z are the same as for CSC. --- Social variables The social variables used, hereafter described as 'axes', were 1) sex , 2) literacy and 3) place of residence . Eight subgroups were generated in each dataset by combining these axes . The subgroups selected a priori as more advantaged on each of the axes were males, literate individuals, and urban dwellers. Box 1: Subgroups created by disaggregating women and men by literacy status and place of residence --- Analysis methods Analyses were conducted using STATA 12.0 . The social distribution of cataract indicators were examined by calculating cataract blindness prevalence, CSC, eCSC, and their 95% confidence intervals for each subgroup listed in Box 1, and constructing graphs for each outcome. The World Health Organization recommends rate differences and rate ratios as simple measures to monitor health inequality. [17] We selected the most advantaged subgroup as the reference group for calculations . RDs and RRs were calculated for CSC and eCSC. A positive value for RD and RR>1 indicates worse health in the disadvantaged. The more RD diverges from zero and RR diverges from one, the greater the inequality. Conversely, a negative value for RD and RR<1 indicates worse health in the subgroups that were assumed to be advantaged. To examine the cumulative effect of multiple disadvantage, literacy status and place of residence were added to sex in turn, and the increased odds of higher cataract blindness, and lower CSC and eCSC in the worst-off compared to best-off subgroup was calculated using multiple logistic regression. Nine models were constructed in each country-one for each of the three outcomes for three axis combinations. The assumed best-off subgroup was used as the reference group, and age was forced into the model. For each outcome, age-adjusted odds ratios and 95% CI were plotted comparing 1) men and women; 2) literate men and illiterate women; and 3) urban, literate men and rural, illiterate women. Goodness of fit was assessed using the chi-squared test. --- RESULTS --- Social distribution of cataract indicators --- Cataract blindness The social distribution of cataract blindness was similar in both countries, with illiterate women and men having a higher prevalence than their literate counterparts. Within literacy subgroups, rural dwellers had a higher prevalence than urban dwellers . Literate urban dwellers had the lowest prevalence in both countries: in Pakistan literate, urban men and in Nigeria literate, urban women . Conversely, illiterate, rural men in Pakistan and illiterate, rural women in Nigeria had the highest prevalence of cataract blindness. In Pakistan, despite two subgroups of men having the highest prevalence of cataract blindness , the overall prevalence was lower in men compared to women . This is due to the higher level of social disadvantage experienced by women-the two most disadvantaged subgroups of men comprised 25% of the sample, whereas the corresponding subgroups of women comprised 45% of the sample . In Nigeria, the proportion of women in the two most disadvantaged subgroups was more than twice that of men , leading to an even greater difference in overall prevalence between women and men . --- Cataract surgical coverage and effective cataract surgical coverage A social gradient was observed among women and men for CSC and eCSC, with coverage increasing as social position improved, being lowest for illiterate rural dwellers and highest for literate urban dwellers . The slope of the gradient was steeper for men than women in Pakistan, whereas in Nigeria the opposite was observed, with greater differences among women than men. In Pakistan, inequality in eCSC was approximately twice as high as inequality in CSC in both absolute and relative terms-absolute inequality was 21.6% for CSC and 45.9% for eCSC, and relative inequality was 1.3 for CSC and 2.7 for eCSC. In Nigeria relative inequality also around twice as high in eCSC compared to CSC . In absolute terms inequality in eCSC was lower than for CSC , partly due to much lower rates of eCSC. --- Cumulative disadvantage Cumulative disadvantage was evident for all outcomes, with increasing odds of higher cataract blindness prevalence, lower CSC and lower eCSC with the addition of each axis of disadvantage . The findings were similar in both countries. --- DISCUSSION The UEH Action Plan states that "All people should have equitable access to health care and opportunities to achieve or recover the highest attainable standard of health, regardless of age, gender or social position", [2] closely aligning with UHC aims. [18] Our results show that at the time these surveys were completed there was much to be done to achieve UEH in Pakistan and Nigeria. Despite differences in sociocultural and health system contexts, both countries had large numbers of people blind from cataract , with a disproportionate number being in disadvantaged social groups . The most disadvantaged groups also had lower CSC , most likely due to barriers to accessing services not experienced by their more advantaged counterparts. Amongst those who accessed surgery, the disadvantaged were less likely to have good visual outcomes than the more advantaged . The larger inequality observed in eCSC compared to CSC highlights the compounding disadvantages of poorer access and worse quality for those already socially disadvantaged. This phenomenon was described decades ago as the inverse care law-those most disadvantaged and in greatest need typically have last access to good quality services. [19] The cumulative disadvantage evident in both settings for all outcomes suggests more comprehensive and equity-focused approaches are required to reduce inequalities and achieve the aim of UEH. Some unpredicted insights also emerged. Men were not uniformly better off than women and certain subgroups of men in Pakistan experienced the worst levels of cataract blindness , while more socially advantaged women fared better than disadvantaged men in all outcomes . Indeed, in Nigeria, literate, urban women fared best in all three outcomes. Such differences highlight the benefit of exploring social axes beyond gender, as well as emphasizing the more nuanced picture that emerges beyond the standard multivariate 'risk factor' analysis, which previously identified associations between female sex and cataract blindness and CSC in both settings . [8,[14][15][16] Moving beyond unidimensional analysis of social position was an approach included in the Equity-Focused Health Research Agenda developed by a World Health Organization Task Force on Research Priorities. [20,21] We believe this is the first example of such an approach in eye health. The enhanced value of more finelygrained information on inequalities, as presented here, is that it provides evidence of specific forms of disadvantage, throws up new questions for research, and facilitates the targeting of interventions to those who have the potential to benefit most. [11,22] Our analysis must be interpreted in relation to several potential limitations. First, the surveys upon which these analyses were based were completed several years ago, and should not be interpreted as representing the current situation in these countries. The data are nevertheless illustrative, may well hold true in the countries studied as well as many others, and provide a baseline to assess future inequalities in these countries. Furthermore, this study demonstrates the types of analyses that are likely to be of value in blindness surveys in the pursuit of UEH. Second, despite the fact that these surveys are amongst the largest eye surveys undertaken in LMICs, the size of some subgroups was small, resulting in large 95% confidence intervals around the estimates. Finally, a hierarchical socioeconomic variable from which to construct wealth quintiles was not collected in the surveys, so commonly used inequality measures could not be calculated and we had to rely on simple measures . Despite this, inequalities were evident across the social axes explored, reinforcing the need to undertake more finely-grained analyses to identify the subgroups of the population with the greatest needs along with potential to benefit from service improvements. [21] There is also much that could be learned from positive outcomes. For example, how did Pakistan achieve near universal CSC for the most advantaged men, and has this been sustained? Furthermore, how can the quality of cataract surgery be improved for everyone while also extending, accelerating and prioritising good quality services to those most disadvantaged? The promotion of UEH is an important contributor to achieving universal health coverage. For UEH to be achieved the prevalence of cataract blindness must be reduced, as well as its unequal distribution addressed. Importantly, cataract blindness can be ameliorated at relatively low cost and with substantial benefits to individuals and societies. [23] More evidence on how and with whom to intervene is required if we are to reduce unfair and unacceptable gaps that arise from social and economic disadvantage. [24] The analysis approach presented here contributes to filling the evidence gap by identifying the nature and extent of inequalities within populations. Such evidence provides insights into how best to design and evaluate scalable interventions and should help inform policymaking. [25] It is now up to societal, health and research leadership to identify and prioritise those most disadvantaged and to make available to them opportunities to access interventions that may transform their lives, thus demonstrating real commitment to "leave no one behind". I=illiterate, L=literate, R=rural, U=urban. For each subgroup, the prevalence of cataract blindness is plotted along the y axis, and the proportion of the sample in each subgroup is indicated at the bottom of each bar e.g. for Nigeria 32% of the sample were illiterate, rural females. Inequality was assessed by age-adjusted logistic regression, comparing the best-and worst-off in each axis combination; the odds ratio quantifies the association with i) higher prevalence of cataract blindness, ii) lower CSC and iii) lower eCSC. For all models, the chi-squared test showed no evidence of lack of fit. --- Pakistan Nigeria 1 axis=female versus male*; 2 axes= illiterate female versus literate male*; 3 axes=illiterate rural female versus literate urban male* Pakistan Nigeria --- Competing interests: None of the authors has any conflicts of interest to disclose. Financial support: Funders of the initial survey in Pakistan were Sightsavers, CBM, Fred Hollows Foundation, Pakistan Office of the World Health Organization, Ministry of Health. In Nigeria the original survey was funded by Sightsavers, Velux Stiftung, CBM, and Ministry of Health. Those funders had no role in the analyses presented here. This analysis received no specific funding. Running head: Inequality in cataract blindness and services

Inequalities in cataract blindness are well known, but data are rarely disaggregated to explore the combined effects of a range of axes describing social disadvantage. We examined inequalities in cataract blindness and services at the intersection of three social axes.Three dichotomous social variables (sex [male/female]; place of residence [urban/rural]; literacy [literate/illiterate]) from cross-sectional national blindness surveys in Pakistan (2001-4; n=16,507) and Nigeria (2005-7; n=13,591), were used to construct eight subgroups, with disadvantaged subgroups selected a priori (i.e. women, rural dwellers, illiterate). In each dataset, the social distribution of cataract blindness, cataract surgical coverage (CSC), and effective cataract surgical coverage (eCSC) were examined. Inequalities were assessed comparing the best-and worst-off subgroups using rate differences and rate ratios (RR). Logistic regression was used to assess cumulative effects of multiple disadvantage.Disadvantaged subgroups experienced higher prevalence of cataract blindness, lower CSC and lower eCSC in both countries. A social gradient was present for CSC and eCSC, with coverage increasing as social position improved. Relative inequality in eCSC was approximately twice as high as CSC (Pakistan: eCSC RR 2.7 vs CSC RR 1.3; Nigeria: eCSC RR 8.7 vs CSC RR 4.1). Cumulative disadvantage was observed for all outcomes, deteriorating further with each additional axis along which disadvantage was experienced.Each outcome tended to be worse with the addition of each layer of social disadvantage. Illiterate, rural women fared worst in both settings. Moving beyond unidimensional analyses of social position identified subgroups in most need; this permits a more nuanced response to addressing the inequitable distribution of cataract blindness.

INTRODUCTION Online sex seeking is an increasingly common practice among men who have sex with men 1,2 . The online environment provides MSM an alternative venue in which men can build social connections and find new sex partners 3 . At the same time, there is concern that sex seeking online may increase the risk of acquiring and transmitting STDs [4][5][6][7] . Technological advances now provide MSM multiple platforms for online sex seeking. Such platforms include websites and gay mobile apps 8 . MSM are likely to seek sex through a wide range of websites . These websites connect MSM from different places, and MSM often develop online relationships before meeting in person 9 . Gay mobile apps are another increasingly popular approach to online sex seeking. Unlike websites, gay apps use geospatial technology to identify physically proximate men 10,11 . Gay apps differ from websites in terms of a) the emphasis on physically proximate partners that decreases barriers to in-person meeting 12 ; b) the real-time nature of partner choosing that facilitates rapid meeting 11 ; and c) the widespread default profile photos which may simplify partner selection 2,13 . These differences may accelerate sex-seeking and promote condomless sex 7 . Along with the rapid proliferation of social media, using multiple types of platforms for social networking is generally on the rise, from 42% of Internet users in 2014 to 52% in 2015 14 . Evidence from online and offline research indicates that diversifying platforms to meet sex partners may increase the possibilities of causal sexual encounters and spread infections among MSM 15,16 . There is uncertainty about whether adopting multiple dating tools, e.g., using both websites and apps together, facilitates riskier sexual behaviors when compared to only using one platform 10,17 . In China, gay dating websites have been emerging since the late 1990s 16 and nearly half of Chinese MSM have used websites for sex seeking 18 . Gay apps have rapidly expanded in China during recent years 5,6 . As of 2015, 27 million MSM have used Blued 19 , the world's largest sex-seeking gay app. Despite the similar purpose of both platforms, their features vary which likely attract different groups of MSM. Identifying the characteristics of MSM groups based on sex seeking platforms could help tailor interventions. Studies on comparison of website users, app users, and men who use both platforms are limited, especially in middle-income countries 2,17 . The purpose of this study was to identify and compare socio-demographics and sexual risk behaviors between Chinese MSM in the following subgroups: website only users, gay app only users, and men who use both platforms. --- METHODS --- Setting In China, approximately 7.7% of MSM have HIV infection. The proportion of new HIV cases attributable to MSM behaviors has increased from 2.5% in 2006 to 25.8% in 2014 20 . The data for this study were collected through a cross-sectional, online survey from September through October 2014. Advertisements for study participation were put on three large, gay web portals that reach a large number of MSM in China. Banner links were presented on each web portal homepage and the announcement was sent out to registered users. The survey was developed based on interviews with MSM and then reviewed by many local stakeholders workers, social media experts, physicians, and public health experts) 6 . To be eligible for the survey, participants must have stated that they were born biologically male, were older than 16, and engaged in anal or oral sex with a man in the previous year. Participants signed a consent form before filling in the survey, and eligible participants received a small phone card reimbursement upon completion. Overall, a total of 1,424 eligible MSM finished the online survey. Among them, we excluded those who did not have sex in the last six months and those who only used offline means for finding sex partners . --- Measures The survey measured socio-demographic information, sexual orientation, and online sex seeking behaviors. Based on survey responses, MSM were divided into three subgroups: website only users, gay app only users, and men who used both platforms for online sex seeking in the past six months. Each subgroup was further questioned about behaviors with sex partners they met in the past six months, including number of partners, condom use, time from initial conversation to in-person meeting, negotiation of condom use before meeting, HIV and STI testing, and exposure to STD-related messages. Sexual behaviors, including group sex, commercial sex, and sex when using drugs, were also assessed. --- RESULTS Of the 1201 individuals who were engaged in online sex seeking, 377 men were website only users, 487 were app only users, and 337 were men who used both platforms. --- Sample Socio-demographics Overall, the average age of men was 25.6 years old , with the majority between 20-29 years old . One quarter had at most attended high school and 10.7% were married. In addition, 89.8% of men lived in urban areas and 81.9% had an annual income less than $9600 USD. In total, 64.4% of men reported sexual orientation disclosure to anyone other than their partners. Nearly half of the participants had a main male sexual partner . --- Socio-demographic characteristics of three MSM subgroups Socio-demographic characteristics were significantly different between the three groups of men . First, there was significant difference in age, χ 2 = 36.29, p < 0.001. Younger men constituted 72.3% of gay app only users and 70% of men who used both platforms, as compared to 59.9% of website only users. Education level was also different between the three groups, χ 2 = 18.52, p = 0.001. For instance, 32.4% of website only users, 24.4% of gay app only users, and 19.9% of men who used both platforms received no more than a high school education. Furthermore, the three groups were different in marital status, χ 2 = 45.36, p < 0.001. The percentages of those who had ever married were 26.0%, 9.4% and 13.6% for website only users, gay app only users, and men who used both platforms, respectively. Lastly, the three groups were different in disclosure of sexual orientation, χ 2 = 14.98, p < 0.001. Among men who used both platforms, 70.6% had disclosed their sexuality or sexual history to someone, as compared to 57.0% of website only users and 65.7% of gay app only users. --- Sexual behaviors between three MSM subgroups Gay app only users were more likely to have multiple sex partners than website only users , χ 2 = 204.81, p < 0.001. However, there was no significant difference in condomless anal sex behaviors between the three subgroups. In total, 57.6% of men reported that they had engaged in condomless anal sex with their last male partner in the past six months. The percentages were 55.7%, 59.5% and 57.0% for website only users, gay app only users, and men who used both platforms, respectively. Also, exposure to STD-related messages was different between men using different platforms ; 58.1% of website only users, 63.2% of gay app only users, and 75.1% of men who used both platforms attained STD-related messages . --- Factors associated with choices of sex-seeking platforms Table 4 shows the association between online sex seeking platforms and men's sexual behaviors after adjusting for potential confounders, including age, education, income, marital status, residency, sexual orientation, sexual orientation disclosure, and currently had a main partner. Compared with website only users, gay app only users have multiple sex partners . There was also an association between sexpartner seeking platforms and the time duration between initial conversation to in-person meeting. Compared with website only users, gay app only users were more likely to meet sex partners within a day or within 2 to 7 days . Men who used both platforms were more likely to meet in person within one day or 2 to 7 days in comparison to website only users and gay app users . Using different online platforms to seek sex partners was also significantly associated with condom use negotiation. MSM who used both platforms were less likely to negotiate for condom use compared to website only users . Finally, the results showed that men who used both platforms exposed to more STD-related information than website only users . --- DISCUSSION Multiple online platforms are now available for MSM to meet sex partners around the world, creating barriers and opportunities for HIV prevention 21 . The online environment creates a new local context that may influence sexual risk taking 6,7 . This study extends the literature on online sex seeking 2,17 by differentiating website and gay app channels, examining men who used both types of platforms as a separate group, and focusing on a middle-income country. Our data may help inform the development of tailored online interventions. Our data suggest that these three MSM subgroups have distinct socio-demographic characteristics. Men who only used websites were more likely to be older, received less education, and married compared to men who only used gay apps. Men who used both platforms were similar to men who were gay app only users, except that men who used both platforms were more likely to have ever disclosed their sexual orientation to others. This observation is consistent with other literature showing that online sex seeking preferences are related to sexual orientation disclosure 22 . Openness of sexual identity is likely to be associated with active engagement in community and sexual networks as such men may feel safer and more comfortable to get involved. Young men are more likely to use either gay apps exclusively or both platforms. This trend may be because young men generally adopt new technologies faster and are more interested in getting involved in multiple platforms 11 . This trend suggests targeting interventions for older MSM through websites and for younger MSM through both websites and gay apps. Although dating websites are decreasing in frequency as compared to gay apps, this study re-emphasizes the importance of performing promotional interventions on websites. A subgroup of MSM remains active on websites because they have become used to websites or they have privacy concerns regarding the geo-locating features of gay apps 13,16 . We found similar rates of condomless sex among different MSM subgroups. Our study showed that 57.6% of men had condomless sex in the past six months, which is consistent with earlier literature 2,18,23 . This high prevalence of unsafe sex behaviors underlines the importance for HIV prevention intervention to reach men on both websites and gay apps. Prior studies reported inconsistent findings on whether sex seeking platforms influence risky sexual behaviors. Some studies indicated gay app users were riskier compared to non-app users 7,24 , some studies suggested gay app users were less risky 11 , and other studies found no difference between these two groups 17,21,25 . Our study also found no difference and is consistent with findings from Hong Kong 21 and the US 25 . More importantly, it is consistent with the only identifiable empirical study in China that gay app users were associated with higher rates of multiple sex partners but not associated with higher rates of condomless anal sex behaviors 5 . Despite similar condomless behaviors, gay app users had greater numbers of sex partners compared to website users. Our findings indicate that sex seeking platforms may be facilitators for meeting sex partners, but they are not necessarily related to risky sexual behaviors. Lastly, this study showed that men who used both platforms received more STD-related messages. This finding suggests that social media interventions may benefit from using multiple platforms. It aligns with prior studies that using a combination of active and passive channels can reach a larger population for sexual health promotion 26 . This study also echoes reports that consider online dating and sex seeking channels as important sources for health information and support 17 . Social and sexual networks may provide more opportunities to conduct innovative interventions for HIV prevention 27 . This study examines men's online sex seeking and risky sexual behaviors in the context of China. Due to the highly stigmatized environment in China, MSM often utilize fast, convenient online channels for sex seeking 15 . In addition, advanced development of mobile internet allows individuals to be connected at all times in any place. The accessibility and affordability of mobile internet in China facilitates men to adopt mobile apps as social networking channels 28 . With more platforms available, men's likelihood of finding a sex partner increases. However, we found no correlation between condomless sex are platforms used. Furthermore, men are likely to transfer from old platforms to new platforms 29 . It is important for this study to note that although an increasing number of men are adopting gay apps in China, some men prefer to remain on websites for sex seeking. This suggests that interventions should consider using both old and new platforms. This study has limitations. First, we only examined men who had sought sex online in the past six months. Although this may overestimate sexual behaviors among the general MSM population, it would not influence the validity of comparing men's online sex seeking platforms. Second, men self-reported their behaviors and this may lead to social desirability bias. Nevertheless, participants' personal information was not collected except for their cellphone numbers for compensation. Third, this study used a cross-sectional survey and thus could not claim causal relationship between men's online sex seeking behavior and their risky sexual behavior. This study has policy and research implications. First, MSM social media interventions may opt to use multiple platforms in order to better reach MSM. Future studies can look more closely into men who use multiple platforms 2,17 . Moreover, as online sex seeking platforms proliferate alongside an expanding HIV epidemic among MSM 20 , how these online behaviors relate to disease transmission requires research. Additionally, men acknowledged receiving health messages when seeking sexual partners online. This finding is of great importance for men who are not motivated to actively search for health information. Further investigation can be done on the behavioral differences associated with actively seeking versus passively receiving sexual health information on safe sex behaviors and disease control. --- CONCLUSION The persistent increase in newly diagnosed HIV infections among gay men in China calls for intensifying interventions in key areas and target groups 20 . Given gay dating websites and apps are both risk environments where low prevalence of consistent condom use are identified, focusing HIV prevention interventions in these online platforms may be useful 27 . Different socio-demographic features between MSM subgroups suggests tailoring interventions on different platforms for different generations. Meanwhile, government and health agencies have collaborated with gay dating apps to fight the spread of HIV 30 . Since MSM who used both platforms reported receiving more online health messages than single platform users, more such cooperation with both gay websites and apps should be encouraged. --- Supplementary Material Refer to Web version on PubMed Central for supplementary material. --- ---

Background-Many men who have sex with men (MSM) seek sex partners online, creating barriers and opportunities for HIV prevention. The purpose of this study was to examine the characteristics of MSM and the risks associated with seeking sex through websites, gay apps, and both platforms in China. Methods-Data was collected through a cross-sectional online survey from September through October 2014 from three large gay web portals. Socio-demographic information, sexual behaviors, and online sex seeking behaviors were measured. Multinomial logistic regression was performed to compare sexual risk behaviors among website users, gay app users, and men who used both platforms. Results-Of the 1201 participants, 377 (31.4%) were website only users, 487 (40.5%) were gay app only users, and 337 (28.0%) were men who used both platforms. These three MSM subgroups have distinct socio-demographic characteristics. Overall, 57.6% of participants reported having engaged in condomless anal sex with their last male partner in the past six months, but there was no significant difference in condomless sex between the three groups. Men who used both platforms viewed more STD-related messages than website only users (aOR = 2.19, 95% CI:1.57-3.05).

unemployment, or other signs of distress . Subsequent studies analyzed trends in income segregation across all income levels, with particular attention to income segregation within racial/ethnic groups. The most recent research has been conducted in a period when social scientists, policymakers, and the public have become more acutely aware of issues associated with rising income inequality . Several recent reports have found that income segregation, too, is on the rise, increasing the estrangement of rich from poor and possibly leading to a decline in support for meeting the needs of less affluent Americans . Two patterns stand out in recent studies. First, past changes in overall income segregation have been unsteady, declining in one decade and rising in another, but segregation has been found to rise substantially after 2000. Second, income segregation is described as higher and rising more quickly within minority populations. These findings are widely enough accepted that they are referenced in public statements by political leaders: "What used to be racial segregation now mirrors itself in class segregation. This great sorting taken place. It creates its own politics. There are some communities where … I don't even know people who have trouble paying the bills at the end of the month. I just don't know those people. And so there's less sense of investment in those children." . There have been many studies of income segregation in recent decades, and these reach different conclusions depending on the time period, the measure of income segregation, and the segment of the population being studied . Several studies distinguish between income segregation among whites and blacks . Representative of the most recent research is a widely cited study by Bischoff and Reardon , using the ACS from 2007-2011. Some of their results are reproduced in Table 1, including an overall summary measure of family income segregation , a measure of the separation of families in the bottom decile of the income distribution from all others , and a measure of the separation of families in the top decile . Table 1 reports their measures for the total population and the black population for 2000 and 2007-2011 for several large metropolitan areas along with the mean and standard deviation of all the large metros in their sample. Values in these and subsequent tables have been multiplied by 100, so their theoretical range is 0 to 100 and differences can be characterized in percentage points. The observed patterns of change vary depending on the measure and population segment that is studied. For the total population H and H90 increased in all 6 example metros and rose on average in large metros by 1.3 and 1.5 points, respectively. H10 increased in 4 of the 6 example metros and rose by 1.7 points in the average large metro . These results support reports of increasing income segregation. Results for the black population show even higher values of segregation and larger increases over time. H, H10 and H90 increased for all 6 example metros and the average value across metros with large black populations rose by 8.2, 9.9, and 11.0 points, respectively . We cast doubt on these findings. Our main insight is that all of these studies rely on sample data collected by the Census Bureau. Yet it is well known that the effective samples to estimate income distributions within census tracts were substantially downsized with the introduction of the American Community Survey after 2000, while sample sizes within census tracts for minority populations have always been smaller than for the non-Hispanic white or total population. This recognition raises the general problem of small area estimation . Estimates from random samples are known to be unbiased, but the variance of estimates can be quite large when samples are small. Social scientists in the past have treated the census's income tabulation in census tracts as though it were not based on a sample, presuming that the one-in-six long form data were sufficiently reliable for their analyses. But as Voss observes, in the ACS "standard errors of most estimates are so large that even substantial differences in numbers lack statistical significance" . For example, the most recent estimate of the median household income in relatively affluent tract 107.01 in Boston's Back Bay neighborhood in the 2013 ACS is $99,234. The Census Bureau calculates a standard error of plus or minus $13,552. So, staying within that confidence interval, the tract's median income may well have been as low as $86,000, or it may have been as high as $112,000. We suspect that the less reliable the income estimate is for every census tract, the larger will be the estimated variation across tracts -a value that is at the heart of all measures of income segregation. If so, it is possible that in the full population -not in the samples enumerated by the Census Bureau -there actually was no change or even a decline in income segregation. Further, because the calculations for the black population in Table 1 were drawn from original samples that were typically less than half as large as those for the total population, and often only 10-20% as large, the apparently stronger pattern of increases in black income segregation may also be illusory. Reardon andBischoff were aware of the problem that "When the number of households in a unit is small , estimates of within-unit variation will be biased downward, meaning that estimates of segregation will be biased upward." They proposed to equalize sample sizes across metropolitan areas by a second stage of sampling, arguing that the resulting estimates would be "comparable across race groups, metropolitan areas, and years, regardless of population size" . We show in Appendix I that procedures of this type do not compensate for differences in sample sizes in the original census samples. --- Pitfalls and solutions in measuring income segregation We are aware of two sources of error in the observed measures of income segregation based on publicly available data. The one that we focus on in this study is bias associated with limited sample size. The other is the Census Bureau's withholding of data in public tabulations of income at the census tract level. The latter can only be overcome by gaining access to the original sample data in a Federal Statistical Research Data Center , but we mention it here to underline the fact that even the bias-corrected measures that we will present below -while an improvement over past results -are not definitive. --- Data suppression in the census The Census Bureau's concern with disclosure of information about individual persons has led to a specific pattern of data suppression: race-specific income data in tracts with few residents of a given subgroup are not published. This practice has consequences for measures of group-specific income segregation, and it may also have affected studies showing that affluent blacks and Hispanics are under-represented in higher income neighborhoods . To illustrate the nature of the problem, we have analyzed suppression of income data for black households in the 2007-2011 ACS. We found that nationally there were over 17,000 census tracts with black residents for which no income distribution for black-headed households is reported. These tracts have a total black population over 1.2 million. The median income of the suppressed tracts is over $70,000, and in most tracts members of all racial groups have similar income levels. This compares to the median income for families with a black householder in non-suppressed tracts of under $45,000. The implication is that small, affluent minority populations in affluent neighborhoods have not been taken into account in past research. The same issue arises for poor whites whose data will be suppressed in predominantly minority neighborhoods. In both cases, suppression could lead to an underestimate of income segregation. Another disclosure practice is that in published tables at the tract level , the highest category of income is $200,000 and above. This top-coding is of concern to studies of income inequality because it obstructs estimation of the income levels of households/families in the highest income category . It is also an issue in studies of income segregation where assumptions must be made about the distribution of values in each category of income to calculate the variation of income within tracts . The top category is the most obvious problem because it has no necessary upper bound and the upper limit could vary greatly across tracts. It is especially problematic for measures calculated from the actual incomes rather than rankordered percentiles. Jargowsky uses such a measure and deals with the problem by assuming that the top category has a Pareto distribution. --- Sampling strategies and weighting We set these concerns aside now to focus on the issue of sample size that the approaches described below can ameliorate. Despite very large national samples, the decennial census and the ACS have relatively small samples for individual census tracts. This is a problem shared with large-scale health surveys, which despite impressive national sample size have insufficient samples for reliable estimates of characteristics of smaller geographical areas. Statisticians define a "small area" as one where "the domain-specific sample is not large enough to support direct estimates of adequate precision" . Hence, depending on the data source a county or even a state may be "small." Demographers and public officials have become more aware of concerns about the nature of estimates of small area characteristics because of the substitution of the decennial long-form census by the annual American Community Survey . At the level of census tracts and even counties , ACS data are "noisier" than comparable data from 2000 and before . This is largely because the ACS samples are smaller. Tract estimates in the ACS are also affected by the use of population controls from estimates made at the state and county level rather than at the tract level. Starsinic estimated that the standard errors from the fiveyear pooled ACS at the tract level would be about 50% higher than in Census 2000 long form data . The Census Bureau has attempted to deal with ACS's large confidence intervals through changes in the sampling design and through weighting techniques to account for probability of selection, nonresponse, and coverage adjustments . The National Research Council report on these efforts concludes that changes in sampling rates have tended to equalize the precision of estimates across tracts of different population sizes, but at the cost of decreasing the reliability of estimates for larger areas, resulting in minimal net improvement. Complex weighting has another cost: while reducing bias, it increases the variance in sample weights, which in turn increase the margin of error of the final estimates. Hence the weighting procedures can be seen as "an implicit policy statement that unbiased estimates are more important than precise estimates" . When applied to measuring the income distribution within tracts, this general problem is exacerbated by the fact that -aside from sample size -some kinds of places have less precise sample data than others. Specifically, Spielman, Folch and Nagle show that in the 2007-2011 ACS the tracts with the lowest and highest median incomes have larger margins of error than tracts closer to the average income. --- Correcting the bias We now discuss in detail two standard types of measures of income segregation that are susceptible to bias related to sample size and derive potential approaches to correcting them. In a subsequent section we will use 100% microdata from the 1940 census to quantify the We will then apply the corrections to the publicly available income data from the 2000 Census and 2007-2011 ACS to show how bias has affected conclusions about levels and changes in income segregation. Researchers have employed several different measures of income segregation. The simplest is to divide the income distribution into a small number of categories, perhaps three, and to compute a standard segregation index between the bottom and top categories, the rich and poor. This is the approach taken by Massey and Eggers and Massey and Fischer . The simplicity is also a weakness, because such measures do not make use of the full income distribution provided by the census. We focus on two types of measures that do exploit the multiple and ordered category nature of the data. The first is the rank-order information theory index used by Bischoff and Reardon and reported in Table 1. 2 It "compares the variation in family incomes within census tracts to the variation in family incomes in the metropolitan area" , having first recoded incomes into percentile ranks. Two other measures are based on variance. One version is the rank-order variance ratio index R, which like H relies on a rank ordering of incomes and has percentile specific variants such as R10 and R90. The rank-order variance ratio index "can be interpreted as a measure of the average variance of the neighborhood cumulative percentile density function" . 3 Another is the correlation ratio, which Jargowsky refers to as the Neighborhood Sorting Index . The NSI is simply the square root of the between-tract variance in income divided by the total variance of income, a familiar measure in analysis of variance. Like H, it "implicitly controls for the overall income level because it is based on deviations from mean household income 2. Bischoff and Reardon describe H R as follows . "For any given value of p, we can dichotomize the income distribution at p and compute the residential segregation between those with income ranks less than p and those with income ranks greater than or equal to p. Let H denote the value of the traditional information theory index of segregation computed between the two groups so defined. Likewise, let E denote the entropy of the population when divided into these two groups. That is, E = plog 2 1 p + log 2 1 and H = 1 -∑ j t j E j T E , where T is the population of the metropolitan area and tj is the population of neighborhood j. Then the rank-order information theory index can be written as H R = 2ln ∫ 0 1 EHdp Thus, if we computed the segregation between those families above and below each point in the income distribution and averaged these segregation values, weighting the segregation between families with above-median income and below-median income the most, we get the rank-order information theory index." 3. The percentile-p specific variance-ratio index may be defined as R p R = ∑ j M j ∑ k M k 1 M j ∑ i ∈ P j I f ij < p -p 2 /) where I is the indicator function, f jj is the cumulative percentile of income of household i in tract j relative to the city, P j indexes the population of households in tract j, M j =| P j | is the number of households in tract j and J is the number of tracts. Note that the total variation, and thus the denominator in this expression is p because the fraction of households with income less than p in the population is exactly p. Similarly, the rank-order variance ratio index is defined as R R = 6 ∫ p = 0 1 M j ∑ j M j ∑ j 1 M j ∑ i ∈ P j I f ij < p -p 2 dp . and also controls for income inequality because it is expressed as a percentage of total income variance" . We now present alternative approaches to reduce bias for each type of measure. First, we propose a method to estimate the upward bias in entropy-based measures that draws solely on knowledge of the tract-level sample sizes and tract population counts. Then we propose an approach to estimate the variances within tracts using either the original interval-scale measure of family income or a rank-order measure, which can be summed to the city or metro level and used to estimate NSI or R. Neither of these approaches solves the problem of variation in segregation estimates across samples, which is inherently greater when samples are smaller . We will show, however, that they minimize the systematic bias that is found in uncorrected measures. --- Correcting the bias in H We have derived a convenient and feasible approximation of the bias in entropy based measures such as H and H90 related to sample size. The idea is to construct a quadratic Taylor expansion of the entropy function and apply this function to the sample income distributions and actual population counts by tract. This procedure builds on insights by Miller . It yields an estimate of the bias that can be then subtracted from the sample estimate to get an approximate true estimate. It is most useful when sample sizes are not so large that the bias is trivial and not so small that the quadratic function is a poor fit to the entropy distribution over the relevant range. We first note that with independent sampling without replacement the proportion of sampled households, Sj, in a tract with income above some percentile of the population has a hypergeometric distribution with meanPj and variance p j 1 -p j N j M j -N j M j -1 where p j is the proportion in the tract population above this percentile, M j is the tract population, and N j is the number in the sample. The population entropy of this tract is E p j = p j ln p j + 1 -p j ln 1 -p j . A second order approximation to the sample entropy is E s j = E p j + ln p j + ln 1 -p j s j -p j -1 2p j 1 -p j s j -p j 2 + O s j -p j 3 . Taking expectations with respect to s i yields E E s j = E p j - 1 2 1 N j M j -N j M j -1 + O j where Oj is the integral over the order statistic. Interestingly the population variable p j does not appear in this expression except through Oj and thus the approximate bias can be calculated without knowing the true tract population income distribution. For example, the formal expression for the expected approximate bias for H 90 adds up the tract-specific terms: E = H90 -∑ j M j M j -N j 2MEN j M j -1 -∑ j M j O j ME where M = ∑ j M j and E = .1ln + .9ln. In order to assess the accuracy of the approximation we need to be able to compute bounds on Oj. For any given M j and N j the expected difference between the actual and approximate entropy as a function of the number K of households in the population below the specified threshold is O j * = ∑ k = 0 K K k M j -K n j -k M j n j k n j ln k n j + 1 - k n j ln 1 - k n j - K M j ln K M j + K M j -1 ln - K M j + 1 -ln K M j -ln - K M j + 1 k n j - K M j + 1/2 N K k n j - K M j 2 K M j -1 -1 Note that while K is not known in practice it can take on only one of M j +1 possible values, each of which may be checked given M j and N j . The bounds of the set of O j * for K ∈ [0,1,…,M j +1] provide bounds for O j . For M j = 1000 and N j = 50, for example, the maximum of the absolute value of O j is .0065, which applies when there is only one household in the tract population above the percentile cutoff. This is a rare scenario, and the average bias across tracts is likely to be much smaller. Thus our approximation provides a useful basis for estimating bias. It is also worth noting that the adjustment for sampling without replacement will be small if the sample is small relative to the population. For example, for M j = 1000 and N j = 50 the term M j -N j M j -1 = . 95. If we drop this term our approximate bias expression depends only on tract sample size: E E s j = E p j - 1 2 1 N j + O j * . Computation of bias for H is a straightforward extension of the above because the bias term does not depend on the percentile under consideration. In particular, E -H ≈ ∑ j M j M M j -N j N j M j -1 As before, if the sample is small relative to the population this expression reduced to E -H ≈ ∑ j M j M 1 N j So the approximate bias in H is just the average inverted tract population size. --- Sparse-Sampling Variance Decomposition In contrast to the entropy-based measures the level of bias for variance-based measures of income segregation depends not just on sample and population sizes, but also on the distribution of incomes within and across tracts. Unfortunately these distributions are only imperfectly observed through the sample, and the bias correction must account for this fact as well. It was therefore necessary to develop a different approach to bias-correction for these measures than was used for the entropy measures, which we refer to as Sparse-Sampling Variance Decomposition . The idea of this approach is as follows. The variance-based measures decompose total variation into variation within and across census tracts. We presume that the total variance in income can be reliably estimated from the sample data in large cities and metropolitan areas. We then estimate the total within variance, from which the between variance can be deduced. We take advantage of being able to leverage results from a large number of census tracts to average-out errors in the estimation of the distribution within each tract. Although to our knowledge this has not been done before in this context, it is parallel to the use of "small t, large n" panel data methods . We begin by applying SSVD to estimating NSI. The expected total variation of income within tracts for the city is the average of the tract-specific variances weighted by the number of households in the tract. Tract populations are of course known and the variance based on the sample in each tract is an unbiased but noisy estimate of the underlying population variance, even with samples as small as two. But the population weights are uncorrelated with the noise . Hence the populated-weighted average of the variance estimates for each tract from the sample converges to the within variation for the population as the number of tracts gets large. In addition, because the total variance for the population only involves the calculation of a single mean for the city , the per-household population-weighted total variance estimated from the sample is consistent for the corresponding population measure as the number of tracts gets large. The population acrosstract variation is just the total minus the within-tract variation in the population. Thus NSI can be estimated as the square root of 1 minus the ratio of the within to the total variation calculated from the sample using population weights. The bias in the SSVD estimate of the NSI can be shown to approach 0 as the number of tracts increases. Assume there are J tracts and that the income y ij -of household i in tract j is a random variable drawn from a distribution F j with mean μj and variance σ j 2 . Further let the means be drawn from a distribution with mean μ and let the variance V μ be drawn from a distribution with mean σ 2 and variance V σ . Let μ M = ∑ j M j μ j / ∑ j M j , P j denote the set of households in tract j, and S j the set of households in tract j that are in the sample. The NSI for the full population is NSI = A/T = /T = 1 -W /T where A is the across, W the within and T the total variation. Then EW = E ∑ j ∑ i ∈ P j y ij - 1 M j ∑ k ∈ P j y kj 2 M = ∑ j M j σ j 2 M ET = E ∑ j ∑ k ∈ P j y ij - 1 ∑ l M l ∑ l ∑ k ∈ P j y kl 2 M = ∑ j M j μ j -μ M 2 + σ j 2 M Denote the sample analogs of W and T as W = E ∑ j M j N j -1 ∑ i ∈ S j y ij - 1 N j ∑ k ∈ S j y kj 2 ∑ j M T = ∑ j M j N j ∑ i ∈ S j y ij - 1 M ∑ j M j N j ∑ k ∈ S j y kj 2 ∑ j ∑ i ∈ S j M j N j The expected values for the mean in tract j, the overall mean, and the overall variance are provided from sample values corrected for degrees of freedom: E 1 N j ∑ i ∈ S j y ij = μ j , E 1 M ∑ j M j N j ∑ k ∈ S j y kj = μ, and E 1 N j -1 ∑ i ∈ S j y ij - 1 N j ∑ k ∈ S j y kj 2 = σ j 2 . Consequently plim A critique of the NSI is that it is affected by changes in the distribution of income even when the ranking is preserved. Fortunately the SSVD can be applied to variance-based measures calculated from rank-order data such as R -simply transform the income data for the sample into cumulative percentiles and then do a variance decomposition of the resulting percentiles. In particular, for each percentile p one can calculate the fraction of households in each tract below that p and compute the tract-population weighted variance of this measure across tracts relative to the total variance in the sample. This measure for p=90, say R 90 R then, like H90 indicates the extent to which the top 10 percent of the population is segregated from the other 90 percent. To compute R r we average the across and total variation across all percentiles and then divide. Because both R r and R 90 R are based on across-variance estimates the SSVD decomposition follows exactly. Moreover, the approach is computationally efficient because the integration over p needed for R R has an analytic solution and thus no numerical integration is needed. 10 The integration in the SSVD estimate of the rank order estimate conveniently has a closed form solution: R R = 1 -6 ∫ p = 0 1 ∑ j M j M 1 N j -1 ∑ i ∈ S j * 2 dp = 1 -6∑ j M j M N j N j -1 ∑ i ∈ S j k f ij s 2i -1 N j -1 where f ij s is the cumulative percentile of a household in the sampled city population, accounting for any differential sampling weights by tract, and S j * indexes the sample S j from tract j ordered such that if i ≤ i* then f ij s ≤ f ij s ∀i, i* ∈ S j *. The number 6 comes from fact that ∫ 0 1 pdp = 1/6 The second expression, which is obtained by bringing the integral inside of the summations, indicates that the SSVD estimate of the rank order variance ratio is simply a weighted average of the cumulative percentiles. Note that since the sample households are ordered within a tract, the track-specific weights are antisymmetric w = -w around the median ranked household in the tract. --- Validating the bias correction procedures Because the Census and ACS have collected income information only for samples of the population, there is no "true value" for income segregation using contemporary data against which we can compare bias-corrected measures. But using full-count historical microdata from 1940 we can carry out an exercise that is impossible with contemporary data: to draw samples of varying proportions, then to calculate measures of income segregation across census tracts from those samples. Because the "real" level of segregation is known from the 100% data, we can determine how the "observed" level is affected by sampling proportion. In carrying out this test we are mindful of the difference in sampling rates for income between the 2000 Census and the ACS. The 2000 Census long-form data were from a onein-six sample of the population. The NRC calculates a generalized design effect for the 2000 Census of 1.12, representing the degree to which the effective sample size from the Census's design differs from a simple random sample. This reflects, for example, how the Census dealt with overall non-response and the use of population controls in developing weights. Hence we treat Census 2000 as approximately a 15% sample. The actual sampling rate for income may be lower than this, because income is among the variables for which non-response is especially high. What sample proportion does the ACS represent for this purpose? The ACS is not conducted as part of a full census enumeration as the long form surveys in decennial censuses used to be. It utilizes a complex system of sampling and weighting, and it has changed in important ways over time. The sample size increased in 2011. At that time also the Census Bureau increased the differential in sampling proportions between smaller and larger census tracts in order to improve estimates for smaller tracts. One estimate for the 2007-2011 sample is that the median tract sample size was 296 households . After taking into account the generalized design effect of 1.41, the effective sample size for the median tract was only 209 households . These calculations convey the order of magnitude in the ACS's reduction of sample size compared to Census 2000. A more precise calculation can be made from the 2007-2011 ACS summary files for census tracts, which report the sample size in every tract. Across all tracts in the United States, the average final sample proportion was 8.2%. This is reduced to an effective sampling rate of 5.8% after taking into account the design effect but again not considering special concerns of non-response on income data. In the following analyses, we will treat sample proportions of around 5% as representative of the ACS. Spielman, Folch and Nagle cite Census Bureau estimates that imputation rates for income variables approach 20%, suggesting that the actual ACS sampling rate for income data may be less than 5%. --- Results from sample draws from the 1940 census The 1940 census was the first to collect data on income. For each employed household member the enumerator listed the person's wage and salary income. Our analysis is for total household wage and salary income, combining the figures for the household head and all other household members. We did not measure family income because family relationships are not clearly defined in the 1940 data. In addition to income segregation for the total population we compute indices for a subgroup whose population share in 1940 was comparable to the share of black residents in the contemporary data: foreign-born whites . Results for foreign-born white households reveal how observed trends in segregation for specific subgroups may be affected by their smaller numbers. We use data for all households enumerated in the 1940 census in Chicago . We drew samples of 5%, 10%, 15%, and 20%, repeating the procedure 100 times for every level of the sampling proportion and calculating every segregation measure for every draw. 4 Non-residential tracts are omitted; a minimum of one household was sampled in every tract. This procedure yields a sampling distribution of estimated income segregation at each sampling rate. Because all of 4. As reviewers pointed out, the reliability of samples depends directly not on the sampling proportion but on the number of sampled cases, particularly the number of cases in each census tract from which the tract's income distribution is estimated. We repeated our analyses for average sample sizes per tract of 50, 100, 150, 200, and 250. Tracts averaged about 1300 households, so a sample of 50 would be just under 4%, while a sample of 200 be around 15%. Results of these analyses showed the same patterns as did the analyses where we varied sampling rate. these values are from the same population in a single year, any differences between sampling distributions for lower or higher sampling proportions are due solely to varying sample size. At each of four selected sampling rates the plot displays the distribution of values from the 100 samples that were drawn: the maximum and minimum values, the median value, and the values at the 25 th percentile and the 75 th percentile . A straight line across the graph shows the true value of the measure as calculated from the original 100% microdata. The figures are useful as a visualization of the differences across measures, and they are especially helpful in displaying the bias in the uncorrected estimates. An important feature of the values in both figures is the sampling variation. For example, even the best performing estimate for H in Figure 1 has a risk of yielding either too high or too low a value, even though on average -with many samples -it is within .001 of the correct figure of .062. In this exercise we are able to draw many samples and compare them, but in actual research there is typically only one sample. Consequently, researchers should be aware of not only the potential bias in estimates but also their sampling variation, which diminishes with larger samples. We draw attention to this statistical principle because although we show here that it is possible to correct for bias, sampling variation remains a concern with the lower sampling rate of the ACS compared to the decennial census long form that it replaced. Correcting for bias does not reduce the risk of drawing wrong conclusions due to sampling variation. Figure 1 applies the correction calculated from average tract sample size to H, H10, and H90. Note that the median values of the corrected estimates fall within .003 of the actual values even at a 5% sampling rate for each of the three entropy measures. Bias in the uncorrected values, in contrast, is very high with a 5% sampling rate, falling substantially at 10%, and then continuing to improve but still present even at a sampling rate of 20%. The bias correction thus works well, but not perfectly, reflecting the fact that a second order approximation was employed in the derivation of that correction. Figure 2 applies the SSVD to NSI and R. Again there is a spread in all sample estimates that is reduced at higher sampling rates. The uncorrected estimates for R and NSI are high at a 5% or 10% sampling rate, and some bias remains even at 20%. In contrast, there is no bias in the corrected estimates of R at any sampling rate. The correction of NSI greatly reduces the bias in the uncorrected estimate, though at a 5% sampling rate the median estimate of .246 is somewhat above the true value of .237. These results support our initial concern about the upward bias in measures of income segregation. A clear implication is that if income segregation in the population has not changed between time 1 and time 2, but the data by which we measure segregation changes from a higher to a lower sampling rate, we will observe an apparent -but false -increase in segregation. We now turn to estimates for population subgroups where we expect greater impacts of variation in the sampling rate. For this purpose we select households with a foreign-born white head of household . Results are displayed in Figures 34. There is a similar pattern with all five segregation measures. Here we describe results for H in detail to point out the key findings. The actual population value of H for foreign-born white households in Chicago in 1940 was .045, lower than the level of income segregation of the whole population . Yet the uncorrected estimate of H at a 5% sampling rate was .093 for foreign born, higher than the estimated .081 for the total population at the same sampling rate. The greater bias for the foreign-born population at this sampling rate might have led researchers to conclude incorrectly that this minority population was especially highly segregated. Based on the corrected estimates, however, at every sampling rate the foreign-born population is shown to be less segregated. Our proposed sample-size correction methods for entropy measures and SSVD for variancebased measures are shown here to be successful in reducing the bias from varying sampling rates. If we compare the corrected estimates to the true values , we find that it is more effective in some instances than in others. For the total population, the corrections yield median estimates that are close to the true values even at a 5% sampling rate for H, H10, H90, and R, but there is a slight upward bias on the median value of the NSI. For foreign born whites, who represent a smaller population segment and therefore smaller sample sizes, the corrected estimates are excellent for H90 and R, slightly high for H10, and somewhat higher for NSI. Still, the corrected 5% median estimate of NSI is closer to the true value than the uncorrected 20% estimate. --- Applying corrections to contemporary data How does bias from the reduced sampling rate of the ACS affect estimates of change in income segregation? Related to this, how does the smaller sample size for subgroups of the population affect estimates of their segregation in comparison to the whole population ? The techniques demonstrated here can be applied to contemporary data, but with several provisos. Most important, without access to internal census files it is impossible to take into account subgroups living in census tracts where their income distribution is not reported in the public data. As noted above, possibly many affluent blacks or Hispanics live in high-income, predominantly white census tracts but in too small numbers in any given tract to be reported. Possibly also many poor white or Asians live in low-income, predominantly minority census tracts but do not reach the threshold to be reported. To the extent that this occurs -and this cannot be estimated with public data -the observed income segregation of blacks and Hispanics may be skewed downward and that of whites and Asians upward. We cannot deal with this issue here. We can, nevertheless, shed light on the potential of bias due to lower sampling rates in the ACS and smaller samples for population subgroups such as African Americans. For this purpose, the key step is to simulate samples of individual families from the grouped data published by the Census Bureau. The grouped data conceal the exact distribution of incomes of families within each category, and we estimate those distributions by making reasonable assumptions. We describe that procedure and examine its efficacy by reference to the 1940 results, concluding that estimates are robust for H, H10, H90, and R, but not for NSI. We then apply the same procedures to contemporary data for each of the rank-order measures to determine how the bias that we have investigated here affects estimates of trends in income segregation. --- Estimating income segregation from grouped data The challenge in estimating measures of segregation using grouped data is that segregation measures may be sensitive to the distribution of incomes within categories and within tracts. Arbitrarily assigning the mean of the interval to each category may be reasonable for some purposes; however, initial explorations using the 1940 data suggested this approach did not work well for the measurement of income segregation. The problem is especially acute for the top category for which there is in principle no upper bound. The limitations of group-level data have been addressed in previous work using assumptions about the smoothness of income distributions within tracts. Reardon proposes calculating a percentile-specific H for each category's lower or upper bound using the fraction of households in each tract above and below the bound in each tract. He then uses a fourth-order polynomial to interpolate values of H10 and H90 and to integrate over the full distribution of percentiles to obtain H. In constructing estimates of the NSI, Jargowsky makes use of published tract means, which the Census Bureau constructs from the unit-level data, but has to interpolate to estimate the metropolitan area variance. In particular, he assumes a linear distribution of households in lower income categories and a Pareto distribution for income categories above the metropolitan area mean. 5 Integration over these category-wise distributions within tracts can then yield the desired measures. 5. The Pareto Distribution {F = 1 -αY β } describes the distribution of a population with incomes of Y or greater. One way to estimate α and β is by using Quantile Method proposed by Quandt . Choose two probability levels P 1 and P 2 and determine the corresponding quantiles Y 1 and Y 2 from the income category: P 1 = 1 -αY 1 β , P 2 = 1 -αY 2 β , log 1 -P 1 = logα + βlogY 1 , log 1 -P 2 = logα + βlogY 2 , Then: Unfortunately, these methods do not apply in a straightforward way to the construction of bias-corrected methods. There are two issues. First, for the variance-based measures the SSVD requires information on the within-tract variation. While measures of variation by tract are reported in some published tables, and indeed could be provided in most cases without violating confidentiality, we were unable to find consistent measures across the two censuses by race. Second, preliminary explorations suggested that simply subtracting a biascorrection from a standard entropy-based measure that integrates over the estimated distributions or interpolates quantiles may over-correct for sampling bias for relatively small samples. The reason is that the bias corrections assume that the data are sampled but smoothing itself removes some of the sampling variability. To address these problems, we developed an approach based on Jargowsky's procedure. Instead of estimating the distribution only for the overall metropolitan area, we applied the procedure to each tract. Then, to preserve the sampling variability inherent in the original individual data, we sampled incomes from each distribution, with the number of sampled cases per category and tract being equal to the actual number of households. This procedure in effect recreates a unit-level data set from a grouped data set. 6 We expect that estimating tract-specific distributions from grouped data will work if the noise in the estimation process averages out across tracts. Because we cannot demonstrate theoretically that it will, we turned once again to experimenting with the 1940 Chicago case. Starting with the 100% household level income data, we sampled from these data by tract using a specified sampling rate without replacement. We grouped the data into 10 prespecified categories, keeping only the numbers of households by tract and category and the category boundaries. We then used the Jargowsky procedure to estimate the full tract and category-specific distributions based on the grouped data. Finally, we sampled from these β = logN 1 -logN 2 logY 1 -logY 2 , α = 1 -P 1 Y 1 β = N 1 /N Y 1 β . Where N 1 and N 2 are the number of household whose income is at least greater than Y 1 , Y 2 , respectively. N is the total number of households among all the tracts. Following Jargowsky we use the Pareto Distribution to estimate the variance in each category: ∑ i = 1 n b δ 2 = ∫ y 1 y 2 2 fdy 6. We also implemented Reardon's approach of estimating H using a fourth order polynomial to the metro-level entropy function, which was explicitly designed for group level data and thus does not require second-stage sampling. In Figures 56, Tables 23, and Appendix Table 1 this estimate of H from grouped data is identified as H-R. We include these results to show that our correction procedure improves the estimates for either approach. The uncorrected H-R estimates are better than the uncorrected H estimates, and we attribute this to its having smoothed out some sampling fluctuations of values within categories. Because we cannot use Reardon's approach for the variance-based measures, we focus here on results using our modification of Jargowsky's method. estimated distributions with the number of households in each category and tract equal to the number retained in the simulated grouped data. The relevant algorithms were then applied to construct biased and bias-corrected estimates of the relevant measures. The second-stage sampling was repeated 100 times and the results were averaged. The whole procedure, starting with the 100% household level income, was then repeated 100 times for each specified sampling rate. Figures 5 and6 plot the estimates of income segregation parallel to results in Figures 12but now calculated from the grouped data . True values are from the 100% individual data. Results of the correction are excellent for the entropy measures H, H10 and for R. H90 is a bit off, but the corrected measures are still a substantial improvement over the uncorrected estimates even at a 20% sample. It is noteworthy that in Figures 5 and6, even when for H90) the intervals of the estimates do not overlap the true value they tend to be equal across sampling values. This relationship is a useful feature as it suggests that our correction procedure does succeed in removing differences that are attributed to changing sample sizes even when the grouping procedure results in some loss of information. Indeed, the process of grouping the data can yield bias even with a 100% sample as a consequence of the interpolation process. This pattern is especially evident with respect to H-R, for which the corrected estimates are identical at .060 across all sampling rates for the grouped data. The comparable figure for the individual-level data, however, is .062. Analogously the gap between the group-based estimates of H90 and the true value in Figure 6 is attributable to the fact that even with a 100% sample, the grouped data yield an estimate of 0.063 as opposed to 0.059 for the individual data. Further, the procedures using grouped data yield uncorrected and corrected estimates of NSI that are very different from and further from the true value than those calculated directly from the individual household data. At a 20% sampling rate, the corrected NSI virtually matched the true value when estimated from individual data for both the total population and the foreign born whites. But at this sampling rate, the corrected NSI from grouped data is too low for the total population and too high for foreign born whites. While even with grouped data the estimates are relatively invariant with respect to the sample size when SSVD is applied, we are not satisfied with the level of precision. The difficulty with the NSI may not be correctable, because it is due to the combination of the great sensitivity of NSI to the highest income values and our inability to reproduce the original distribution of these values from the grouped data. Consequently, we will not offer estimates of contemporary changes in NSI here, deferring to a future time when the internal census files may be available. --- Correcting measures of contemporary income segregation We now turn to an effort to estimate income segregation in 2000 and 2007-2011. As already noted, these estimates are not "true" because they are distorted by working with grouped data and they do not compensate for suppression of income data for blacks in census tracts with small black populations. However they do remove the bias due to smaller sample sizes in 20072011 and for blacks compared to the total population. To maximize comparability with results previously reported by Bischoff and Reardon we use family income data at the tract level from Census 2000 and ACS 2007-2011. Income segregation for the total population is for 118 metropolitan divisions or statistical areas with population above 500,000 in the 2007-2011 data; segregation for black families is measured for a subset of 66 metros with more than 30,000 black families. Metropolitan area boundaries are fixed by the definitions used in Census 2010 for primary MSAs. We applied the method described above with the 1940 data and summarize results in two tables, Table 2 for the total population and Table 3 for families with black household heads. Table 2 lists values for the whole population for the same six metropolitan regions that were presented in Table 1 as well as the mean and standard deviation for large metropolitan regions. Again values are scaled by 100 and can be interpreted as percentage points. Our uncorrected figures differ from those reported by Bischoff and Reardon ; theirs are generally larger because they imposed a cap on sample size whose effect was greater for larger regions . For every measure and at both time points the mean is lower for uncorrected measures than for those that have been adjusted. The adjustments also result in changes in the trend over time. Averaging across cities, an increase of H of 1.8 points becomes an increase of 1.2 points; the corrections reduce the apparent gain in H in all six metros listed in the table. The increase in the mean H10 of 2.0 points reduces to an increase of only 1.1 point . The average increase of H90 for all cities was 2.6 points without the correction but only 1.7 points after correction. Similarly, R rose by 1.9 points before correction, and by 1.3 points after correction. Did income segregation increase for the total population? Pending a more comprehensive analysis that may be completed in the future with internal census files and acknowledging variability in patterns across metropolitan regions, we conclude that it did increase, but that about a third of the previously reported increase is attributable to the lower sampling rate in the 2007-2011 ACS. We turn now to income segregation among black families. We have emphasized that the upward bias from sample data is more pronounced for estimates of segregation within subgroups, so we expect corrections to be greater for black families. This bias predates the ACS, because the tract-level sample sizes for the black income distribution have always necessarily been smaller than for the total population. Hence several studies over the years have reported that segregation among blacks is greater than in the general population. Bischoff and Reardon draw particularly strong conclusions: "The trends in income segregation among black and Hispanic families are much more striking than those among white families… By 2009, income segregation among black families was 65 percent greater than among white families. [.] Indeed, in the nine years from 2000 to 2009, income segregation among black families grew by almost two standard deviations." These conclusions are muted or contradicted by the corrected estimates for black families in Table 3. First, is segregation higher among black families? Compare the mean values for 2007-2011 in Table 2 and Table 3 . On two measures blacks are more segregated by income, but on the other two blacks are less segregated. Further, on the two measures showing blacks to be more segregated, the differences after correction are much smaller than before correction. In 2007-2011 H was higher for blacks by 1.1 , and H10 was higher by 0.5 . Did black income segregation increase and did it increase more than in the total population? Results from uncorrected measures fit readily into a common narrative about race and concentrated poverty, according to which the rapid growth of income segregation among black families has accentuated the isolation of the black poor. Uncorrected measures show that H increased by 1.8 in the total population and by much more -4.6 -for black families; H10 increased by 2.0 in the total population and 5.8 among blacks; H90 by 2.6 and 6.3, respectively; and R by 1.9 and 4.6. Although on every corrected measure the increase among blacks was greater than in the total population, the disparity is considerably smaller after correction. There are also important differences in the comparisons across the six specific large cities reported in Table 3. For example, the uncorrected estimates for R show increases in income segregation among blacks in every city but Detroit. However, the corrected estimates show declines in three of the cities but increases in the other three . What seemed a simple set of results from the uncorrected measures is more nuanced when we adjust for sampling issues. --- Conclusion It is plausible that income segregation has increased since 2000 along with the rising level of overall income inequality in the United States. Income segregation may also be higher and increasing faster for minorities than for the general population. We will not know for sure until corrected measures of income segregation are estimated from the original individuallevel sample data without suppression, but certainly past research based on the 2000 Census and 20072011 ACS was systematically biased to show an upward trend. Estimates using methods of correction for this bias suggest that income segregation did increase for the total population but not by as much as has been previously reported. Corrected estimates suggest that income segregation may have increased faster among blacks, but there are mixed results on whether segregation among black families was higher than among the total population in 2007-2011. This research is immediately consequential our understanding of patterns and trends in income segregation. More generally these findings add to a growing awareness of the relevance of sampling issues to research based on the population census and ACS. Statisticians have always been aware of these issues, but in the past most researchers took the reliability and precision of census data for granted. The one-in-six long form data from 2000 and earlier were routinely treated as though they fully captured population characteristics. We are now in a better position to evaluate how this approach can lead researchers to wrong conclusions, thanks to the 100% census samples that are now being made available by the Minnesota Population Center. The 1940 data are especially useful because they include a more complete set of socioeconomic indicators . By drawing repeated samples from these full population data, researchers can now readily study the sampling variability of any parameter of interest. We have exploited this opportunity to examine the behavior of measures of income segregation. We have shown that there can be considerable bias even in a 15% similar to the long-form data collected in decennial censuses up to 2000. Now the Bureau of the Census has shifted to smaller annual samples in the American Community Survey that are no larger than 5% even when data are pooled over five years, and measures of income segregation have greater upward bias. The problem is necessarily accentuated when dealing with subgroups of the population. We have given attention here to the foreign-born white population in 1940 and black families in the last decade. The same issues may be relevant for other subgroups, such as families with children . Fortunately sampling bias in these measures can be corrected. Results from repeated sample draws from the 100% data in 1940 give a high level of confidence in the efficacy of a simple population-and sample size-correction for entropy-based measures, and the SSVD estimation of variance-based measures. The remaining obstacles can be overcome by access to confidential census data centers, where researchers can study the original individual-level sample data without suppression. Despite the reduced sample size of the ACS, it will be possible in this way to draw clear conclusions about changes in income segregation in the country as a whole, in different regions, different parts of the income distribution, and different categories of people by race, family composition, or age. A closely related question for further research is whether other measures of spatial inequality, such as traditional measures of racial/ethnic segregation, are affected by sampling in the same way. Fortunately some data are collected from full population samples, including even the counts of Asians and Hispanics of different national origins. Studies of segregation of some other potentially important categories of people, though, rely on sample data. These include distinctions within the black population between people of recent African origin, Afro-Caribbeans, and African Americans with a longer history in the U.S. Immigration researchers typically wish to distinguish people not only by race/ethnicity but also by nativity, and the Census Bureau provides the necessary tabulations by census tract from sample data. As noted above, some researchers wish to distinguish between family households with and without children. Studies of this type are being conducted, and it is important to know whether the usual indices of segregation are subject to the same kind of bias as the income segregation measures evaluated here. Appendix I: Two-stage resampling as a possible solution to bias We use 1940 census data here to test a resampling procedure developed by Bischoff and Reardon to correct for differences in sampling rates of H in the 2000 and 2010 census using aggregate tract level data. The idea of the resampling approach is that if similarly sized small samples are drawn based on observed distributions for the two censuses then small-sample biases, while present, will be comparable across time and thus trends will be correctly estimated. The procedure, as explained in a private communication from its authors and similar in concept to their handling of data from 2000 in a previous study , is that a simulated micro data set is created with the number of households in each category of income by tract equal to the number implied by the published distributions. Then 100 samples of 50*N, where N is the number of tracts, are drawn from the simulated micro data for each year. H is calculated for each sample using grouped data procedure for which Bischoff and Reardon provided Stata code. The procedure involves fitting 4-th order polynomials to the cumulative distributions and then integrating over this distribution to H. The 100 sample Hs are then averaged to get a final H estimate for each year. We carried out a two stage procedure to mimic the data generating process underlying this procedure, and we applied it to H, H90, and NSI with similar results for each measure. Here we display the results for H. In the first stage we constructed 100 5% and 15% samples and one 100% sample of the 1940 micro data and then categorized the data by tract to create the equivalent of the published tables for each sample. In a second stage we carried out the proposed resampling procedure for each first-stage sample. We then plotted the difference between the estimate and a "true" estimate based on the 100% sample without any second stage sampling. A box plot of the resulting differences by original sampling rate is presented in Appendix Figure 1 and labeled "N*50 Cen". We also carried out a variant in which the microdata consisted of the number of households implied by the census tract sample sizes rather than the census tract population counts . A third variant was constructed in which there was no second-stage sampling. It is evident from Appendix Figure 1 that despite the fact that the second stage sample is of the same size for the 5 and 15 percent samples, the Bischoff and Reardon procedure does not yield similar biases for them. The average difference is .009 or 16% of the true value of .056. This difference is roughly equal to the difference of .01 when no second stage is sampled at all . In short, the "N*50 Cen" approach increases the bias by roughly the same amount for the 5 and 15 percent samples and thus does not undo the difference in bias that is created from the different sampling rates used in the first stage. The bias estimates for "N*50 Samp" are more comparable across samples. This suggests that the Bischoff and Reardon procedure would work better if it were based on the counts of the sample rather than the counts of the population that are estimated from the sample. In effect this works because it approximates what you would get by randomly discarding 2/3 of the 15% sample so that sampling rates are in fact the same. There are two drawbacks with this approach. First, in contrast to the "N*50 Cen" procedure, the "N*50 Samp" procedure will not work to compute segregation among subgroup populations for which 5% of the population per tract on average is less than 50. Second, bias depends on the within tract distributions as well as on sampling size, so there is no guarantee that bias will be the same in different cities with different levels of segregation. Simulations available from the authors on request suggest that this latter source of bias is not large for H, but it can be important for other measures such as H10 and H90. --- Appendix Figure 1: Bias from resampling procedures based on 1940 Chicago Census Data

Recent studies have reported a reversal of an earlier trend in income segregation in metropolitan regions, from a decline in the 1990s to an increase in the 2000-2010 decade. This finding reinforces concerns about the growing overall income inequality in the U.S. since the 1970s. Yet the evidence may be systematically biased to show an upward trend because the effective sample for the American Community Survey (ACS) is much smaller than it was for Census 2000, to which it is being compared. There is a possibility that the apparent changes in disparities across census tracts result partly from a higher level of sampling variation and bias due to the smaller sample. This study uses 100% microdata from the 1940 census to simulate the impact of different sampling rates on estimates of several measures of segregation and to propose and test the effectiveness of approaches to correcting the bias. It then applies those approaches to publicly available data for 2000 and 2007-2011. The reduction in sample sizes associated with the ACS results in exaggeration of evidence for increasing income segregation for all measures tested here, especially for subgroups (African Americans are studied here as an example). The methods of correction applied here will yield more conclusive and unbiased results when applied to the original sample data that is held internally by the Census Bureau.Because neighborhoods are so consequential in people's lives and futures (Sampson 2012), urban social scientists have long been interested in neighborhood-level segregation. Although most of this literature focuses on separation by race and ethnicity, attention has also been given to segregation by social class or income. An early version of this research was devoted specifically to "underclass" neighborhoods, areas with high levels of poverty,

Background Global under-five deaths have decreased from 12.6 million in 1990 to 5.4 million in 2017 [1]. After Sub-Saharan Africa, South East Asia has the highest burden of under-five deaths. Diarrhoea and pneumonia are still the major killer diseases among the under-5 year children globally and nationally [2,3]. Acute Respiratory Infection and associated fever, dehydration from diarrhoea and malnutrition are the important contributing causes of childhood morbidity and mortality in developing countries. Most child deaths are caused by diseases that are readily preventable or treatable with proven, cost-effective and qualitydelivered interventions [2]. In developing countries, health system suffers from a chronic shortage of human resources. In this context, the mobilization of community health workers or volunteers has been seen as a cost-effective and promising health system intervention [4]. Nepal started the Female Community Health Volunteers program in 1988. This program was designed to enhance Nepal's primary health care service delivery network, improve community participation, and expand the outreach of health services in doorsteps. The program was established in all 75 districts of the country by 1995 [5]. Nepal initiated community-based diarrhoeal disease control program in 1982 involving local health workers to promote home-based oral rehydration therapy and case management at the health facilities. This intervention, which also involved FCHVs after their introduction in the health system, had an important role in reducing child morbidity and mortality attributed by diarrhoea [6]. In 1997, an evaluation of community-based pneumonia case management supported by the World Health Organization further paved the way to involve the FCHVs to treat the pneumonia cases [7]. Given that the percentage of diarrheal cases treated with Oral Rehydration Salt and zinc by FCHVs among the total diarrhoeal cases reported throughout the county was 64% in 2014/15 and 65% in 2015/16, they are an important cadre to curb child deaths and sickness [3]. Recognizing the importance of community-based interventions, the Government of Nepal initiated community based-integrated management of childhood illness program in 1999 to address major childhood killer diseases . This intervention involved FCHVs to recognize and treat pneumonia and diarrhoea in children under 5 years of age, and refer the sick neonates and young infants with any danger signs including malnutrition [6]. The role of FCHVs was crucial to achieving Millennium Development Goal 4 targets in Nepal [8,9] which was also exemplary to the rest of the world. However, evidence regarding the FCHV program in Nepal [10][11][12] has shown that the FCHV program has been inconsistent in delivering high-quality community health services across the nation. Given their contribution to bridging the gap between health facility and community, it is necessary to motivate and capacitate this workforce for providing community health services including child health. To do so, it is important to understand the factors that are affecting FCHVs performance. However, very few studies have explored the factors affecting their performance. Previous studies done in small settings [10,11] suggest that local leadership, work burden, geographical difficulties, inadequate training, limited monitoring and evaluation and incentive issues affect the service delivery by FCHVs. In 2014, the Ministry of Health conducted a national survey on FCHVs [13] for assessing the performance of the FCHV program. In this context, this study aimed to identify child health services provided by FCHVs and determine the factors associated with it using the data from the national survey. The findings of this study will serve as evidence for the policymakers and program managers to focus on management aspects of FCHVs to improve the child health status of the country. --- Methods --- Study design and sampling population Data for the study was obtained from a nationally representative FCHV Survey conducted in 2014 [13]. The survey was a cross-sectional study based on cluster design. For the study purpose, the country was divided into 13 domains considering ecological and developmental regions of the country. These 13 domains included 36,050 wards. Wards were the primary sampling unit. In each domain, the sampling frame of wards was created and wards were selected randomly to get sample size per domain. In total 257 urban and 4045 rural wards were selected for the sample. The details of the sampling procedure have been reported in the survey report [10]. The study included 4302 FCHVs. They were interviewed using structured questionnaires which can be accessed in survey report [13]. These questionnaires were finalized based on the feedback received through pretest and experts. Trained enumerators were mobilized with a window-based tablet to collect the data. Information related to service delivery was obtained through review of the FCHV service register. A mobile platform namely "Enketo" linked with SurveyCTO was used to collect and analyze data. Detailed data collection methods have been discussed in the survey report [10]. --- Variables --- Outcome variable The outcome variable in this study was the child health service provided by the FCHVs. It was defined as any one of the five services recorded in the service register of the FCHVs. These five services included: distribution of ORS packets to diarrhoea cases, distribution of Zinc tablet to a child suffering from diarrhoea, examine the child for ARI, distribution of cotrimoxazole pediatric tablet to a child suffering from pneumonia, and counselling or referral service to the malnourished child. --- Independent variables The independent variables included characteristics related to socio-demography, work profile of FCHVs, supports received by FCHVs, governance of the FCHV program, availability of commodities and service delivery by FCHVs. Socio-demographic variables included the age, caste, educational status, residence and province. In the survey, FCHVs were labelled as "literate" if they had an education level of sixth grade and above, or if those with less than that level of schooling but could fully or partially read a simple sentence written in the card. The work profile of the FCHVs included time taken to reach the health facility, use of mobile phone and years of work experience. The time taken to reach health facility was grouped in three categories and years of work experience was also grouped into three categories . Support received by FCHVs included information about whether they have received basic training , dress allowance , incentive received and use of FCHV fund . Information on the availability of key commodities included the availability of ORS packets and zinc tablet to treat diarrhoea and Cotrimoxazole tablet to treat pneumonia. Service delivery by the FCHVs included information on the involvement of FCHVs in antenatal care services: counseling to pregnant women, involved in testing pregnancy or providing misoprostol for preventing postpartum haemorrhage in case of home delivery , conducted health mothers' group meeting , supported in immunization clinics and supported in conduction of Primary Health Care-Out Reach Clinics . Governance part of the FCHV program included information about affiliation with FCHVs right based organization and involvement with other local committees . --- Statistical analyses Descriptive analysis was performed and presented as frequencies and percentages. Then, the bivariate analysis was performed to examine the association between dependent and independent variables. The multi-collinearity analysis was performed and variables which had correlation value less than + 0.5 were considered in the regression analysis. Multiple logistic regressions were employed to determine which variables could best explain the child health service delivery by FCHVs. Adjusted Odds Ratios were presented with their 95% confidence intervals and p-value of less than 0.05 was considered as statistical significance. Sampling weight was used to adjust the sampling distribution using "svy" command in Stata 13 [14]. --- Ethics The ethical approval for this study was obtained from the Nepal Health Research Council which is the government body that approves human researches and monitors such studies within the country. Written informed consent was taken from the study participants before the data collection and personal identifiers were removed during the data analysis. --- Results --- Services provided by FCHVs Overall, 62.6% of the FCHVs had provided at least one child health service within the last 3 months before the survey. Among the five child health services, 51.6% of the FCHVs had treated at least one case of diarrhoea with ORS in the last 3 months, which was the highest among all the services provided and 9.3% had counselled or referred the malnourished child which was the lowest among all the services provided . --- Relationship of child health service delivery with the socio-demographic profile of FCHVs In the bivariate analysis, child health service delivery by FCHVs was associated with their literacy status, ethnic group, residence and province. FCHV's age was not associated with child health service delivery . --- Relationship of child health service delivery with other background characteristics In the bivariate analysis, child health service delivery was associated with the use of mobile phone, time to reach the nearest public health facility, training received, dress within the last 3 months had significantly higher odds of providing child health services than those who did not. The FCHVs who utilized money within last 1 year from the FCHV fund were more likely to provide child health services compared to those who did not use the money from the fund. Similarly, FCHVs who used a mobile phone had higher odds to provide child health services as compared to those who did not use the mobile phone. Likewise, FCHVs who were involved in different local committees . were more likely to provide the child health services than those who were not involved in such committees. In the adjusted analysis, literacy, residence, time to reach the nearest health facility, training received and dress allowance received were not significantly associated with the child health service delivery provided by FCHVs . --- Discussion The sharp decline in child mortality in Nepal after the 1980s has been attributed to inexpensive communitybased interventions [7,[15][16][17] led by local health workers and FCHVs. Studies around the world have shown that community-based interventions with minimal training to community health volunteers and ensuring the availability of life-saving key commodities contributes to the decline in child mortality [18][19][20][21]. Our study findings demonstrate that nearly two out of three FCHVs were involved in providing at least one child health service in the last 3 months before the survey. There is no doubt about the importance of FCHVs in rendering child health services. For instance, 67% of all the diarrhoea cases in the country were reported by FCHVs in fiscal year 2017/18 [22]. However, it is also relevant to understand the reasons for not providing child health services by the remaining FCHVs. This In our study, availability of key commodities with FCHVs was significantly associated with providing child health services. The limited availability of commodities severely restricts FCHVs' ability to provide services and affects the community trust on them. The major reason for this is due to the inadequate stock of drugs in health facilities [23] which is again the consequence of procurement and transportation hurdles. Reducing commodity stock-out rates across health facilities in Nepal and equipping FCHVs with these commodities could thus reasonably be assumed to contribute to improved service delivery and child health outcomes. Though the importance of local health workers cannot be ignored, mobilization of FCHVs has supported in bringing services closer to the community in a country which has been suffering from a chronic shortage of human resources for health. A study from Nepal [12] has indicated that women did not prefer to contact FCHVs during the illness of their child because of their incompetency and lack of medicines. Thus, for an effective FCHV program, their mobilization needs to be continuously monitored and supervised by local health workers with regular competency-based training and sufficient supply of logistics. The Government of Nepal has created an FCHV funda micro-credit fund, which is managed by FCHVs. They use this fund in income-generating activities. Our study revealed that the utilization of money from the fund was positively associated with child health service delivery by FCHVs. The fund might have strengthened their economic status, increased the sense of belongingness and improved their performance which is also supported by a qualitative study done in Nepal [10]. The FCHVs who supported PHC-ORC and ANC related activities were more likely to provide child health services as compared to those who did not support these activities. In these activities, the FCHVs are involved in educating and counseling couples on family planning; distributing commodities; promotingANC, institutional delivery, and postnatal care for mother and newborn [22]. Involving FCHVs in these wide ranges of interventions would aid in the integration of maternal, child and newborn services thus leading to the continuum of care and better outcomes in child survival [21,24]. It is however equally important to consider the work burden of FCHVs while task shifting and also capacitate FCHVs to deal with cultural and religious issues that surround during pregnancy and childbirth in Nepal as identified in other studies [25,26]. Our study showed that the participation of FCHVs in health mother's group meeting was associated with the delivery of child health services. These meetings are unique platforms to discuss different health issues and are attended by local women. Studies from Nepal have shown that frequent interactions between mothers and FCHVs were related to the use of child health services from FCHVs [12] and reduction in underweight and stunting status among children [27]. Similarly, a study from Makwanpur, Nepal [28] and Jharkhand and Orissa of India [29] had shown that participatory intervention involving women's group can decrease both maternal and neonatal mortality and improve service utilization. Our study findings showed that the use of the mobile phone by FCHVs was associated with child health service delivery indicating that a mobile phone could play a potential role to improve child health outcomes. The use of mobile phone and wireless technology has a huge potential to improve the health and wellbeing of the resource constraint communities through communication and exchange of skills among health workers and with communities [30,31]. The use of the mobile application to improve maternal and neonatal health outcomes has been studied among FCHVs in Baglung and Dhanusha districts of Nepal which also has shown promising results [32][33][34]. This might be because FCHVs are likely to be contacted through mobile phone in case of emergency or other health needs by the communities leading to increased service utilization of child health services [33]. Our study findings demonstrated that FCHVs who were involved in the local level committees were more likely to provide child health services than those who were not involved in similar committees. RB Khatri, SR Mishra and V Khanal [35] however are of the view that the involvement of FCHVs in other non-health programs such as forest user groups, community development groups, education, and microcredit and saving groups could compromise their working hours in the health sector leading to poor performance. We put forward that such involvement may open opportunities for social networking leading to higher self-esteem and Our study findings indicate that incentives do not affect the delivery of child health services by FCHVs. This might be because of their volunteering role and that FCHVs are motivated by social recognition which is also supported by a qualitative study done in Nepal [36]. Studies have however shown that the issue of fair compensation for FCHVs needs to be addressed [10,36,37] as economic insecurity is a strong barrier to volunteering. Motivational activities like the provision of training and dress allowance for FCHVs were not significantly associated with the delivery of child health services in this study. The reason might be that FCHVs had joined volunteerism with the least expectation. In our study, education of FCHVs was not significantly associated with the child health service delivery. This finding contrasts with a study from Dhanusha, Nepal [38] where the educational status of FCHVs was associated with their knowledge and performance on maternal and child health services. Nepal is in the early stage of implementation of federalism and the constitution has identified health as a fundamental right. While the federal government is responsible for overseeing broader policy, the province and local governments are responsible for the management of the health services [39,40]. In the changed federal governance, the task of managing FCHVs comes under the direct responsibility of the local level governments who were previously managed by local health posts and primary health centres. The role of local-level governments would thus be crucial to motivate these cadres for contributing to the health of the communities. The FCHVs are complimentary cadres to improve the health of the communities and their effort alone might not be sufficient. It is also necessary that there is increased community demand for health care and the availability of quality health services at health institutions. This study is based on a large sample size and the findings can be generalizable to the entire country. However, there are inherent limitations to the study. Firstly, its cross-sectional study design does not allow establishing causation. Secondly, this study only focuses on the FCHV's perspective and thus there is an indication of further research to understand the community and service provider's side perspectives. Furthermore, this study doesn't identify the main reasons for not providing or providing child health services. Future research may be needed to explore these factors in detail. Despite these limitations, this study provides useful evidence for the policymakers and program managers for understanding the factors influencing the performance of FCHVs in providing child health services. The evidence can be utilized for efficient and effective utilization of these cadres for improved child survival. --- Conclusions This study revealed that availability of the essential commodities , use of mobile phone, conduction of health mothers' group meeting, involvement in local level committees and support in other health programs and utilization of FCHV fund were the determinants of child health service delivery by FCHVs in Nepal. Ensuring a sustainable means to supply these essential commodities and encouraging the involvement of FCHVs in communitybased health services through adequate support system will help to improve child health status in Nepal. The existing FCHV program can benefit from this evidence by motivating these cadres for strengthening child health services at the community level. --- --- Abbreviations AOR: Adjusted Odds Ratio; ARI: Acute Respiratory Infection; CB-IMCI: Community Based Management of Childhood Illness; CI: Confidence Interval; FCHVs: Female Community Health Volunteers; ORS: Oral Rehydration Salt; PHC-ORC: Primary Health Care Out Reach Clinic; WHO: World Health Organization Authors' contributions HKB and LD conceptualized the study and monitored the survey. HKB, LD and SS performed data analysis. KR was the principal investigator of the survey and supported to write the discussion. VK and PK interpreted the results and wrote the discussion section. NRP supervised the study. All authors reviewed the final draft of the manuscript. --- --- --- Competing interests The authors declare that they have no competing interests. ---

Background: Nepal has made a significant improvement in child survival in the last few decades and the involvement of female community health volunteers (FCHVs) has been crucial in such achievement. While there have been many studies on child health in Nepal however, rarely explored the status and factors associated with the child health service provided by these volunteers. This study aimed to identify the factors associated with the child health service delivery by FCHVs. Methods: A national survey was conducted in 2014 in Nepal that included 4302 FCHVs using the structured questionnaire across the 13 geopolitical domains of the country. Factors associated with the use of child health services was examined using Chi-square test (χ 2 ) followed by logistic regression. Results: Overall, 62.6% of FCHVs provided at least one child health service. Those FCHVs who utilized money from the FCHV fund, conducted health mothers' group meeting, involved in local committees and those who supported antenatal care and outreach clinics related activities had higher odds of providing child health services. Similarly, FCHVs equipped with the stock of Cotrimoxazole tablet, Zinc tablet, Oral Rehydration Salt packets were more likely to provide child health services. The province-wise analysis showed that FCHVs from Province 5 and Sudur Paschim Province were more likely to provide child health services compared to their counterparts from province 1. Technologywise, FCHVs who were using mobile were more likely to provide child health services. Conclusions: FCHVs are important human resource in providing child health services in Nepal. To improve child health service delivery by FCHVs; availability of key commodities, involvement of FCHVs in regular health mothers' group meeting, use of mobile phone, involvement in other public health programs and social networks, and utilization of the FCHV fund need to be taken into consideration.

Introduction Severe acute respiratory syndrome 2 , which is the virus responsible for coronavirus disease 2019 , is a public health issue of the highest magnitude, affecting diverse populations around the globe. To date, there have been over 95 million positive cases and approximately 2.06 million confirmed deaths worldwide , and 24.1 million positive cases and over 400,300 confirmed deaths in the U.S. . Notably, there have been significant health disparities in COVID-19 morbidity and mortality that reflect social determinants of health, biases in our medical system, and interacting forms of social marginalization leading to a higher reported number of cases among racial/ ethnic minorities and communities with fewer resources or are otherwise marginalized including people who report substance use . These disparities are likely due to population density of these neighborhoods, lower access to primary care, and social disadvantages that make it more challenging to engage in social distancing. Social inequities have exacerbated these disparities, as these communities are also more likely to experience job loss and resulting unemployment , grief associated with loss due to the virus, and associated emotional impacts including shock, disbelief, anger, and sadness . There have also been increased mental health and psychosocial problems resulting from loss of employment, other personal loss, and restricted access to social support . Among people who use substances, there have been unique challenges posed by the pandemic including access to safe drug supply, clean needles and harm reduction supplies, and opioid substitution treatment . One subpopulation of individuals at high risk of poor COVID-19 health outcomes, but rarely acknowledged, includes people who trade sex for money, goods, or other services . Broadly, this includes individuals who engage in commercial sex work, transactional or exchange sex, and survival sex, who will be referred to as "sex workers" and/or "street-based sex workers" throughout the rest of the manuscript. Syndemics theory recognizes that "epidemics" create a "synergy" that places individuals at risk of multiple health and social conditions . Applied to the current population and COVID-19, syndemics theory can be used as a lens to understand the multiple forms of structural and social disadvantage that affect street-based sex workers and their risk of COVID-19 exposure and acquisition. According to syndemics theory, social conditions create a set of circumstances that make individuals more vulnerable to various health conditions and worse health behaviors. Syndemics theory posits that co-occurring epidemics work synergistically to place marginalized individuals at risk for multiple health and social conditions. With COVID-19, we have already seen the ways in which the disease has traveled from those with relative privilege to those with relative disadvantage . Individuals with multiple marginalized identities are more likely to experience a range of health conditions resulting from structural disadvantage and discrimination including worse mental health and quality of life . Sex workers represent a vulnerable population for myriad health conditions with notable systemic inequities which perpetuate disadvantage, including for COVID-19 . Interestingly, societal patterns emerging throughout the world suggest that COVID-19 may not be just pandemic, but rather a syndemic as well, as it is interacting with social forces and systematically causing more harm to those with less advantage . Several commentaries have attempted to raise awareness around the potential impact of COVID-19 on street-based sex workers; however, the impact of COVID-19 on this population is still unknown . With respect to other conditions of public health significance, street-based sex workers have higher rates of HIV/STIs, substance use disorders, and mental health conditions . This population also experiences a disproportionate number of social factors that affect health including trauma, poverty, homelessness, and substance use . For example, preliminary data from the state of Rhode Island and the country show increases in substance use severity across all substances and increases in overdoses and overdose-related deaths as a result of COVID-19 ; and, anecdotally, this has directly impacted the lives of sex workers since the start of COVID-19. These health disparities are synergistic and best understood through a syndemics lens whereby street-based sex workers are more vulnerable to these conditions as a result of societal marginalization, lack of access to health care, and stigma. Recent publications have documented the impact of COVID-19 on vulnerable populations including ethnic/ racial minorities , persons experiencing homelessness , and individuals who use substances ; however, few have been published on individuals who engage in exchange sex and sex work specifically. This population is a high-needs demographic that is overrepresented in homelessness and minority populations . A few studies have examined the impacts of COVID-19 on sex work internationally ; however, to our knowledge, there has not been a COVID-19 paper published on individuals who engage in exchange sex or sex work in the U.S. despite the fact that this population often experiences multiple co-occurring conditions, including trauma, poverty, homelessness, and substance use. To address this gap, the current study examined the impact of COVID-19 on sexual behavior, substance use, access to services, and quality of life for sex workers in a small U.S. city. --- Method --- --- Procedure Data were collected between April and May 2020 by CBO staff. Peer specialists explained the risks and benefits of participating, the voluntariness of the survey, and that data would not be linked back to anyone personally because it was being collected without their name or other identifying information. Peer specialists also informed individuals completing the survey that information would be shared with researchers and other stakeholders in the community who may be able to raise awareness about some of these issues. All participants verbally consented to complete the survey. Surveys did not include identifying information or protected health information. After obtaining consent, staff administered the survey orally to participants in English. Surveys were administered either in-person employing social distancing and infection control measures in the community during outreach events, indoors at the CBO location, outdoors on the CBO property, or over the telephone. A total of 46 surveys were analyzed. Participants were compensated with a five dollar gift card for a local convenience store chain or fast food restaurant. The Miriam Hospital Institutional Review Board provided a waiver of consent and approved the study methods to retrospectively review and analyze deidentified data for research purposes. --- Measures Surveys included questions about demographic information , as well as questions about frequency of substance use and sexual behavior in the past 30 days and whether there were changes due to the COVID-19 pandemic. Participants were also asked if they were using personal protective equipment to protect themselves and/or their clients while participating in sex work, such as use of face masks or gloves. Participants were asked open-ended questions verbally including "How has COVID-19 changed how you have sex or otherwise engage with clients?" and "Is there anything else you would like to share with us about how your life has changed as a result of COVID-19? This can include changes in access to services, jobs, housing, food, or other needs; changes in your relationships; and, changes in your own mood or thinking. Please feel free to share openly about how your life has been affected." --- Data Analysis Plan Quantitative data analysis. Data were analyzed using SAS version 9.4 . The number of individuals and percentage of the sample providing specific responses was obtained using the PROC FREQ procedure. Qualitative data analysis. Qualitative data were analyzed descriptively by two raters and a third rater reviewed codes and rated to agreement . Domains were identified based on reported behaviors, which included stopping sex work or engaging in online sex work and the impacts of COVID-19 on life circumstances, such as employment and housing. Each survey answer was assigned to one or more domains, and the number of responses in each domain was calculated. Exemplary quotes were selected to represent each domain for this paper. --- Results --- Quantitative Findings A total of 50 individuals completed the needs assessment, and n = 46 were available for analysis. Participants were incredibly diverse and ranged in age from 18 to over 55 and represented multiple racial groups , ethnicities , genders , and sexual orientations . Most individuals were homeless and or marginally housed . For additional details about the sample, see Table 1. Sexual behaviors in the past 30 days included anal sex insertive and receptive and vaginal sex insertive and receptive . When data were aggregated, within the past 30 days, individuals reported 144 total anal or vaginal sex partners and 30 of these partnerships did not use condoms. The median of sexual partners was two , and the median number of condomless sex partners was zero . However, 1 3 there were a number who did not endorse of any insertive or receptive anal or insertive or receptive vaginal sex partners in the past 30 days. The use of condoms did vary by type of sex. However, the numbers are too small to draw any meaningful conclusions. See Table 2 for more details. We also examined condom use by PPE use during sex . We found that of the eight individuals who wore PPE while doing sex work, three reported no use of condoms, while the remaining five reported using condoms at least some of the time. Of those individuals who did not wearing PPE during sex, a similar number did not use condoms as did use condoms . Additional details on the frequency of sex, use of condoms, and use of PPE are reported in Table 2. Individuals reported a wide range of substances used during the past 30 days including marijuana , cocaine , prescription stimulants , methamphetamines , prescription opioids , street opioids , sedatives , hallucinogens , inhalants , or some other substance . Additional details on the frequency of use are reported in Table 2. Approximately 40% of the sample reported that their sexual behavior and substance use "stayed the same." Others reported a decrease in the number of partners: "a lot fewer" or "somewhat fewer" partners. Changes in substance use were more varied as there were just as many individuals indicating increased use as reporting decreased use . Additional details about changes in sexual behavior and substance use due to COVID-19 are shown in Table 3. --- Qualitative Findings Changes in Sex Work Several domains emerged as important themes in the practice of sex work during COVID-19. For many, there had been no reported changes to their sex work behavior. When asked about how COVID-19 impacted sex work, these individuals simply responded "it hasn't." One individual expanded on that idea and said, "It hasn't because I'm young and healthy. I don't want to harm elderly people." This indicates that perhaps some thought and decision-making was going into selection of sexual partners and choosing older partners was a concern for spreading the virus. For others, they reported a reduced number of clients either by their own choice, by their "regular" clients' choices, or by lack of available new partners. For example, one individual indicated making a choice to cut down on partners, "Have cut down on partners because of stay at home order" and another indicated, "sex work is very slow." For some, cutting down was not enough, and they decided to become celibate or completely stop sex work. Several individuals stated that they had completely stopped doing sex work during COVID-19 due to their fears of becoming sick. Many indicated they were "at risk" of COVID-19 based on their HIV status and/ or were concerned that doing sex work would make them vulnerable to illness. Of note, it is possible that for some individuals this was already part of their goals. For these individuals, COVID-19 was an additional source of motivation. For those who chose to continue doing sex work, some attempted to implement precaution measures including the use of personal protective equipment or other strategies. One individual noted a "no kissing" rule with clients. Another individual stated, "I've had to come up with creative ways to stay safe ." Finally, one individual noted a transition to online sex work, "Doing sex work online via video instead of in person" as a way to protect their own Client did not answer 1 2.2 health and health of others. Overwhelmingly, many individuals expressed concern for themselves and for fellow streetbased sex workers due to COVID-19 and were trying to adapt to their new reality, as one individual stated, "My fear of COVID has scared me out of relapse. If I go back out there to do sex work during this, I will probably die. Worry about the girls still out there" . Impact on Other Areas of Life Respondents identified significant challenges at structural, social, and individual levels when asked about how COVID-19 had impacted other areas of their lives. With regard to structural challenges, individuals noted more difficulty accessing services, transportation, employment/income, housing, and food. Generally, regarding access to services, an individual noted difficulty with applying for state-based services, "cannot apply for food stamps and RI state ID because registry is closed." With regard to transportation, many rely on public transportation, and noted that these services had changed, "The city buses Housing was also a difficulty as many housing programs had restricted access, reduced capacity, and/or clients were concerned about trying to get into shared housing environments for fear they would become infected with COVID-19. Others indicated they were forced to live at home with family and the issues that were causing distress, "I am very unstably housed and being quarantined with family has been awkward for me." Because many in the sample are unstably housed, food access also became an issue during this time. Many rely on churches or local community organizations providing food on set schedules, but with the pandemic these services were reduced or stopped completely. Many indicated there were needs for additional food services, for example, one individual stated, "There needs to be more places for homeless people to get food." Clients also noted reduced social connections including personal relations. One client noted, "I don't see my family as much as I used to." Another specifically pointed out the challenges of being in early recovery and reduced recovery support due to COVID-19, "Being in early recovery, it's hard to access meetings and support systems." Although only one individual specifically mentioned this limitation, anecdotally, peer specialists shared that this was a more common occurrence and there was frustration among clients about their ability to attend in-person support through 12-step meetings, which many perceive is key to their recovery from substances. Finally, not surprisingly, many individuals noted that in addition to reduced access to services and social connection, they were experiencing worsened mental health and increased substance use. Individuals noted increases in mental health symptoms including depressive, anxious, and psychotic symptoms like delusions and paranoia. For example, one individual noted "worse mood, more paranoia and delusions" and another noted "Isolation has been a trigger for depression and anxiety." Others noted these increases in worsening mental health symptoms interacted with their increased use of substances. Many expressed that their existing mental health and substance use issues were exacerbated by fears of getting COVID-19. For this population, it is extremely difficulty to avoid potential exposures given marginalized housing, sex work, and existing health conditions including HIV. For example, one individual noted, "Makes it more concerning doing sex work. Already have a compromised immune system." Finally, through all of these challenges, there were expressions of resilience. As one client noted, "I'm just grateful to be alive and staying safe as much as possible" . --- Discussion Our study evaluated the impact of COVID-19 on the lives of street-based sex workers in the U.S. In the open-ended questions, many respondents expressed concerns about the possibility of transmission and the challenges posed by wearing personal protective equipment or practicing social distancing guidelines while trying to support themselves financially. The number of total sex acts also indicates there were many episodes of close, interpersonal contact, which is a known risk factor for COVID-19 transmission. While we did not ask about the frequency of use of PPE, the low number of persons using PPE suggests that several of these encounters had the potential for COVID-19 transmission. In addition to potential risk of COVID-19 transmission, there was also potential for HIV transmission, as only half of the sample endorsed using condoms during sex. Interestingly, when we examined condom use by PPE use, we found that at least three individuals who had not used condoms were using PPE during sex, suggesting that for those individuals COVID-19 might have posed a more salient health risk than STIs/HIV. These findings suggest that in addition to COVID-19 transmission, there was also the possibility of STI or HIV transmission, a dual model of risk. To our knowledge, we are the first to report on the use of PPE or other precaution measures taken by street-based sex workers during COVID-19. Syndemics theory posits that co-occurring social epidemics create a synergy that places individuals at risk of disease and has primarily been used to describe the forces that place individuals at increased risk of STIs and HIV . Our data are interpreted through this lens, recognizing that behaviors such as condomless sex and poor adherence to social distancing are influenced by larger structural and social systems that affect individual behavior. From the syndemics lens, our data indicate that multiple, co-occurring risk factors place street-based sex workers at higher risk of COVID-19 and associated social impacts. The syndemics framework clarifies how multiple factors are also associated with increased risk of COVID-19. Streetbased sex workers often operate at the margins of society and experience multiple, syndemic conditions including homelessness, hunger, poverty, substance use, and mental health issues . Paradoxically, despite challenges experienced by this group, they are often forgotten in the identification of vulnerable communities and not considered in relief programs . We also recognize that in addition to experiencing multiple co-occurring syndemic conditions, identity as a sex worker likely also intersects with other unique identities including race, gender, and sexual orientation, all of which impact the ability to navigate through society with relatively more ease or difficulty. Our sample of street-based sex workers was incredibly diverse with regard to sex, gender, ethnicity, and race; and, although our sample is not large enough to compare groups and find significant differences, we may expect that individuals' intersectional identities might differentially impact how they are affected by COVID-19. This group is also heterogenous with regard to whether they selfidentify as sex workers, consider sex work a form of employment, and/or engage in exchange sex only occasionally for money, drugs, or things they may need. We recognize that each of these identities likely intersects with empowerment and perceived control or ability to engage in behaviors to protect their health. Many individuals in our sample expressed they did not feel they had the power to assert themselves to wear PPE with clients or practice other precautions. Individuals who did not use condoms may have also been affected by similar concerns. The theory of gender and power has been applied to HIV risk among women and posits that individuals may be unable to assert their sexual health needs due to inherent power differences with their sexual partners . Street-based sex workers, regardless of gender identity, often hold less social capital and power than their clients and may similarly struggle with asserting their health needs with regard to wearing PPE to prevent COVID-19 and/or condoms to prevent STIs/HIV. Our findings that street-based sex workers have been greatly affected by COVID-19 are consistent with work conducted by Callendar et al. in an international study of male sex workers using an online forum. They reported that COVID-19 impacted the sex industry and reduced the number of individuals entering sex work, obtaining sex work, and engaging in online services instead of in-person as much as possible. However, the study also identified that sex workers with fewer options for other work downplayed the significance of the risks posed by COVID-19 , suggesting that cognitive dissonance may contribute to justifications of persons who need to expose themselves to COVID-19 to survive. Cognitive dissonance theory suggests that when individuals notice a discrepancy between their attitudes and behaviors, they are motivated to change one or the other to reduce dissonance or internal conflict. Applied to this situation, when individuals are forced to continue to work in unsafe conditions and potentially expose themselves to COVID-19, they may alter attitudes to perceive this as safer than it is to protect their selfimage and reduce dissonance . For street-based sex workers experiencing homelessness, COVID-19 has represented an additional challenge. People who are experiencing homelessness are at a significant disadvantage when it comes to preventing infection as they often lack access to regular hygiene and sanitation . Many shelters or housing facilities that would normally take new residents declined to do so because of COVID-19, and those that did take in residents often saw outbreaks of COVID-19 due to poor hygiene and close living quarters . Additionally, soup kitchens and religious organizations who provide meals to people experiencing homelessness ceased operations leaving these individuals without sources of food . Access to PPE at the time of data collection globally and locally was limited, and this impacted not only risk of COVID-19 acquisition and transmission during sexual encounters, but also barriers to accessing general services. For example, one participant mentioned that the public bus system was requiring masks, which means that for those individuals without masks who rely on the bus system, they were not allowed to enter. Since data collection, PPE has become much more available and the CBO has also made efforts to distribute PPE to their clients and the larger community of street-based sex workers to ensure that individuals are able to wear masks when accessing public services like transportation. Several changes have been made to try to address some of the issues identified in this survey to better improve the lives of clients through statewide initiatives, non-profit support, and actions taken at the CBO. These include but are not limited to providing clients with surgical masks during outreach, continued HIV/HCV testing and referrals to PrEP and other HIV prevention services, delivering basic groceries and food items to clients, and providing clients with cell phones to access various service organizations and make and attend telehealth appointments. These efforts have been useful in addressing the needs of clients and helping improve their quality of life. However, we recognize that systemic challenges that affect street-based sex workers generally, and even more so during COVID-19, will not be ameliorated with small efforts. As such, we hope that dissemination of these findings can be used to raise awareness and increase change at a larger level to address some of these systemic disadvantages that continue to place street-based sex workers at risk of COVID-19 related morbidity and mortality. Our study was limited by the use of data collection methods employed. Namely, peer specialists administered brief surveys during community outreach. Thus, there were missing data, and a few surveys were lost during the data collection process. However, the use of trusted outreach workers and peer specialists with lived experience through a CBO that works with the population likely also conferred specific benefits including truthful responses and willingness to participate from a population that is often excluded from traditional research on the impact of COVID-19. Our study was also limited by the same factor as all cross-sectional data collection-we are unable to comment on the temporality of associations observed. Finally, we were limited by a small sample size in that we were not able to undergo inferential statistics to understand more complex relationships between the variables studied. --- Conclusions In conclusion, our study documents the impact of COVID-19 on sex work, substance use, and other areas of life among a group of street-based sex workers utilizing survey data with open-ended qualitative sections. Importantly, we identify street-based sex workers as a population at risk of COVID-19 due to their marginalization from society, as well as life circumstances including relying on shelter housing and close interpersonal interaction with multiple partners to survive. The use of a syndemics framework provides a lens to understand the multiple social circumstances that affect the health of this high-needs population and potentially place them at high risk of COVID-19. We hope these findings serve as a call to action for institutions and communities and nationaland state-level funding agencies to identify sex workers as a population of public health importance given the risks implied by the nature of their work and limitations to services, housing, and food which, in turn, places them-and thus the larger community-at greater risk of COVID-19. --- --- Compliance with Ethical Standards Conflict of interest There is no potential conflict of interest to report. --- Ethics Approval All procedures were approved by The Miriam Hospital Institutional Review Board. The study was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. Informed Consent A waiver of written informed consent was obtained from the Institutional Review Board at The Miriam Hospital. The study was a secondary analysis of needs assessment data conducted with partners from a community-based organization that provides harm reduction services to at-risk individuals. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

COVID-19 has disproportionately affected vulnerable populations across the U.S. Street-based sex workers are one vulnerable population whose health and impact of COVID-19 have been understudied to date. The goal of this study was to evaluate findings from a community needs assessment with street-based sex workers on impact of COVID-19 on health behaviors and social circumstances. A brief survey was developed at a community-based harm reduction and recovery services organization. Surveys were administered by peer specialists to street-based sex workers during street outreach in April and May 2020. A total of 46 surveys were analyzed. Many individuals reported continuing to do sex work and use substances during the COVID pandemic. Slightly more than a quarter of individuals (n = 13; 28.3%) indicated using personal protective equipment while doing sex work and described challenges to using precautions when working with clients. Individuals had used marijuana (n = 32, 71.1%), cocaine (n = 17, 39.5%), prescription stimulants (n = 9, 21.4%), methamphetamines (n = 5, 11.9%), prescription opioids (n = 12, 27.3%), street opioids (n = 12, 27.3%), sedatives (n = 11, 25.0%), hallucinogens (n = 3, 6.8%), inhalants (n = 3, 7.0%), or some other substance (n = 4, 8.7%) in the past 30 days. About half (48.8%) reported that COVID-19 had a major impact on their lives. This study is among the first to characterize the impact of COVID-19 on street-based sex workers. From a public health standpoint, this group also represents a high-priority population given their vulnerability and close contact with others, which increases the potential for community spread.

Background In developed countries, many children live in a digitally enmeshed world. Time spent with screen media devices, including televisions , computers , electronic video games, smartphones and tablets, saturates the waking hours of young children [1][2][3][4]. Specifically, in recent years there has been a rapid uptake of mobile screen media devices among young children living in developed countries [5][6][7] and a recent study found that over onehalf of 3 year olds were given their own tablet [8]. Concerns about the effects of screen time on physical and biopsychological well-being among children have been raised [9][10][11]. Longer periods exposed to screens have been associated with higher risk of overweight and obesity due to the lack of physical activity and a negative impact on diet [12,13]. Other potential adverse effects of media exposure includes delayed cognitive and language development, attention deficits and behavioral problems such as violent behavior and aggression [14][15][16]. Overall, children with higher screen time had increased risk of having a lower well-being, [17] and obesity, depression and anxiety have also been found associated with the use of newer technologies, such as smartphones [18][19][20]. Studies like those prompted authorities to recommend limits for children's daily screen: less than 1 h per day for children aged 2 to 5 years and less than 2 h per day for older children [3,21,22]. Moreover, the American Academy of Pediatrics [21] encourage parents to develop a family media use plan specific for each family and each family member. However, worldwide, a significant proportion of children are not following the recommended exposure time [23][24][25]. It is critical to understand when and how young children make use of the different screen media. However, very few studies on screen time have included children below the age of 5 years and most of the findings were based on passive forms of technology prior to the widespread introduction of mobile and touch-screen formats [26,27]. In addition, identifying population groups with the highest risk of accumulating excessive screen time enables the appropriate targeting of intervention programs. To provide further information on this issue, this study aims to determine the patterns of use of traditional devices as well as emerging devices by Portuguese young children according to their gender, age and socioeconomic position . --- Methods --- --- Variables collected This study is part of the national project "Inequalities in childhood obesity: the impact of the socioeconomic crisis in Portugal from 2009 to 2015". The purpose of the referred project is to gain an understanding on the Portuguese prevalence of obesity and to explore a number of possible-related behaviors, including sedentary activities. The screen-viewing behavior of the children was assessed by parental questionnaires previously used in a similar population [28]. Questionnaires are the most common method to measure screen time, particularly in young children, and the literature suggests that a questionnaire with strong psychometric properties can be a useful tool that estimates screen ownership and time in a simple, fast, no-cost, and completely anonymous way. Specifically, parents were asked to report the average time per day that the child spent watching TV, playing electronic videogames, using the PC, the tablet or a smartphone. Each device was accessed in a different question and separate responses were collected for weekend and weekday use given that previous research suggests that level of children's screen time may differ between weekdays and weekend [29]. Response options were none = 0, less than 1 h per day , 1 h/d = 60 min, 2 h/d = 120 min, 3 h/d = 180 min, 4 h/d = 240 min, and more than 4 h/d . Total screen time was determined by adding television, computer, video games, smartphone, and tablets use. Given the AAP recommendations for daily screen use [21] and considering that previous studies have found that, for children, ≥2 h daily TV viewing was associated with reduced physical and psychosocial health [30] and <1 h/d of video game playing was associated with positive psychosocial health [31], screen time was categorized into 1) less than 1 h/d, 2) between 1 h/d and 2 h/d, and 3) 2 h/d or more. Items referred to behaviors occurring outside of school hours but did not distinguished recreational from educational use. Factors such as gender, age, parental education and occupation, and urbanization were collected. Father's and mother's educational level was based on the Portuguese Educational system -9 years' education or less, 10-12 years' education , and higher educationand grouped into three categories . Father's and mother's occupation was self-reported and later classified according to the Portuguese Classification of Occupation , which provided ten groups, namely: 0) Armed Forces, 1) Managers, 2) Professionals, 3) Technicians and associate professionals, 4) Clerical support workers, 5) Service and sales workers, 6) Skilled agricultural, forestry and fishery workers, 7) Craft and related trades workers, 8) Plant and machine operators and assemblers, and 9) Elementary occupations [32]. Place of residence was reported by the parents and urbanization was classified according to the criteria of the Portuguese Statistical System [33]. Father education was used as a proxy measure to the socioeconomic position , as seen in previous studies, including in the Portuguese context, since the country does not have an official measure of this variable [34,35]. --- Ethical approval Prior to commencing the study, the protocol was approved by Direção Geral do Ensino and Comissão Nacional de Proteção de Dados , the Portuguese Data Protection Authority . All procedures were in accordance with the 1964 Helsinki declaration and its later amendments. Written informed consent was obtained from children's parents. --- Statistical analyses A descriptive analysis was done using the screen time for all 7 days of the week according to gender and SEP, using age as a continuous variable. Later, the sample was divided in two groups according to children's age: 3 to 5 years and 6 to 10 years . Linear regression models with 3 steps were used to predict children's screen time according to socioeconomic position. Model 2 was adjusted for children's gender. Model 3, included the variable in previous model, plus mother's education, parental occupation and urbanization. Model 3 allowed us to explore the interactions of other socioeconomic indicators. The adjusted R-square was determined at each step. A final analysis was done using the categorized screen time and possible statistical differences between age groups were calculated using chi-square tests. All analyses were performed in SPSS version 23 using a significance level of 0.05. --- Results Among the 3 to 5 years old children , 52.6% were males and 47.4% were females. The mean age was 4.47 ± 0.67 years. Most preschool-aged children were from medium SEP families , followed by high and low SEP . A total of 6570 children aged 6 to 10 years old were observed; 50.3% were males and the mean age was 7.94 ± 1.35 years. Most children were from high or medium SEP families . Daily total screen time was higher during the weekend than the weekdays, both for younger -183.15 min/day and 97.59 min/day , respectively -and older children -251.61 min/day and 99.91 min/day , respectively . Figures 1 and2 shows the frequency of weekly use by children's age and according to their gender and SEP. In comparison with other devices, screen time allocated to TV was the highest, independently of children's age, remained almost constant across all age groups, and extremely high since a young age. Among girls, the time allocated for computer increased rapidly across ages, while for boys, the most rapidly increase was found for electronic game devices followed by computer. Boys, compared to girls, spent significantly more minutes per day using screen media devices and differences according to gender are bigger among older children. Overall, socioeconomic disadvantaged children spent significantly more time per day using screen devices than children from medium and high SEP and the screen time of children from medium SEP was more similar with their counterparts from lower SEP than from high SEP. Screen time was progressively higher among older children, primarily driven by more time spent on electronic devices such as computers, electronic video games and tablets. The linear regression shows that SEP was a negative predictor of screen time, meaning that children from higher SEP were more likely to spend less minutes per day using screen media devices compared with socioeconomic disadvantaged children . However, for most media devices, SEP lost is significantly association when adjusted for confounding variables exceed that recommendation during the weekend. Television alone accounted for the most screen time with most children spending more than 1 h/d using that device, particularly during the weekend . The prevalence of <1 h daily TV viewing, PC and electronic video games use was significantly lower in older children compared with the ones aged 3 to 5 years for both weekdays and weekend. The prevalence of ≥1 h/d using tablets and smartphones was significantly higher among older children, compared with their younger counterparts, but differences were only found for the weekend. --- Discussion This study describes the patterns of use of electronic devices in Portuguese young children, according to their gender, age and SEP, measured by father education. Rapid advancements and increased ownership of information and communications technology in recent years has increased the variety of screen-based media available to young people however, present findings show that TV viewing remains the predominant source of children's electronic media use. This is consistent with results from other countries like the United States [4,7,36] and the United Kingdom [37]. At age 3, children were already using screen base devices on a daily basis and the time spent on screen devices progressively increased with age. This age-related increase in screen time has been commonly reported [37]. However, TV time was already high among 3 and 4 years old children, occupying approximately 91 and 96 min per day, respectively. TV time was similar or even higher than the ones reported in the last decades, before the advent of mobile devices [7,38,39] which increases concerns on excessive screen time and potential harmful effects on children's health [40], given that, the increasingly popularity of new technologies are being added to daily traditional screens use . Chen and Adler reported a similar pattern in which screen time doubled among children aged 0 to 2 years, from 1997 to 2014, mainly because young children's TV time did not decrease after the advent of mobile devices [7]. Tablet was the second equipment that consumed more time from children, independently of their age group, gender, and SEP, which is consistent with the latest screen media report carried out in 3-4 and 5-15 years old children from the United Kingdom [41]. Also, a study from North America, found that most children had their own tablet by age 4 and that by age 2, 3 out of 4 children were using mobile devices on a daily basis [8]. Increase ownership and patterns of use suggest that mobile devices, more specifically tablets, may displace TV as major sources of media consumption for young children in the years to come. However, tablets may not be very different from TV, given that a previous study found that photo and video viewing were the most common activities performed on touch-screen devices [42]. In this case, some of the issues that arise with passive watching TV still apply such as, exposure to unsuitable material, passive eating, increased consumption of energy-dense foods and sugar-sweetened beverages, and displacement of other developmentally important activities [43,44]. Another emerging device, the smartphone, was extremely common, surpassing the use of computers and electronic video game devices among the youngest children. Latest findings show that household ownership of tablets and smartphones doubled in the last decade, including within preschool children [4,8] which may help to explain present results. Frequent mobile device use is likely to increase children's social isolation, and hinder opportunities for social interaction with family and friends [45]. In aggregate, the total screen time reported for many of these children, including the young ones, exceed previous recommendations to limit daily screen time to less than 1 or 2 h [21,22]. In this sample, more than 90% of children spent more than 1 h/day using screen media devices. Moreover, close to 77% of children aged 6 years and above use media devices for 2 h or more on a daily basis which can significantly reduce the physical and psychosocial health [30,31,46]. These findings are in line with previous analysis that demonstrated widespread usage of screen-based media in young people living in a variety of European countries, United States, Brazil and Australia, in which at least two thirds of 4 to 17 years old children exceeded 2 h/day of screen time [27,47]. Similarly, data collected in 2002/2003 among Portuguese children aged 7-9 years showed that approximately 70% spent more than 2 h/day watching TV, which was the most predominant sedentary behavior followed by electronic games and computer use [38]. Consistent with other studies we found that screen time activities differed between boys and girls [23,47]. Of all the activities covered in this study, gaming had the biggest gender disparity: boys in each age group spend more minutes than girls in a typical week playing games. The same tendency was recently reported for British pre-school and school-aged children [41]. Also, differences by gender increased between age groups and the higher levels of screen viewing by older children could be at least partially explained by increased use of computers and electronic video games [48,49]. The family SEP was significantly associated with screen time, even after adjusting for confounders, and the disparities were similar to previous studies that found inverse associations between SEP and screenbased media use [7,47,50,51]. SEP explained between 4 to 8% of the variance in children's screen time which is consistent with the work of Carson and Janssen, [52] where family demographic explained approximately 8% of the variance in screen time among Canadian children aged 0-5 years. Generally, the results indicate that socioeconomic disadvantaged children spent, in average, more time using media devices compared with children from higher SEP. Moreover, in most devices, the results for medium SEP children were more closely related with the ones from low SEP children than with their counterparts from higher SEP. Decreasing costs and marketing strategies can explain the access to screen devices by all SEP groups. In addition, it might be possible that parents from more educated backgrounds set limits when children use screen devices. Other studies have shown that a lower parental education is associated with lower parental modeling, less parental co-viewing, more chance to have a TV in the bedroom and to eat meals in front of the TV [53]. Other possible explanation is that poorer children may spend more time indoors due to a greater likelihood of living in unsafe neighborhoods [54]. The difference in screen media use between lower vs. higher SEP was found between all ages but the gap was even larger in older than in younger children, particularly in some devices like electronic games and smartphones. Present findings suggest the need to intervene at different stages of childhood, starting in young ages, in order to decrease specific screen time behaviors. Physicians and other health care providers should counsel parents and caregivers of young children on the appropriate use of screen time. Strengths include the large sample size collected at a national level in pre-school aged and schools aged children. In addition, the screen time was obtained for all of the electronic devices, including traditional and emerging/mobile devices such as tablets and smartphones. The survey assessed weekday as well as Saturday and Sunday screen time and the study took both time periods in consideration. Also this was one of the first studies to date to examine screen time in traditional and emerging devices in young children according to the SEP. Limitations of the study include the cross sectional design which makes us unable to draw cause-effect conclusions and to make observations over the time. In addition, since it is not feasible to obtain direct measurements of screen time in large population-based studies, the screen time measures were parental-reported. The information bias associated with these measures may have resulted in an underestimation of screen time. --- Conclusions This study shows that screen time in children aged 3 to 10 years, is longer than the recommended, particularly among boys, that children engage in these activities starting at a young age and increases in screen-based entertainment use occurs from younger to older children. Moreover, SEP explained some of the variance in children's screen time, with screen time being higher in socioeconomic disadvantaged children. Effective strategies targeting children and/or their parents are needed to equitably reduce unhealthy behaviors among children. --- --- Abbreviations SEP: Socioeconomic position; TV: Television; PC: Computer; EG: Electronic games; d: Day; h: Hour; min: Minutes; SE: Standard error; CI: Confidence interval; AAP: American Academy of Pediatrics Authors' contributions DR conceived the study aim, analyzed and interpreted the data and drafted the article. AG, AM-R, VR, M-RS, HN and CP designed the original study, developed the study methodology, acquired the data and critically revised the manuscript. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work. --- --- Consent for publication Not applicable. Competing interests Aristides M. Machado-Rodrigues is an Editorial Board Member. The remained authors have indicated they have no potential conflicts of interest to disclose and that they have no financial relationships relevant to this article to disclose. ---

Background: Children are often exposed to too much screen time but few studies have explored the use of old and new digital media among young children. This study assesses screen time, including traditional and mobile devices, in pre-school and elementary school-aged children, according to their gender, age, and socioeconomic position (SEP). Methods: A total of 8430 children (3 to 10 years; 50.8% boys) from the north, center and south-central Portugal were included in the present study. Data was collected by a parental questionnaire during 2016/2017. Children's screen time (by media device, weekdays and at the weekend; calculated by mean minutes per day) were reported by parents. Analysis were carried to compare screen time by children's age, gender and family SEP (classified using father's educational degree). Results: Daily screen time was high both in children aged 3 to 5 and 6 to 10 years -154 min/day (95% CI: 149.51-158.91) and 200.79 min/day (95% CI: 197.08-204.50), respectivelyand the majority of children, independently of their gender, exceed the recommended 2 h/day of screen viewing. Children are still primarily engaging in screen time through television but the use of mobile devices, particularly tablets, were already high among 3 year-old children and increased with age. SEP was a negative predictor of screen time in the linear regression analysis, including after adjustment. Conclusions: Considering the negative health impacts of excessive screen time, recognizing subgroups at risk of excessive screen time and identifying how each device is used according to age is fundamental to enable appropriate future interventions. The screen time in children aged 3-10 years is longer than the recommended, particularly among boys and in those children from lower SEP. Parents and policymakers should have in mind that children spend most of their screen time watching television but mobile devices are becoming extremely popular starting at a young age.

Introduction Since its onset, experiences with COVID-19 in China, Europe and the United States provide clear testimony of its far-reaching health, economic and societal impacts. It is becoming increasingly clear that the pandemic's consequences are likely to be even graver in many low-and middle-income countries . Higher transmissibility due to overcrowding in densely populated areas, larger household sizes and poor quality of water and sanitation, higher infection-to-case ratios due to high prevalence of comorbidities such as hypertension and diabetes, and higher fatality rates due to lack of health care capacity contribute to more severe health consequences in LMICs . Lockdown measures and restrictions on movement that have been put in place to quell infection rates also have far-reaching adverse effects. Large proportions of the workforce in the informal labour market are left without any source of income and informal supply chains are disrupted . Negative health and economic consequences are exacerbated by limited coverage of social protection . Lack of income support makes it more difficult to adhere to physical distancing by foregoing work outside the home. Estimates suggest that the number of people expected to fall into extreme poverty ranges anywhere between 80 and 395 million . To date , policy measures have focused on providing shortterm economic relief for those most vulnerable and on developing a vaccine that is available and affordable for all . However, as experiences with HIV/AIDS, Ebola, cholera and other transmissible diseases have highlighted, responses to pandemics often suffer from a narrow biomedical focus and overlook the social, economic, political and cultural contexts in which they take place . This limits understandings of how disease might spread, who is affected and how, and how policies should be shaped to offer the most effective and equitable response. This is no different for COVID-19: the sheer scale of the pandemic, the speed at which it continues to spread and the urgency with which response measures are being put in place means that socioeconomic dynamics and inequities receive relatively little attention . Moreover, LMICs' replication of policies that originate from high-income countries call into question the extent to which such policies are adequately contextualised and pay attention to existing patterns of socioeconomic injustice . Stigma is integral to patterns of injustice, and experiences with other pandemics highlight that paying heed to stigma is vital for providing an effective and equitable response to COVID-19. Long-standing research on HIV and neglected tropical diseases as well as more recent experiences with Ebola indicate that stigma is experienced by those being affected by disease or by groups that may be seen as 'vehicles' for, or most structurally vulnerable to, infection. Stigma also intensifies along fault lines such as class, age, gender and race . Stigma is detrimental to halting and controlling pandemics, to support societies' recovery from a pandemic, and achieving equitable development. It puts up barriers to accessing health care and social services, impedes healthseeking behaviour, undermines adherence to treatment and leads to social exclusion, discrimination, mental distress and-in some cases-violence . Although COVID-19 is still relatively new, the many unknowns and uncertainties associated with this 'newness' may give rise to fear that can in turn fuel stereotyping and stigmatisation . Indeed, stigmatisation of those infected with or having survived the virus as well as health workers who face high exposure is already mounting . The objective of this paper is twofold. First, we assess potential driving factors of COVID-19-related stigma, and how this intersects with existing stigma fault lines. Second, we explore mechanisms through which COVID-19-related stigma may be counteracted and prevented. We do so by taking an interdisciplinary approach and building on well-established research on disease-related stigma in relation to pandemics such as HIV, NTDs and Ebola as well as literature on stigma associated with old age, disability and poverty. The remainder of this article is structured as follows: First, we provide a conceptual framework to underpin our analysis of COVID-19-related stigma. Second, we assess driving factors that contribute to COVID-19-related stigma. Third, we explore fault lines that intersect with and compound COVID-19-related stigma. Fourth, we explore ways in which such stigma may be avoided in the response to COVID-19. Finally, we offer critical reflections for future action. --- Framing Stigma In his seminal work on stigma, Goffman posits that stigma denotes an attribute that is discreditable, leading to a 'spoiled identity'. Stigma represents a characteristic that is different from 'normal' and triggers disapproval or devaluation in certain social contexts with those being stigmatised somehow deemed to be less human . Stigmatisation constitutes others' responses to stigma and can be overt, such as avoidance and humiliation, as well as more subtle, including lack of eye contact . This, in turn, interacts with 'potential internalization of the negative beliefs and feelings associated with the stigmatised condition' , understood as self-stigma, internalised stigma or shame. One can also distinguish between 'existential' and 'achieved' stigma . Existential stigma derives from a condition that is mostly outside of the control of the target of the stigma, while achieved stigma applies to those 'who have earned a stigma because of their conduct and/or because they contributed heavily to attaining that stigma ' . Stigma has far-reaching intrapersonal, interpersonal and societal consequences, ultimately undermining efforts to counteract pandemics and promote equitable development. It leads to social isolation, hampers access to care, and reduces treatment adherence . During the Ebola pandemic in western Africa from 2014 to 2016, Ebola patients, survivors and orphans , as well as health workers involved in the response faced "shunning and isolation, ill-treatment, [and] the rejection of treatment and political or economic exclusion" . Fear of stigma led Ebola-affected families to hide symptoms from their communities and health services . Similar reports emerged from survivors and health workers as a result of the SARS pandemic in some Asian countries . There is also published evidence of widespread stigma related to NTDs such as lymphatic filariasis, podoconiosis, Buruli ulcer, onchocerciasis, leprosy, and leishmaniasis . These examples of disease-related stigma can be linked to Goffman's notion of 'spoiled identity'. A common critique of Goffman's theory, however, is that it provides an apolitical understanding of stigma, placing the onus of being 'discredited' on those that are stigmatised. Instead, as pointed out by scholars across the health and social sciences, stigma takes hold and is perpetuated within a social context with inherent power structures, serving to put in place and maintain hierarchies and social order . Building on conceptualisation of stigma within the field of HIV , COVID-19-related stigma can be understood as a social phenomenon whereby people infected with or affected by COVID-19 are labelled, stereotyped or discriminated against and that intersects with existing fault lines of socioeconomic disadvantage, thereby perpetuating and reinforcing power structures and structural inequalities. Stigma is the visible 'tip of the iceberg' of much wider systems of beliefs about who carries disease and why that are grounded in and compounding social inequalities . The Health Stigma and Discrimination Framework by Stangl et al. helps to ground understandings of stigma within a wider context and to map the process from drivers and facilitators of stigma through to outcomes of stigma. Drivers of stigma tend to be specific to the health condition, including fear of infection, lack of awareness, stereotypes and prejudice. Facilitators of stigma apply more broadly and include social and cultural norms, levels of inequality and legal and policy environments. The health-related stigma itself intersects with other stigma 'markers' such as age, gender, race and class. This then leads to experiences, manifestations and outcomes at individual, societal and institutional levels, ranging from internalisation of stigma to social exclusion and institutionalised discrimination . The framework's multi-layered focus and the way in which it locates the process of stigmatisation within the wider socioeconomic context makes it particularly useful for coming to grips with the unfolding and evolving nature of COVID-19-related stigma in LMICs. It affords an opportunity to move away from biomedical explanations of how the pandemic may evolve and provides a biosocial perspective. Experiences in relation to HIV and AIDS have evidenced the importance of understanding social, economic and political processes surrounding disease in order to create an effective response . Given the evolving nature of the pandemic and its drivers, dynamics and manifestations, we zoom in on emergent driving factors of COVID-19-related stigma and fault lines along which existent stigma may be reproduced and compounded. --- Driving Factors of COVID-19-Related Stigma Based on a review of stigma in relation to other transmissible diseases, attributes such as old age, disability and poverty and unfolding knowledge about COVID-19-related stigma, we identify three interrelated driving factors. --- Lack of Information and Misconceptions First, lack of information, misinformation and misconceptions serve as key conduits for stigma, often assigning blame for disease. Whilst biomedical explanations may be widely available, alternative explanations or misinformation may emerge with specific individuals being singled out as causes or carriers of disease, playing into 'achieved' stigma. For example, during previous Ebola outbreaks in the Democratic Republic of Congo , the disease had been attributed to witchcraft or sorcery, supposedly enacted by those who were perceived to 'benefit' from a patient's death, widows or mothers of the diseased, or people who are seen as outcast and in contact with non-natural forces . In Pakistan, the general public incorrectly identified dementia symptoms as a normal part of aging, or perhaps more concerning, that it is attributable to the actions of the individual. The latter inevitably led to people believing that the person with dementia should be responsible for their own healthcare and outcomes . Similar dynamics have been observed in relation to disability with cultural and religious beliefs playing into false perceptions about causes of disability . Across Africa, stigma-inducing beliefs include disability being caused by ancestral curses or transgressions in previous generations, demonic possession and supernatural or otherworldly interference, promiscuity or other sinful behaviour by the mother or other family member or a punishment for wrongdoing by the person with a disability . Stigma may also arise from beliefs that people with a certain condition pose a burden to society. These experiences are clearly visible in relation to disability with low expectations about how people with disabilities can or cannot contribute to society leaving them labelled as less able or productive . They also play out in relation to poverty, with people in poverty being blamed for their predicament on presumptions of laziness or unwillingness to work . Misinformation and misconceptions in relation to COVID-19 are already widespread. For example, in Ghana there was a voice recording where it was believed President Nana Akufo-Addo was endorsing numerous false conspiracies, including the claim that the pandemic was a planned event. Ghana's Information Minister has confirmed the voice on the recording is not that of the president . In Nigeria and Sudan politicians, have been seen wearing 'protective badges' that are marketed as 'virus blockers' or 'virus stoppers'. It is claimed the badges kill viruses and bacteria by releasing a type of bleaching agent, but experts report that such 'blockers' do not work and could actually cause harm . There are still many unknowns about the virus , offering fertile feeding ground for misconceptions and alternative explanations. Internet and social media are widely used to obtain information about COVID-19 but often information is of low quality . The WHO has referred to an 'infodemic' to denote the excessive volume of information that is available , much of which is incorrect . There is some suggestion that the spread of misinformation may be larger in LMICs , possibly due to greater reliance on social media for information and lower capacity for fact-checking. Misinformation about COVID-19, including about its origin, spread and treatment, is widespread across Africa, hampering an effective response and risking demonisation of those infected . BBC News Africa has launched a COVID-19 misinformation hub, offering a service that fact-checks and debunks false information about the pandemic . --- Fear of Contagion Second, fear of contagion is a significant driver of prejudice, stereotyping and stigmatising behaviour. This has been widely observed as in relation to disability and NTDs . Widespread anxiety about the cause and spread of SARS constituted an important mechanism for stigma to take hold at the time of the outbreak in Asia in the early 2000s . In China, fear of contracting HIV was an important source of stigmatising attitudes associated with HIV and AIDS among rural-urban migrants . This also relates to 'existential' and 'achieved stigma'; disease-related stigma is often related to the extent to which the person being stigmatized is deemed at fault for having contracted the illness if their behaviour contributed to infection. Such 'victim-blaming' has been widely observed in relation to HIV and AIDS, including in the Caribbean . The infectious nature of COVID-19 and its potentially life-threatening consequences coupled with ubiquitous updates about infection and fatality rates leads to high levels of anxiety, stoking fear of infection . Already examples are emerging of how such fear leads to prejudice, avoidance and demonisation among those having been infected with or having survived COVID-19. In Ecuador, the family of the first person having contracted the virus fell victim to a harsh social media campaign, accusing them of irresponsible behaviour . In Indonesia, names of the first few cases were leaked online, with one woman being falsely blamed for having contracted the virus due to selling sex to a foreign client . In Kenya and Zimbabwe, survivors of COVID-19 were shunned by their communities after having returned from hospital for fear of them infecting others . In Bangladesh and India, social shunning of those being infected with COVID-19 reportedly led to suicide . Healthcare workers around the world, including in LMICs, also face harassment and stigmatisation, largely out of fear for infection . In Mexico, doctors and nurses reportedly used bicycles to travel to and from work as they were denied access to public transport; in India, healthcare workers have been asked to vacate their rental homes and face social isolation . --- Targeted Policies Third, policies that single out and separate vulnerable groups and the narratives that underpin them often inadvertently act as conduits for stigma. Such policies are the most visible way of labelling those who are 'deviant' and distinguishing them from the rest of society who are deemed 'normal', laying the foundation for legitimisation and institutionalisation of stigma . The cholera response in Brazil, for example, unwittingly 'earmarked' specific communities or persons as foci of disease transmission through their targeting, treatment and risk communications strategy . In relation to disability, there is a long history of segregating people with disabilities from the rest of society by placing them in institutions, leading to perpetuation of negative stereotypes . With respect to old age, there are stereotypes that older adults are weak and need help , feeding into a notion that elderly persons represent a burden to society. Support to people in poverty is commonly targeted on premises of 'deservingness' , leading to a toxic narrative about who should get what and why . Many policies that seek to contain the spread of COVID-19 rely strongly on practices of singling out, separating and labelling, thereby laying the foundation for stigma. Quarantine, a measure that has also been used in Ebola and SARS outbreaks, is likely to feed COVID-19-related stigma . In India, air crew staff felt stigmatised after health authorities stamped 'quarantined' on their houses . Gender-based physical distancing measures may also bring to the light people's sexual orientation or non-conforming gender identities, putting them at risk of harassment and prejudice. In Colombia, measures to limit the number of people on the streets allowed men and women to go outdoors for essential tasks on alternate days, leaving trans people at risk. Multiple reports of abuse and violence against trans people were reported . --- Fault Lines for COVID-19-Related Stigma Disease-related stigma does not stand on its own and interacts with other fault lines, layering onto already existing processes of discrimination and marginalisation and leading to intersecting or co-stigmas. For example, when cholera reemerged in Latin America in the 1990s, the disease was attributed to poor people, who had been previously neglected and marginalised by the state in terms of provision of basic water and sanitation infrastructure and health services. Cholera was also racialised, and poor black and indigenous people were 'othered' as dirty or filthy . Those living in informal settlements were unable to implement the government guidelines on cholera, because of inadequate infrastructure . In China, HIV/AIDS-related stigma was found to co-exist and interact with commonly associated stigma attached to injection drug use and commercial sex . Across LMICs, we can also see such fault lines emerging that make certain groups more vulnerable to COVID-19-related stigma. We identify four main fault lines. --- Poverty and Informality First, people living in poverty, precarious conditions or informal settlements are at a heightened risk of COVID-19-related stigma. Poverty and deprivation itself is a potent trigger for stigma, leading to internalisation of stigma and shame . This in combination with the many difficulties that people in poverty and precarious conditions face in adhering to distancing measures provides a potent mix for stigma. Guidelines on hand hygiene, physical distancing or lockdown are often not plausible in densely populated, informal and impoverished contexts due to crowding, insufficient water and sanitation services and low and precarious incomes . Similarly, farm workers, refugees and people living on the street are both less able to adhere to lockdown and physical distancing measures while also more affected by those measures . This makes them vulnerable to negative stereotyping and being labelled as vehicles of infection, which compounds stigma that pre-dates COVID-19 . --- Ethnicity, Origin and Nationality Second, ethnicity and nationality represent important fault lines for COVID-19-related stigma. Blaming a 'foreign other' for the outbreak of disease is commonplace in order to make sense of and distance oneself from disease . For example, during the 2018 Ebola outbreak in DRC, the Twa indigenous group was singled out by the majority Bantu community as transmitters of the disease, and experienced discrimination in their access to health services . Stigmatisation of Asians, or anyone deemed of Asian origin, was widespread in North America and Europe in relation to SARS and Avian Flu . Similar co-stigma is likely to occur in relation to COVID-19. The fact that the virus first emerged in China has given rise to or compounded pre-existing negative attitudes, discrimination and xenophobia directed at Chinese people or those who are perceived to be of Asian descent in many high-income countries, such as Canada . Few reports of Sinophobia in relation to COVID-19 have emerged from across LMICs, although experiences involving Chinese nationals in Uganda indicate that mistrust based on assumed nationality and origin is on the rise . Instead, a more disparate group of people 'on the move' have been stigmatised for spreading the virus. Depending on who and where these groups are, this may be associated with ethnicity, nationality or place of origin . In Mozambique, for example, mine workers who migrated back home from South Africa were met with suspicion when integrating back into their communities . Finally, with a growing evidence base now indicating that Black, Asian and Hispanic patients are at a higher risk of contracting COVID-19 , ethnic minority is further compounded as a co-stigma. --- Age Third, age-and old age in particular-is likely to emerge as a fault line for COVID-19-related stigma. Deaths associated with COVID-19, disproportionately occur in older adults . This can be attributed to older adults having increased likelihood of underlying health conditions such as diabetes and cardiovascular disease , hypertension , and respiratory problems , all of which increase the risk of COVID-19-related deaths . In addition, ageing results in a general decline in immune responses , making older adults more vulnerable to such viruses. As such, it is unsurprising that many health policies are put in place to protect this group. However, this focus on old age as a factor of vulnerability is likely to compound existing patterns of ageism. In Kenya, ageism has led to the marginalisation, loss of power and dignity of older adults, opening the way to elder abuse . Within LMICs more broadly, older adults describe that they are being denied autonomy related to finances, work, family life and their everyday activities . Governments may not have the healthcare and support systems in place to support older adults, further restricting older adults' choices. This in itself may reflect structural discrimination, as age limits imposed by global health policy and targets may lead to governments not allocating health resources to older adults . Current literature from HICs indicate that a significant minority of public discourse about older adults and COVID-19 express ideas that older people are less valuable or that they should just be isolated . Such narratives, particularly when perpetuated by those in power, could lead to policy that isolates older adults and discriminates against their basic human rights . Lack of health care capacity to provide care for all patients and has led to debates about who should be prioritised for treatment, often finding younger people more 'deserving' Whilst ageist issues exist in LMICs, such as in Ghana and Kenya , there is a dearth of data derived from these regions. Cultural differences and government responses to the pandemic are likely to drive how COVID-related stigma may form in relation to age. For example, within China a key theme from social media discourse was that older adults were still seen as "contributing to the community", with views of warmth and competence during the pandemic . --- Gender and Sexualities Fourth, gender and sexualities are emerging as important fault lines for COVID-19-related stigma. In many countries, LGBTI people have previously been blamed for disasters or disease. In Bhutan, for example, transgender women were held responsible for natural disasters . In Zimbabwe, LGBTI people affected by HIV and AIDS were blamed for their illness and for placing a curse on their family . Reports are already emerging of such blaming taking place in relation to COVID-19, such as in Israel where one Rabbi propagated coronavirus as the revenge to gay parades and in Iran where a Shia cleric warned that the pandemic would not subside as long as same-sex marriages were legal in Iran . The United Nations has also highlighted that surveillance and control of populations in order to contain the virus can galvanise oppression of sexual and gender minority peoples , often in settings where transphobic and homophobic actions are already prevalent. In Uganda, for example, police reportedly used laws that prohibited more than 10 people gathering in public to arrest 23 people living in a sheltered housing project for LGBTI people . A study among young people in Asia and the Pacific found that those who identified as LGBTI individuals or members of the transgender community were more likely to experience stigma during the pandemic . --- Countering Stigma There is a wide and well-established body of literature that assesses the role and effectiveness of interventions to counter disease-related stigma. These interventions engage at the intrapersonal level-seeking to counter internalised stigma-as well as at interpersonal and community level. With respect to NTDs, these commonly include self-care and self-help , educational programmes , assistive technology , and community engagement approaches . Other measures against stigmatisation tend to be health education for patients and the community and disease management, yet the effectiveness of these measures has not been demonstrated in the literature . Additional measures include rehabilitation, counselling and support groups . Most of these measures can be considered reactive, aiming to counteract or reduce already existing patterns of stigma. Moreover, they are predominantly premised on a biomedical approach and are therefore highly individualistic, aiming for empowerment on a top-down basis that is largely disconnected from the structures that facilitate and legitimise stigma in the first place . With COVID-19 still evolving and its longer-term responses beginning to take shape, a more proactive and strategic approach to avoid stigma taking hold at a large scale may be more equitable and effective. Adverse unintended social consequences are to be avoided rather than to be grappled with once they have taken hold. Crucially, any policy response needs to acknowledge and address the human rights violations that caused COVID-19 to flourish in the first place, and new violations that have resulted from the disease that allow stigma to reproduce itself and new stigmas to emerge. Without addressing the root causes of violations, the next disease or the next pandemic will repeat the stigma cycle where certain violations cause disease and new violations are created due to disease. Experiences in relation to other disease-related stigma, including NTDs and HIV, points to the importance of a rights-based and inclusive response. Public health programmes that take a rights-based approach have been shown to 'improve service delivery and enhance equality, equity, inclusiveness, and accountability' . The WHO has set out a rights-based approach to NTDs, and that could apply to COVID-19 and other transmissible diseases, which emphasises the principles of participation, non-discrimination, and accountability . This resonates with a recently published report by the United Nations that places human rights at the centre of the response to COVID-19 and recommends, among others, that policies should guarantee meaningful participation of all sectors of society, that efforts need to reach those most vulnerable and traditionally marginalised, that political leaders and other actors should speak out against discrimination and that freedom of expression is guaranteed in order to hold governments to account . We discuss key considerations for a rights-based and inclusive approach to avoiding and mitigating COVID-19-related stigma within the context of LMICs. --- Participation and Inclusion Participation and engagement of those affected by COVID-19 will be vital for ensuring that responses are contextualised, localised and responsive to needs . In order to avoid that social and economic inequalities are overlooked or compounded, participation needs to cut across intersecting stigmas and fault lines such as poverty, ethnicity and age . Depending on the disease and the context, working with communities and those affected by disease will allow for tailored and community-led responses. For example, the Twa in DRC were incorporated into the response to deliver their own services, and communication materials were prepared in the local language . In Liberia, the inclusion of Ebola survivors is recommended for developing interventions for Ebola-related stigma . In addition, states have an obligation to ensure that laws, policies, and the distribution and delivery of resources, health services, and underlying determinants of health adhere to the principles of equality and non-discrimination. For example, in Sierra Leone, one of the initiatives promoted by the Ebola Response Anthropology Platform was a step-by-step reintegration process in which a social contract was signed by the community and the survivor, including the negative results of the tests, the commitment to pursue infection control practices, and the willingness to welcome the individual . Inclusion should not be taken at face value. Experiences with the response to HIV are instructive in this respect. Inclusive actions, and their associated forms of knowledge about HIV, have themselves often been subject to stigma . Indeed, still today gender and sexual minority peoples continue to be failed by the international HIV and AIDS response, even as it bears the imprint of sexualities activism. Inclusion may look like it is happening when 'LGBT recognition' is inserted into policy documents and programme guidelines, but substantive and on-the-ground action is called for, based in the lived realities of people of diverse gender and sexuality . Inclusion may also be at odds with preventing stigma. Focusing efforts on the inclusion of vulnerable and traditionally marginalised groups requires identification and targeting of such groups. Targeted service delivery inevitably requires labelling and separating out, which in itself is a source of stigmatisation. Service provision premised on the principle of universalism may be powerful in preventing stigma . In the case of cholera in Brazil, for example, affected poor communities felt less stigmatised when the target of hygiene promotion activities and communication was universal , and when treatment was given as part of broader services for diarrheal disease or broader health system reinforcement, rather than stigmatising cholera treatment centres . At the same time it should be noted that 'shared stigma' can also galvanise positive action within affected groups , helping them to generate solidarity and reclaim identity . The very nature of COVID-19 also poses challenges to a truly inclusive approach. With many of the containment measures focused on physical distancing, the potential for groups to have contact and learn from each other is inevitably hampered. This poses a real conundrum in terms of trying to avoid stigma . In line with experiences with SARS, balancing the protection of public health while preventing stigmatisation of groups that are greater risk of contracting or spreading the disease during a rapidly evolving pandemic like COVID-19 is fraught with complexities . Experiences with Ebola in Liberia suggest that minimising social isolation, reintegration in recognition of the end of transmission risk and community-wide economic support serve as pre-emptive approaches for mitigating long-term stigma . Indeed, there is evidence that contact is far more effective than education in approaches to engage stigmatizers in efforts to reduce diseaserelated stigma. In addition, meaningful contact has been seen as a key pathway to reduce ageism and stigma towards those with mental illness . However, the potential viral spread of promoting contact between groups, particularly those that are the most vulnerable, calls for caution. Video contact may serve as an alternative, assuming that the infrastructure is in place, though this may be less effective than face-to-face contact . In addition, contact needs to be 'voluntary, equal, intensive and/ or rewarding, prolonged, or where a number of people are involved' . If the overall impact of social interactions is 'troubling, harmful, or otherwise disturbing,' then contact will most likely have a negative impact . --- Language and Labelling Careful use of language is also crucial to a non-discriminatory and inclusive approach. Analogies between disease and war by referring to being 'invaded', to having to 'fight' the virus or having to win the 'battle' feed anxiety and fear of contagion that in turn play into stigma. Common use of the word "vulnerable" could reinforce existing stereotypes of weakness and helplessness, particularly in relation to old age . Experiences with NTDs are also instructive in terms of how language can contribute to discriminatory discourse. With the term NTDs spanning almost forty diverse diseases, the words 'neglected' and 'tropical' provide little understanding of the diseases and their health impacts. Instead, they point to a socially constructed category that is reflective of environmental determinism and denial of systematic marginalisation that have allowed these diseases to persist . Thus, the term and category 'Neglected Tropical Disease' is itself stigmatising, with those suffering from NTDs have been 'cast off' and often experiencing isolation, discrimination, and prejudice in their own communities. These lessons are instructive in terms of language used to categorise flu-like diseases as Asian, and how it gives rise to accompanying discriminatory language that borders on hate-speech. While WHO introduced guidelines in 2015 to halt the naming of diseases in relation to their origin, this does not prevent the association of diseases with their place of origin from popular discourse with politicians and media being common culprits, as is the case for COVID-19 . WHO published recommendations regarding appropriate language in relation to COVID-19, highlighting that 'words matter' . This includes avoiding references to place of origin of the virus, using accurate terminology, and refraining from hyperbolic language that instils fear such as 'plague' . Referring to those infected with COVID-19 and spreading the virus as 'super spreaders' or those who breach distancing measures as 'intentional murderers' creates a division between the 'moral us' and 'immoral other' . Similarly, referring to those who may be infected as 'COVID suspects' may prevent people from seeking healthcare . --- Transparency and Accountability It follows that governments and decision makers need to be transparent about the processes and actions that underlie their policy and legal choices. Monitoring and adequate reporting of policies, laws and legal cases is integral to the rights-based approach, also referred to as 'policy surveillance' . Policy surveillance can provide an analytical tool to assess prevalence and trends, and 'can facilitate the evaluation of health laws, health rights, and accountability mechanisms' . Rights-based approaches have diverse meanings and are subject to different interpretations, depending on the actors involved . It is therefore important to consider and make explicit the differing agendas that different actors might have. Doing so lays the foundation for the agreement that 'duty-bearers' meet their obligations and 'rights holders' claim their rights according to the availability of facilities, good and services, accessibility based on non-discrimination, physical, economic, and information accessibility, acceptability in regards to medical ethics, culture, and sensitivities to age and gender, quality of services and care both scientifically and medically . Bringing the issue of stigma front and centre is crucial to ensure transparent and accountable policy design and implementation. Disease elimination criteria often stipulate that programs 'count' disease burden and morbidity to claim success, while issues like stigma are less tangible and more difficult to address. However, given the early onset and persistence of stigma, seeking to prevent and minimise it during and after disease outbreaks is paramount . Doing so necessitates the inclusion of measures to assess and evaluate reductions in stigma. Although the measurement of stigma is complex, the scientific understanding of stigma depends on our capacity to observe and measure it . In a review of healthrelated stigma identified five common approaches to measurement, ranging from surveys of attitudes towards those with certain health conditions to interviews with those affected by such health conditions. These tools ought to be located in specific structural and cultural relations to facilitate the important distinction between the consequences of health-related stigma and a cross-cultural theory of health related-stigma . For example, stigma experiences and practices associated with HIV and NTDs might manifest similarly, yet they are products of entirely different structural drivers and facilitators . This exemplifies the need for a biosocial perspective of COVID-19-related stigma, locating it within wider structural and institutional systems as opposed to narrowing understandings to what happens at intra-and interpersonal levels . Pressure towards continued focus and attention can come from communities themselves and, in part, stigma can serve as an important catalyst for doing so. As noted above, affected groups can become agents of change with stigma serving as a catalyst. What we have learned from HIV activism over many years is that struggles are on-going; they require enduring attention, often beyond the mantras of inclusive policy actions or anti-stigma frameworks in the international community. Better conceiving stigma can provide a model of addressing such concerns, helping to promote meaningful health promotion action in terms informed by understandings of the relation between lived experience, structural vulnerabilities and knowledge communities. Crises typically entail the coming together of experts in attempts to define and resolve a problem to hand-such as a virus and its associated social, political economic effects. What we advocate for here is a conscious broadening of expertise, to include sociological and other experiential knowledge about stigma as it informs life-worlds of inequality within which COVID-19 is intrinsically interconnected. --- Conclusion It is widely acknowledged that stigma negatively impacts health practices and outcomes, reinforces structural inequalities and undermines equitable development. Emerging evidence suggests that stigma associated with COVID-19 is unfolding rapidly, both manifesting itself as a new stigma and facilitating the expression of existing stigma. Against this backdrop, this article aimed to assess the driving factors and markers of COVID-19-related stigma and to reflect on how the response to COVID-19 can counteract such stigma, particularly in LMICs. We did so by drawing on lessons learned from other disease-related stigma and stigma markers such as old age, disability and poverty in LMICs. We find that COVID-19-related stigma is already taking hold across LMICs. Rapid spread of misinformation fuels misconceptions and false beliefs about how the virus spreads and who may be 'carriers' of disease. This in turn feeds already strong fears of contagion, offering further fertile ground for stigmatisation of those who have been or are suspected of being infected by the disease or risk infecting others. These dynamics are further reinforced by efforts to contain the virus that hinge on separating and isolating vulnerable groups and infected individuals. Poverty, precarious and unstable living conditions, ethnicity and origin, age and gender present fault lines along which stigma is compounded and reinforced. A rights-based and inclusive approach to policy that includes the principles of participation, non-discrimination, and accountability is vital. COVID-19 policy and health programmes need to focus on disease elimination or transmission interruption, but they must also 'count what counts' through research into stigma and intersecting vulnerabilities, as well as the conditions which produced the stigma in the first place. Accounting for and monitoring the interplay between factors at the interand intrapersonal, institutional and structural levels is vital for moving beyond a topdown and individualistic approach. Before stigma policy and programmes can be implemented, we must explore what rights violations have contributed to COVID-19 stigma and what new violations have emerged due to COVID-19 stigma. In sum, COVID-19-related stigma in LMICs is a real force to be reckoned with. Not only does it undermine basic levels of dignity, it also causes new types of vulnerabilities, reinforces pre-existing inequalities and ultimately poses a threat to effective containment of COVID-19. In order to be equitable and effective, the response to COVID-19 in LMICs requires the issue of stigma to be placed front and centre to its approach. --- Compliance with Ethical Standards Conflict of interest On behalf of all authors, the corresponding author states that there is no conflict of interest. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

COVID-19 has caused unprecedented health, economic and societal impacts across the world, including many low-and middle-income countries (LMICs). The pandemic and its fallout have laid bare deep-seated social and economic inequalities with marginalised groups being at greater risk of infection and being disproportionately affected by containment measures and their socioeconomic consequences. Stigma is a central element to such inequalities but remains largely overlooked in the debate on the response to COVID-19, including in LMICs. Yet we know from experiences with other infectious diseases such as HIV/AIDS and Ebola that disease-related stigma is detrimental to halting and controlling pandemics and achieving equitable development. Emerging evidence suggests that stigma associated with COVID-19 is already taking hold. This paper assesses potential driving factors of COVID-19-related stigma, and how this intersects with existing stigma fault lines and explores mechanisms through which COVID-19-related stigma may be counteracted, with a focus on LMICs.La COVID-19 a eu un impact sanitaire, économique et sociétal sans précédent à travers le monde, y compris dans de nombreux pays à revenu faible ou intermédiaire (PRFI). La pandémie et ses retombées ont mis à nu de profondes inégalités sociales et économiques, les groupes marginalisés étant plus exposés au risque d'infection et étant touchés de façon disproportionnée par les mesures de confinement et par leurs conséquences socioéconomiques. La stigmatisation est au centre de ces inégalités mais reste largement négligée dans les débats sur la réponse à la COVID-19, y compris dans les PRFI. Grâce à l'expérience que nous avons sur d'autres maladies infec-* Keetie Roelen

Background In recent years, the ageing trend has become prominent in China. In 2018, it was recorded that there were 249 million people more than 60 years old and 167 million people more than 65 years old, accounting for 17.90 and 11.90% of the total population of the country, respectively. The proportion of the population aged over 65 years old has increased year by year, and the old-age dependency ratio has also increased year by year [1]. The number of people aged more than 60 years old is expected to increase to about 255 million by 2020, accounting for 17.8% of the total population; and the oldage dependency ratio is expected to increase to around 28%; the number of elderly people of venerable age is expected to reach up to 29 million, while the elderly living alone and the empty-nested elderly are expected to reach up to 118 million [2], according to China's 'Thirteenth Five-Year Plan' for the Planning of Developing the Aging Industry and Constructing the Endowment System. As the ageing trend accelerates, the number of elderly people and the disabled and semi-disabled elderly will increase in China. There were more than 40 million disabled and semi-disabled elderly people in China at the end of 2016 and 7% are cared for long term by their family. Those who need direct care also desperately require the involvement of medical services [3,4]. Compared with developed countries in Europe and the United States, China shows a unique ageing characteristic of getting old before getting rich, at a larger scale, at a faster speed and showing a heavier dependency burden etc. What is more, China's pension security system needs to be improved and Chinese society is facing tremendous pressure from the challenge of an everincreasing ageing trend. With the basic national family planning policy and economic and social transformation in play, the family supporting function has been weakening, while elderly people's demands for professional nursing institutions and community services have been steadily on the increase. In particular, people aged over 80 years old with a high morbidity rate of chronic disease desperately need systematic, comprehensive, convenient and low-cost medical services. Moreover, both medical and elderly care, as a matter of record, are indispensable for the disabled and semi-disabled elderly [5]. The elderly's medical needs cannot currently be satisfied either in most of the old-age nursing institutions that provide low-level medical services with only a few qualified nursing staff and limited beds or in medical institutions which cannot provide long-term hospitalisation services for the elderly due to their limited resources. Besides, care resources at the community level cannot fully cover the medical and nursing demands of the disabled and non-disabled elderly suffering from diseases. The ageing of the population has exacerbated the shortage of resources for medical services and elderly nursing, which has put forward a request for improving the allocation and utilisation of social resources. The traditional elderly nursing model cannot satisfy the allround care needs of the elderly. It is therefore imperative to implement a new model of health care for the elderly. Nevertheless, medical treatment and eldercare resources are inadequately supplied and mutually independent, which, therefore, cannot meet the needs among the elderly nowadays. Therefore, it is of great necessity to provide them with a 'medical-nursing combination' service that organically combines medical and elderly care. Since there is no standard definition of a 'medicalnursing combination' in China, it is defined differently by a wide range of scholars. Guo et al. believe that the 'medical-nursing combination', denotes a process that gradually forms a cooperative service model integrating medical treatment, recovery and nursing from service providers providing a medical and nursing service conforming to elderly nursing to the elderly in demand according to different health needs at different stages of diseases [6]. Liu et al. defined the combination of medical care and nursing as satisfying the needs of health problems at different levels for the elderly at different stages in the care process, through integrating medical resources and pension resources to optimise the allocation of medical and nursing resources [7]. Liu et al. consider that elderly people can achieve the purpose of obtaining medical treatment while suffering from diseases, and enjoying care while not suffering from diseases under the new elderly nursing model combining medical and elderly care [8]. To Huang et al., the medical-nursing combination possesses the same concept as 'long-nursing' overseas, which focuses on meeting the basic living needs of the elderly, as well as physical and psychological care; moreover, medical treatment should be highlighted, while the enhancement of daily living skills, the adaptation of social environment and the realisation of self-worth are also important [9]. To solve the medical problem of the ageing population, the concept of a medical-nursing combination was first proposed in 'Several Opinions on Accelerating the Development of the Elderly Nursing Service Industry' issued by the State Council in September 2013. It pointed out the need to provide multi-level elderly nursing services, actively respond to the ageing population and accelerates the development of the elderly nursing service industry through actively driving the combination of medical and elderly care services. 'Guiding Opinions on Promoting the Combination of Medical Treatment and Elderly nursing Services' issued by the State Council in November 2015 indicated two tasks for promoting the combination, firstly, encouraging elderly nursing institutions to conduct various forms of agreement and cooperation with other medical and health institutions and establish a sound cooperation mechanism; secondly, promoting the extension of medical and health services to communities and families. The 'Thirteenth Five-Year Plan' for the Planning of Developing the Aging Industry and Constructing the Endowment System issued by the State Council in March 2017, focused on assigning nine tasks including the active promotion of the medicalnursing combination service and improving the allocation and utilisation of social resources. As of 2017, China has set up 90 national-level pilot cities for combining medical and elderly care [10]. On that basis, the 'medical and elderly care combination' is a new elderly nursing model that provides the elderly with services such as uninterrupted daily care, mental consolation, disease diagnosis and treatment, health guidance, recovery from serious illnesses and hospice care through effectively integrating medical and elderly care resources to satisfy the varied health care needs of the elderly at multiple levels. Currently, four medical and elderly care models can be found in China [11]. The first model is 'nursing in hospital', that is, a geriatric department is set in some large hospital with conditions to provide medical treatment, nursing, care for the elderly, rehabilitation, health education, hospice care and similar services; or some low-level primary hospitals with spare resources are transformed into nursing institutions for medical rehabilitation, convalescence and elderly nursing, to achieve the goal of integrating medical and elderly care. The second model is 'constructing a hospital in nursing institutions', and providing professional medical and nursing teams according to the standards of national hospitals in large-scale elderly nursing institutions or welfare homes. Meanwhile, basic medical departments such as a comprehensive medical-surgical department, rehabilitation department and pharmacy are set up to form a new elderly nursing institution, integrating elderly nursing with healthcare functions. The third model is the union of medical and elderly care, namely, a cooperation mechanism is established between medical institutions and elderly nursing institutions. In this way, medical institutions provide medical care training to nursing staff in elderly nursing institutions, and regularly conduct basic diagnosis and treatment services such as detection of common diseases, the management of chronic and geriatric diseases, as well as health education. Meanwhile, the hospital also offers a green channel to provide a timely medical referral service for the elderly in need, and conduct subsequent recovery treatment in the eldercare institutions after the individual's condition is controlled. By doing so, a two-way continuous care model is generated. The fourth model is 'home nursing', which is, in essence, a family doctor model. A service team provides outpatient services and life nursing services for the elderly. It is a model that is primarily designed for the elderly in good health, allowing them to enjoy their old age in peace with familiar surroundings. As developed western countries entered the ageing society paradigm earlier than China, they have developed a new elderly nursing model called 'long-term nursing' that is consistent with the medical-nursing model in concept, connotation, service purpose, content and objects. The United Kingdom, the United States and Japan are the most typical countries that have developed their unique representative research results concerning elderly nursing. The elderly nursing model in the United Kingdom is dominated by community and home care. The main service providers are composed of managers, professional staff and caregivers, who provide four major services including life care, material support, psychological support and overall care. Specifically, life care is mainly to provide home-care services and short-term care services for the self-care or semi-selfcare elderly; material support includes the government upgrading the infrastructure of the elderly's living place and providing tax subsidies or preferences to taxpayers more than 65 years old; psychological support is where service staff visit the elderly for health inspection, publicising health care knowledge, making rehabilitation and treatment suggestions and providing psychological counselling; overall care is delivered in community activity centres funded by the government or the society and is designed to inject fun into elderly people's lives, and some low-intensity jobs are provided to increase the elderly's income and maintain their mental health [12]. The elderly nursing model combining medical and elderly care in the US is dominated by a programme of allinclusive care for the elderly that is set up for the disabled, the semi-disabled and over-55-year-old low-income groups requiring long-term medical care. Covering medical services, rehabilitation services and social support services, the purpose of PACE is to assist the elderly and the debilitated to live as long as possible in the community or family, improving the living quality of the elderly with weak self-care ability and maximally protecting the dignity of the elderly [13]. The elderly nursing combining medical and elderly care model in Japan is dominated by the following models. First, the daycare centre; this model mainly provides rehabilitation and life care services for the elderly of more than 65 years old who are unattended at home in the daytime and need rehabilitation training. Second, the nursing centre; this is supported by a service consisting of nurses, caregivers and welfare workers to provide daily services for the disabled elderly living in the centre. Third, the elderly welfare centre is targeted at the elderly in the community; service staff, mainly health care therapists, provide services including health examination, health education, health care services and family guidance. Fourth is the apartment for the elderly, which is mainly designed for the healthy elderly who can take care of themselves. It provides basic medical services and daily care services. An all-round legal system is a major reason why the elderly nursing combining medical and elderly care has been well developed in Japan [14]. Research on the 'combination of medical and elderly care' have been identified abroad with proven systems moving from policies to actual services, which can provide a reference and basis for researching and implementing the combination of medical and elderly care in China. Researching the 'combination of medical and elderly care' service model is still in its initial stage of development in China. Related research studies in the domestic literature focus on introducing and analysing foreign elder care cases based on the 'combination of medical and elderly care', which propose the status quo of the development of the 'combination of medical and elderly care' model before making suggestions or conducting a case study in the pilot region of combining medical and elderly care in China. Nevertheless, few studies cover the service requirement and influencing. Instead, most of the research discusses elderly people's basic situation, health status, social support and income status, etc. Generally speaking, the better the health condition of the elderly, the higher the self-care level, the lower the income and the lower the social support, the less demand for a service combining medical and elderly care [11,15]. Li et al. found that the number of children, health status, children's support and willingness to pay have significant impacts on the demand for services combining medical and elderly care after investigating more than 420 elderly people aged more than 60 years in four major urban areas of Chongqing [15]. Hu et al. discovered that degree of education, ideal method of elderly nursing and willingness to pay are significant factors affecting the elderly's demand for the service combining medical and elderly care in urban areas after surveying the elderly in Yinchuan [16]. According to Zhou et al.'s research, age, degree of education, number of children and occupation type before retirement are main factors affecting the elderly's demand for the new service [17]. Through investigation, Wang et al. believe that the elderly in Changchun have a high willingness to participate in a combined service. Gender, age, education and occupation type are major factors affecting their choices [18]. The basis of medical insurance in China is comprised of a basic medical insurance system for urban workers, a basic medical insurance system for urban residents and a new rural cooperative medical insurance [19,20]. A unified basic medical insurance system for urban and rural residents should be gradually established nationwide according to the Opinions on Integrating the Basic Medical Insurance System for Urban and Rural Residents issued by the State Council in 2016. The number of people insured with basic medical insurance in China has exceeded 1.35 billion with a participation rate of over 95% by the end of 2017, basically realising a full coverage from 'insurance for few' to 'insurance for all' [21]. The basic medical insurance system for urban employees is raised jointly by social medical unified planning and individual account, forming a social medical unified planning fund and individual medical account fund. The individual account is not set in the basic medical insurance system for urban and rural residents. In other words, only the social medical unified planning fund is established to raise funds through quota. The premiums consist of individual residents' contributions and financial subsidies. Most scholars in our country believe that the demand for the service of combining medical and elderly care is affected by the design and implementation of medical insurance systems and the elderly's capacity to pay under the current medical insurance system. In terms of system design, there is a lack of long-term care insurance specifically for elderly nursing, and the elderly nursing service combining medical and elderly care is not involved in the designated medical insurance units. Medical insurance in China focuses on economic compensation for the loss caused by a given disease and lacks compensation for preventive health care, rehabilitation, long-term care and similar services needed by the elderly, whereas basic pensions are mainly used for daily care of the elderly [22]. On the aspect of system implementation, the reimbursement practice of medical insurance in China is characterised by 'designated medical care with three medical directories'. Since setting up medical institutions in nursing institutions is not included in the designated medical organisation, additional medical services in the nursing institution cannot be paid through medical insurance. In this case, the elderly living in the nursing institution have to visit hospitals for treatment, reducing access to medical services [23]. Besides, a plurality of issues such as the admission of nursing institutions, the verification of medical qualifications, medical insurance designated hospitals and review and distribution of charges can be found in the nursing institution involved in combining medical and elderly care [24]. Regarding the elderly's capacity to pay, the medical insurance only covers medical expenses and examination costs incurred during the medical process. With the lack of a long-term care insurance system, rehabilitation medical programmes, life care programmes and auxiliary equipment programmes are fully incurred by the elderly. However, elderly patients who are economically disadvantaged, especially the disabled and the semi-disabled, the elderly suffering from diseases and more than 80 years old have a limited capacity to pay for the long-term care cost [25]. The elderly are the main service object of the 'medical-nursing combination' model, whose demand willingness plays a decisive role in the development of the 'combination of medical and elderly care. Therefore, it is essential to proceed from the elderly's demand for specific services before conducting an in-depth exploration of the new model. As of the end of 2018, the number of elderly people aged 65 years old or above in Lanzhou has reached 498,800, accounting for 16.50% of the total population [26]. Moreover, the proportion of the population aged 65 years old or above has been higher than the average level of the whole country and Gansu during the same period . As can also be seen from Fig. 1, the degree of the ageing phenomenon has become more serious in Lanzhou from 2010 to 2018, in which the ageing rate was 8.20% in 2010 and jumped to 16.50% in 2018, indicating that the growth of the ageing population has been accelerating by 8.3%. By comparison, the national ageing rate was 11.9% in 2018, which clearly shows that the ageing rate in Lanzhou had accelerated. What is worse, the ageing problem in Lanzhou would be crucial, as the degree of ageing population could become serious over time. The accelerating population ageing in Lanzhou has put tremendous pressure on elderly nursing. Furthermore, elderly nursing involves a variety of requirements such as medical rehabilitation and spiritual happiness with the social progress, rather than merely basic daily care. It can be seen that a tremendous requirement has been proposed to develop multi-integrated nursing services combining nursing and medical treatment based on the huge elderly group and the serious ageing status quo in Lanzhou. Incomplete statistics show that there are 27 nursing institutions in Lanzhou as of now, including seven institutions run publicly and 16 run privately, and four institutions combining medical and elderly care, providing a total of 6107 beds. Specifically, 18 hospitals have set up geriatric services and geriatric beds, providing a total of 500 beds, accounting for 69% of the total number of hospitals; 26 hospitals above the county level have set up green medical treatment channels for the elderly; and 19 nursing institutions can provide medical services, accounting for 70.4% of the total nursing institutions; the contracted service rate of the home-based elderly aged more than 65 years old in Lanzhou reached 73% [29,30]. As a national pilot city for combining medical and elderly care, Lanzhou has made some progress in the process of developing a service combining medical and elderly care. However, the follow-up work remains cumbersome since the policy obstacles in the combination of medical and elderly care should be overcome. What is more, concrete service content and links should be improved, such as constrained nursing conditions in medical institutions, missing service function of nursing institutions, high cost, constrained reimbursement of medical expenses, pessimistic cognitive status of the concept of combination of medical and elderly care and the institution management system requiring enhancement [30]. Scholars tend to be more willing to concentrate on the process and obstacles of combining medical and elderly care at the macro level for such a new type of elderly nursing. However, few studies analysing the elderly's needs for a combined medical and elderly care services can be found. The social and economic foundation of the undeveloped region of western China is relatively weak with a low level of social security and welfare. In particular, the elderly long-term care system in remote rural areas is in its infancy. Lack of a well-defined medical-nursing mechanism seriously affects the well-being and happiness of residents in the area. As the driving strategy of combining medical and elderly care has been vigorously promoted at the national level, theoretical introduction and countermeasure are essential to regional strategic layout. A questionnaire survey concerning the need for an elderly nursing service combining medical and elderly care was conducted on residents in Lanzhou. On this basis, the specific needs of residents for elderly nursing services combining medical and elderly care were analysed. By sorting out factors affecting the demand for elderly nursing service combining medical and elderly care, policy proposals were made accordingly. The case study of Lanzhou was taken as an example to provide a referring significance of developing a combination of medical and elderly care in the undeveloped region of western China, to raise the health care levels of residents in the undeveloped region of western China, satisfy their medical and nursing requirements and improve their nursing services. --- Methods --- --- Design and procedure Questionnaire concerning elderly people's needs for combined medical and elderly care: Questionnaires covering personal characteristics, health status, economic status and elderly people's understanding of a combined medical-nursing service, have been frequently adopted by scholars at home and abroad for investigating the need for an elderly nursing service combining medical and elderly care. To better compare with the results of related literature [11][12][13][14][15][16][17][18][31][32][33][34][35][36][37][38][39][40][41][42], and perform an in-depth analysis of factors affecting the need for combining medical and elderly care, a questionnaire related to the eldercare needs of combining medical and elderly care has been designed by epidemiological and statistical experts and jointly compiled by the research team and elderly nursing management experts based on a full reference to relevant literature [11][12][13][14][15][16][17][18][31][32][33][34][35][36][37][38][39][40][41][42] and national health policies. The in-house designed 'Survey on the elderly's needs of combination of medical and elderly care in Lanzhou' was used as a survey instrument. The questionnaire was designed following the literature [11][12][13][14][15][16][17][18][31][32][33][34][35][36][37][38][39][40][41][42], the China Health and Pension Tracking Survey and the Fifth National Health Service Survey-Family Health Questionnaire, as shown in Table 1. Before the questionnaire was officially distributed, 100 retired residents were selected for pre-survey through convenience sampling. And a formal questionnaire was finalised after the questionnaire was revised according to the pre-survey results. The questionnaire is composed of general demographics, health status, medical endowment insurance and the knowledge of and expressed a Fig. 2 The geographical location of the study area in China. requirement for combining medical and elderly care. A detailed questionnaire is provided by us as Supplemental Materials. See the Supplemental Materials. --- Quality control method The survey was conducted face to face by trained investigators familiar with working with elderly respondents. Investigators explained related concepts in the questionnaire during the survey considering that the elderly were not understanding the combination of medical and elderly care. Questionnaires were collected and checked by a specially-assigned person who was responsible for removing the invalid and the incomplete questionnaires. Questionnaires that qualified based on the investigators' reviews were retrieved in time. --- Statistical analyses Software Epidata3.1 was used to log and proofread data in duplicate. Statistical software SPSS 18.0 was used for statistical treatment. The enumeration data used χ 2 test, and the influencing factors were analysed with a binomial logistic regression. Statistical significance in the differences would be confirmed in case of P < 0.05. the service combining medical and elderly care) for conducting binomial logistic regression analysis. The relationship between the elderly's demand for the service and various variables were obtained, as shown in Table 3. --- Results --- Sociodemographic characteristics of senior citizens in Lanzhou Influencing factors were screened using the stepwise regression method. The inclusion criterion was α = 0.05 and the removal criterion was α = 0.1. The results of logistic regression on the combined service need are shown in Table 4. According to regression results, major influencing factors are comprised of the number of children, health self-rating, type of medical insurance, current model of elderly nursing, elderly nursing need, self-care ability of daily living, prior or acquired knowledge of the proposed service model and willingness to pay for the combination of medical and elderly care . --- Discussion Comparison of the understanding of and demand by senior citizens for the combination of medical and elderly care services Surveyed data of this study show that only 8.13% of the 7320 elderly people aged more than 60 years old in Lanzhou knew about and understood the service combining medical and elderly care. Among senior residents in Datong 25.0% were aware of the model of combining medical and elderly care, while only 6.0% understood well the model of combined medical and elderly care [11]. As can be observed from the study conducted by Hu et al. in Yinchuan, only 7.6% of the residents had a basic knowledge about the elderly nursing model of combining medical and elderly care and only 0.5% understood the combination model well [16]. Studies conducted by Zhou el al. in Urumchi show that 13.67% of the residents had heard of or at least knew about the service [17]. According to research conducted by Wang et al. in Changchun, 2.54 and 6.21% of the elderly people in the community well understood and had a basic knowledge of the elderly nursing model combining medical and elderly care, respectively [18]. The study conducted by Liu et al. in Karamay found that only 11.30% of the local residents knew and understood the proposed combined model [37]; and the study conducted by Wang et al. in Beijing shows that 12% of the elderly residents understood the model, while only 4.0% of those understood and actually used the combined model [38]. The results shown in this study are basically consistent with the above findings, indicating a low knowledge rate of the combined service model. The combination of medical and elderly care is still at the exploratory stage in China and few investigations could be found on the demand rate for some combination of medical and elderly care, since most of the related research is dominated by theories, problems and countermeasures. This survey shows that the demand rate for the service combining service provision for the elderly in Lanzhou is 51.53%. The elderly people's demands for this service were also reported in some domestic studies, for example, the demand for the elderly nursing service combining medical and elderly care for the elderly in cities and towns in Datong makes up 50.5% [11]; the demand for the service for the elderly in Chongqing is 53.00% [15]; the demand for the service for the elderly in Yinchuan accounts for 50.80% [16]; 56.21% of the elderly in Changchun demand the combined service [18]; the demand for the service for the elderly in downtown Zhanjiang is 54.60% [31]; 61.10% of the empty-nested elderly in Quanzhou communities need the service [32]; the demand for the service for the elderly in Karamay is 53.01% [37]; the demand for the service for retired residents in Tianjin is 61.9% [39]; a strong demand for the elderly nursing service combining medical and elderly care for elderly patients can be witnessed in Weifang, accounting for 97.40% [40] and 43.40% of the elderly in Qiqihaer have a demand for the combined service [41]. Compared with the above studies, a lower demand for the elderly nursing service combining medical and elderly care can be observed in Lanzhou. It is generally recognised that the knowledge rate and the service demand for combining medical and elderly care are directly proportional, namely, the higher the knowledge rate, the higher the demand for the service combining medical and elderly care [31,42]. For example, the research conducted by Wu et al. in Zhanjiang proves that elderly people's knowledge rate of the service combining medical and elderly care in downtown Zhanjiang is 59.40%, and the demand rate for the service combining medical and elderly care is 54.60% [31]. Nevertheless, some studies also show that the knowledge rate of the service is not a factor influencing the demand for the service. For example, the studies conducted by Hu et al. in Yinchuan and by Liu et al. in Karamay show that the knowledge rates of the service are 7.60 and 16.11%, respectively, and their corresponding demand rates for the service are 50.80 and 53.01%, respectively [16,37]. These figures are inconsistent with the knowledge rate and the demand rate for the service for the elderly in Lanzhou. The possible cause of the high demand rate is that Yinchuan, Karamay and Lanzhou are located in the undeveloped region of western China with a low level of economic and social development, and a high degree of an ageing population. The low knowledge rate of the service is affected by a plurality of factors, which is primarily due to the lack of publicity from the perspective of policy. The elderly aged 60 years old or above receive information via different channels from those of younger people. Effective channels that are accessible by the elderly, including television, newspapers, communities and nursing institutions should be selected for publicising the combined service, instead of new media such as the Internet, Sina blog, Weibo and WeChat etc. The publicity should also cover the elderly people's children and family members. Government departments are responsible for strengthening the publicity in this regard. Favourable policies should be released promptly and rigorously publicised so that the elderly can obtain the favourable information quickly and effectively utilise various preferential policies provided by the country. --- Influencing factors on the demand for the combination of medical and elderly care by senior citizens in Lanzhou The willingness to pay for the combination of medical and elderly care The demand rate of the elderly willing to pay 2000 to 3000 yuan per month for purchasing the service is 49. 64%; the demand rate of seniors who are willing to pay more than 3000 yuan per month for purchasing the service is 64.50%; and 36.54% of the elderly are willing to pay no more than 1000 yuan per month for the service. According to the investigation conducted by Fan et al. on the elderly in the cities and towns of Datong, more than 50% of the elderly are willing to pay 1000 to 2000 yuan per month; 35% of the elderly are willing to pay 2000 to 3000 yuan per month; 13% of the elderly are willing to pay less than 1000 yuan per month, and only a few elderly people are willing to pay more than 3000 yuan for the service [11]. After surveying the demands of the service for the elderly in downtown Chongqing, Li et al. show that the demand rate for the service for the elderly who are willing to pay 500 to 999 yuan per month is 38.7%; the demand rate for the service by the elderly who are willing to pay 2500 to 2999 yuan per month is 8/9, and 31.0% of the elderly are willing to pay no more than 500 yuan per month for the service [15]. The result obtained by Hu et al. surveying the elderly in Yinchuan shows that 39.59% of the respondents are willing to pay 500 to 999 yuan per month for the service; 32.99% of the respondents are willing to pay less than 500 yuan per month for the service, and only 27.41% of the respondents are willing to pay 1000 yuan or above per month for the service [16]. It is different from the result obtained in this study, which might be due to different income levels in different regions. The average monthly income of residents in Lanzhou in 2018 was about 6800 yuan [28], which was higher than that in Datong , Chongqing and Yinchuan [26]. In that case, the elderly express more willingness to pay for the service in Lanzhou than in the other three cities mentioned above. Furthermore, the demand rate for the service in Lanzhou is on the rise with the rising price that is willing to pay, which is similar to related literature [11,15]. This is due to conditions such as social status, income and degree of education of the elderly with a high willingness to pay are better than those with a low willingness to pay [11,15]. --- Type of medical insurance In this study, the type of medical insurance is a factor influencing the demand rate of the service combining medical and elderly care for the elderly. The elderly with urban medical insurance have an increasing demand for the service. In general, the cost of nursing the elderly in nursing institutions is higher than home-based nursing. Also, the reimbursing proportion of urban medical insurance is high. Although many nursing charges are not included in the reimbursement category, costs incurred by the service combining medical and elderly care can be covered by urban medical insurance, which can alleviate part of the costs for the service combining medical and elderly care. In consequence, the elderly with urban medical insurance are more willing to select an institution with the elderly nursing service for their nursing needs. Some studies indicate that medical insurance is a factor influencing medical rehabilitation and health care services for the elderly [43]. Another study demonstrates that the elderly participating in the new rural cooperative medical system are more willing to have the service, in comparison to those with the medical insurance for urban residents [44]. It is probably because of different reimbursement proportions of various medical insurance systems. --- Number of children This study shows that the number of children is a factor influencing the demand rate of the service. The elderly with fewer children have an increasing demand for the service. This may be caused by the family miniaturisation and the increase in empty-nest families. In this way, the children cannot take care of their elderly parents. Generally speaking, the elderly with more children have richer family-supporting resources, more secure familysupporting care and lower demands for the service combining medical and elderly care. When a core family is formed, the number of elderly people requiring care is increased, while the number of family members who can take care of the elderly has decreased. In that case, the elderly cannot be cared for in an all-round way. What is worse, family members could be helpless in handling the special requirements of treatment, nursing, rehabilitation and hospice for the disabled elderly, the elderly with chronic diseases and being susceptible to disease as well as those suffering from terminal diseases. Therefore, the uncared for elderly person who has difficulty in medical treatment due to a few children or children not around has a higher demand for the service combining medical and elderly care. --- Level of awareness of the combination of medical and elderly care The knowledge rate of the service combining medical and elderly care in this study is one of the factors influencing the demand for the service. As can be seen from the studies conducted in Zhanjiang and Urumqi, the elderly who knew and understood the service have a higher demand rate for the service in comparison to those who had never heard of the service [31,36]. However, the knowledge rate of the service is greatly inconsistent with the demand rate of the service. It is Moreover, some of the elderly cohort cannot understand the service mode, service purpose or service concept accurately and have doubts about the service model. Some even consider that the service is equivalent to the nursing service in traditional nursing centres and care homes [11]. The elderly nursing model proposed is a new elderly nursing model. The elderly with an in-depth understanding of the new model and its advantages are more willing to select the institution combining medical and elderly care for nursing. It shows that the development of the elderly nursing mode combining medical and elderly care is greatly correlated to the understanding of residents. Therefore, relevant government departments should publicise the elderly nursing mode well to enhance the knowledge and recognition of the new mode among the people. --- Self-assessment of health The elderly's health self-rating status in this study is a factor in measuring the health status of the elderly. The elderly's conscious need for elderly nursing or medical treatment can be reflected in the health self-rating status. According to the single factor analysis results, a statistically significant difference can be found in the health self-rating status in the demands of the new service . The elderly with poor health self-rating status have a higher demand for the service than those with better health self-rating status. The demand rate for the service for the elderly with poor health self-rating status is 52.78%; the demand rate for the elderly with not good enough health self-rating status is 52.27%, and the demand rate for the elderly with a good health self-rating status is 41.57%. After surveying the elderly in cities and towns of Datong, Fan et al. found that the elderly with a poor health self-rating status had a higher demand for the service; the demand rate for the service by the elderly with poor health status reached 66%, while the demand rate by those with good health status was only 37%. Another study conducted in Urumqi shows that the ageing population with poor health self-rating status have a higher demand for the service than those with good health self-rating status, and the demand rate for the service for the elderly with poor health self-rating status reached 68.74%; the demand rate among the elderly with not good enough health self-rating status was 46.19%; and the demand rate by the elderly with a good health self-rating status was 53.45%. The findings in this study are consistent with those arrived at in the above studies [11,36]. The elderly's physical function will gradually deteriorate with an increased incidence rate of various diseases and worsening health conditions with increasing age. What is worse, some of the elderly might not be able to take care of themselves. Along with these situations, there will be an increasing demand for the service of medical and elderly care. That is the reason why the elderly with poor health status have a strong demand for the service. --- Current elderly care model This study indicates that the current model of elderly nursing is a factor influencing the demand for the proposed new service for the elderly. Most of the elderly consider home-based care the ideal way of delivering elderly nursing due to the influences of their physical conditions, children's time and energy, local cultural habits, economic factors and psychological needs. Nevertheless, 42.13% of the elderly consider that there is a gap between the existing elderly nursing mode and the ideal one due to financial difficulties, unattended living, poor health and suffering from illness, no entertainment and loneliness and other reasons [16]. The mode combining medical and elderly care prioritises the elderly's health and medical services, which differs from the traditional way of providing high-quality elderly nursing services for satisfying the elderly's basic living needs. It is appealing to the elderly to some extent. Comparison between primary influencing factors on the demand for the combination of medical and elderly care in a different study We also compared primary factors influencing the elderly's needs for combining medical treatment and nursing in a different study. To ensure the comparability of the data, we collected data from literature reports using the same questionnaire as in this study [15,[31][32][33]. In the above literature, the influencing factors on the demand for the combination of medical and elderly care were analysed by binary logistic regression analysis. Therefore, we used the forest map to compare the differences between the factors affecting different studies. As can be observed from Fig. 3, the primary influencing factors of the elderly's demand for the service are varied. The primary influencing factor of the elderly's demand for the service is a willingness to pay in Lanzhou, children's support in Chongqing [15], the type of medical insurance in Zhanjiang [31], health management in Quanzhou [32] and age in Shihezi [34]. The above results show that related departments should formulate and develop planning and policies for the service following the actual situation. --- Conclusions A low level of understanding on the part of the elderly and the demand for the service combining medical and elderly care can be witnessed in Lanzhou. Meanwhile, the number of children, types of medical insurance and willingness to pay for the combination of medical and elderly care are major influencing factors among the complex factors influencing the elderly's demand for the service proposed Lanzhou. Relevant departments in Lanzhou should improve relevant laws and regulations and release systems and standards concerning the service according to local conditions, while vigorously enhancing the publicity of the service to raise the demand for it. What is more, the construction of institutions tasked with developing the service should be accelerated to safeguard the capacity of providing the service combining medical and elderly care. --- --- Supplementary information Supplementary information accompanies this paper at https://doi.org/10. 1186/s12877-020-01616-6. --- Additional file 1. Abbreviations OR: Odds ratio; PACE: Program of all-inclusive care for the elderly; 95%CI: 95% Confidence interval; US: United States Authors' contributions GXB, TKG, and HC conceived and designed the study. YHW, JXZ, and BP performed the data collection. YHW and BP managed the data and performed the analysis. YHW and BP were responsible for data analysis and interpretation. JCW and YHW wrote the initial draft of the paper. HC critically reviewed, revised and supplemented the manuscript. All authors read and approved the final manuscript. --- --- Consent for publication N/A. --- Competing interests The authors declare that they have no competing interests. ---

Background: This study aims to analyse the current demand by senior citizens in Lanzhou, China for a combination of medical and elderly care services and to identify the factors influencing their needs. Methods: 7500 participants aged 60 or above living in Lanzhou, China, were recruited, a unified questionnaire concerning elderly people's demand for a service combining medical and elderly care has been adopted to survey these subjects. The status quo of the demand of the service combining medical and elderly care and its influencing factors were analysed with the single-factor Chi-square test and multi-factor binomial logistic regression method. Results: 3772 of 7320 older people have the demand for the service combining medical and elderly care, accounting for 53.15% of survey respondents. Many factors are in play, including gender, marital status, degree of education, occupation before retirement, number of children, monthly income, health self-assessment status, endowment insurance type, medical insurance type, current nursing arrangements, old-age demands, self-care ability and the knowledge of combining medical and elderly care and the willingness to pay for the combination of medical and elderly care have statistical significance (P < 0.05) with the elderly's needs, different ages, living styles and the prevalence of chronic diseases, have no statistical significance (P > 0.05) with the elderly's care needs in Lanzhou. The number of children, type of medical insurance and willingness to pay for the combination of medical treatment and nursing care are major influencing factors among the complex factors influencing the elderly's demand for the proposed service.

Background Researchers have long been interested in whether the experience of being part of a research study can change the behaviour of participants. When these changes are due purely to the experience of being in a studyrather than as a result of any interventionthis is sometimes called a Hawthorne effect, measurement reactivity, or mere measurement effect [1][2][3]. Such an effect might occur as a result of filling in a questionnaire, being observed by a researcher, or taking part in an interview. This behaviour change might be the result of increasing participants' awareness by asking about a health behaviour, or through prompting participants to reflect on their own choices. Existing research suggests that being involved in a research study can produce statistically significant, though generally small effects in participants' health behaviours. A recent meta-analysis examined the effect of asking questions on participants' behaviour [4]. From 41 studies the authors found a small overall change in behaviour for those who had answered questionnaires or surveys compared to participants with no measurement, or those with other forms of measurement. However, studies did show publication bias, suggesting the effect size may be overestimated. Another meta-analysis examined the mere measurement effect on a more homogenous sample of studies: randomised controlled trials of brief interventions for alcohol consumption [5]. The review identified eight nested trials carried out to assess the impact of measurement on alcohol use independent of the intervention. The meta-analysis did not find a statistically significant effect , although the authors highlighted that the effect size was relatively large , and statistical power was low. These randomised studies are particularly valuable as they avoid a number of biases included in observational studies, and more such studies would help to increase the power for meta-analyses and provide a fuller picture of the role of mere measurement. In a narrative review, French and Sutton [2] found evidence of small, but significant effects from mere measurement. For example, one study was a cluster randomised controlled trial of physical activity promotion in which participants in the control and intervention groups were further randomised either to have measurements taken at baseline, eight weeks and six months, or just at six months. The measurement was a thirteen-page questionnaire, and height, weight and waist circumference were also measured at baseline. The number of people meeting the recommended exercise level was 50% higher in those who filled out the questionnaire three times, compared to those who filled it out once . No significant association was found, however, for the proportion of participants spending at least 150 min per week on physical activity [6]. Using patient information gathered via questionnaires is popular both in the areas of practice oriented research in psychotherapy [7] as well as in more general health contexts in order to collect patient reported outcomes [8]. In these settings repeated assessments during the course of care are obtained from the individual patients and generally these assessments, if fed back to clinicians, have been shown to improve treatment outcomes especially for patients that were on a trajectory of stagnation or even deterioration [9,10]. Whether this effect is due to changes in practice or just the repeated assessments is a topic of current debate [11][12][13]. Many of the health behaviours examined in studies of mere measurement have a social gradient, whereby people who are more advantaged in income or education have better health behaviours than those lower down the social ladder. These include behaviours such as physical activity [6] or smoking [1]. However, there is almost no evidence of how, if at all, mere measurement differentially affects people of different socioeconomic status . Given that many of the studies of mere measurement are nested within larger studies it is likely that data on SES were available, but this was rarely studied. One exception was found: a randomised controlled trial by van Sluijs et al., measuring the effect of surveying physical activity, found a significant change in behaviour due to measurement, but no significant change in the model when adjusting for confounders, among which were employment and education [6]. Very little research has been carried out exploring the causal processes behind mere measurement, beyond a general raising of awareness [2]. It is possible that mere measurement could prompt behaviour change through similar processes as health promotion activities. Clearly, the comparison is not exact; unlike mere measurement, health promotion generally involves the provision of information or explicit suggestions for behaviour change. However, it has been shown that information on its own is often ineffective at changing behaviour [14]. In the absence of direct studies on mere measurement, then, the way in which health promotion interacts with SES was examined. One systematic review explored the effect of health promotion and education campaigns according to SES [15]. Studies examining the effects of accident prevention schemes, educational books for pregnant women and immunisation found that people with higher SES improved more, so that the overall improvement was at the expense of widening health inequalities. In the US, Pickett et al. found that the widely commended 'Back to Sleep' campaign to prevent Sudden Infant Death Syndrome not only increased race and SES inequalities in rates of SIDS, but the odds ratio for SIDS associated with lower SES actually increased during the campaign [16]. The history of smoking is one example of the inequitable effects of health promotion. Although there was no clear social gradient in smoking in the first half of the twentieth century, a social gradient emerged with the 'Smoking Kills' campaigns [17,18]. Smoking prevalence changed very little in the lowest social class quintile between 1973 and 1998, whereas all other quintiles saw significant reductions [19]. An evaluation of England's smoking cessation services in 2005 found a similar trend. Out of those smokers who took up smoking cessation services, disadvantaged groups had cessation rates of 8.7%, compared to a rate of 17.4% in the most advantaged groups [20]. If mere measurement and health promotion affect health behaviours through similar causal pathways then the negative effects of health promotion on equality raise the question of whether mere measurement, too, may increase health inequalities. This study therefore aimed to address two sequential questions: 1. Does simply taking part in a research study improve participants' health behaviour? 2. If so, is this effect mediated by socioeconomic status? --- Methods The Born in Bradford family cohort study Born in Bradford is a longitudinal multi-ethnic family cohort study aiming to examine the impact of environmental, psychological and genetic factors on maternal and child health and wellbeing [20]. Bradford is a city in the North of England with high levels of deprivation and ethnic diversity. Approximately half of the births in the city are to mothers of South Asian origin. Women were recruited while waiting for their glucose tolerance test, a routine procedure offered to all pregnant women registered at the Bradford Royal Infirmary, at 26-28 weeks gestation. For those consenting, a baseline questionnaire was completed via an interview with a study administrator 1 . The baseline questionnaire for the mothers was transliterated into Urdu and Mirpuri using a standardized process, so that words and phrases corresponded with the original English version. As Mirpuri does not have a written form trained bilingual interviewers administered the transliterated questionnaires to Mirpuri speakers. The full BIB cohort recruited 12,453 women during 13,776 pregnancies between 2007 and 2010 and the cohort is broadly characteristic of the city's maternal population. --- Study design During the course of the cohort study some women became pregnant more than once. These women were invited to include their additional babies in the cohort and fill out another baseline questionnaire. This feature of the cohort was used to assess the effect of BIB on mothers' behaviour during pregnancy, as the health behaviours of mothers having their second baby in the cohort could be compared with the health behaviours of mothers enrolled into BIB for the first time , who had not yet been exposed to the study. In our sample, health behaviours differed according to parity 2 . For example, using data from all BIB P1s, the odds of smoking during pregnancy decreased with each additional child, with an adjusted odds ratio of 0.86 . We therefore restricted the study to include only women on their second pregnancies. The study design is presented in Fig. 1. The control group is circled on the left hand side. These women had one pregnancy before BIB started, and a second child who was included in BIB. Their BIB P1 acts as our control. Their baseline questionnaire was filled out as they entered the study, so being part of the study should not have influenced their responses. The exposed group is circled on the right hand side. These are women who had two children and both of them were included in the BIB study. They filled in the BIB questionnaire for each pregnancy. The data collected during their second pregnancy in the cohort acts as our exposed group because they have been exposed by filling out the BIB questionnaire during their first pregnancy . This study used only their responses from the questionnaire completed during their second BIB pregnancy. The data from their first BIB pregnancy was disregarded, so the control and exposed groups consist of different women. --- Outcomes and data collection Health behaviours The current official NHS advice is not to smoke, to consume only small quantities of caffeine during pregnancy and to take vitamin D supplements 3 . Women are also advised not to drink alcohol, although there is some leniency in the consumption of small quantities. Two measures for alcohol were therefore included to reflect differing levels of alcohol consumption. Five health behaviours were therefore analysed, all of which were dichotomous variables: Smoking. Women were coded yes if they reported smoking one or more cigarettes a day during pregnancy. Drinking any alcohol. Women were coded yes if they reported drinking any alcohol during the first three months of pregnancy. 5 or more units of alcohol once a month or more. Women were coded yes if they reported drinking five or more units of alcohol 1-3 times a month or more during the first three months of pregnancy. Caffeine consumption. To reflect NHS advice that consuming small quantities of caffeine is acceptable, women were coded yes if they reported drinking more than one caffeinated drink per day, during the previous four weeks. Vitamin D supplementation. Women were coded yes if they reported consuming vitamin D, Pregnacare or Sanatogen prenatal , which both contain vitamin D, during the previous four weeks. All five behavioural outcomes were collected as part of the BIB baseline questionnaire at 28 weeks gestation. --- Clinical outcome Given the potential biases in self-reporting, birth weight was also used as an objective measure to act as a check on the reliability of self-reported behaviours. Research suggests that the risk of low birth weight is increased by smoking [21], alcohol [22], and possibly caffeine [23]. There is also some evidence that vitamin D is preventative of low birth weight [24,25]. Birth weight was a continuous variable and taken from the maternity IT system. Missing data was negligible for all outcome variables . --- Covariates Estimates were adjusted for age in years, ethnicity , mother's education , and date of questionnaire. The date of questionnaire was included because BIB P2s would occur, on average, later in the recruitment period than BIB P1s. A difference between BIB P1s and BIB P2s could therefore reflect changes in health behaviour in the population over time, rather than the effect of being part of BIB. All covariates came from the BIB baseline questionnaire except mother's age, which was obtained from the maternity record. These covariates were chosen a priori, on the basis that they would be associated with either health behaviour outcomes, or with the likelihood of a mother having two pregnancies within the BIB recruitment period. --- Statistical methods Given the possibility of bias in the absence of randomisation, we used propensity score matching, which is increasingly recognised as a robust method for assessing exposure effects in a non-experimental study design [26]. The propensity score is the conditional probability of having been exposed, given a set of observed covariates [27]. In the current study, the propensity score was calculated based on a multiple logistic regression model predicting previous exposure to BIB. This way, each mother's propensity score represented the probability that she was a BIB P2 based on her age, ethnicity, education, and the date she filled out her questionnaire. Nearest neighbour matching with common support was then used to create matched pairs. This involved firstly matching each BIB P1 in our control group with a BIB P2 with the closest propensity score. If there was more than one BIB P1 with an identical propensity score, one was chosen at random. Secondly, common support was applied, whereby BIB P2s whose propensity score was higher or lower than the range in propensity scores of the BIB P1s, and BIB P1s with a propensity score higher or lower than all BIB P2s were all excluded, ensuring that both groups had propensity scores within the same range. All the other BIB P1s were then dropped from the analysis to leave only the matched pairs. A number of different matching techniques were tested, and nearest neighbour matching with common support was chosen because it created the closest matches. This process involved comparing the distribution of the propensity scores within the exposed and control groups, and also comparing the mean of each covariate in both groups. Ideally, the percentage difference in the means of the two groups should be less than 5%. In this case, none of the matching techniques were quite able to achieve this . The matching approach was chosen before doing the final logistic regression, i.e. blind to the effect of the matching method on the outcome. Finally, a simple logistic regression was carried out within the matched cases only between the six behavioural outcome variables and exposure to BIB to obtain the relevant odds ratios testing our hypotheses. A simple linear regression was carried out for birth weight. Testing the differential effect of mere measurement by socioeconomic status Mother's education was chosen as the best available indicator of socioeconomic status. Education was dichotomised into those educated up to GCSE level, and those with higher levels of education. Separate logistic regressions were then carried out stratified by mother's education, in order to compare results. Logistic regression was carried out using pairs matched on propensity score. Only outcomes showing the largest effects in the previous propensity score analysis were testedany alcohol consumption and birth weight. Significance levels for all analyses were two-sided and set at 5%. Stata 12 was used, with additional user-written programmes psmatch2 [28] and pscore [29] for propensity score matching. --- Results --- Summary statistics Table 1 shows summary statistics of our control and exposed groups. BIB P1s, who had not yet participated in the study, were very slightly older and filled out their questionnaire earlier than BIB P2s. --- Propensity score matching Logistic regressions and balancing for propensity score matching are shown in appendix 1. Nearest neighbour matching produced a sample size of 156 matched pairs . Table 2 shows the results of logistic regression between BIB P1s and BIB P2s, matched on propensity score. When restricting to women on their second pregnancy, the only significant finding was for any alcohol consumption where the odds of drinking was 65% less for women who had been exposed to BIB compared to those who had only just joined the cohort . The effect estimates of smoking and drinking five or more units of alcohol once a month or more were both in the direction of an improvement in health behaviours, and the effect estimate for birth weight also showed an increase. Effect estimates for vitamin D consumption and caffeine consumption were both marginally in the direction of a deterioration in health behaviours, though the effect sizes were very small . --- Testing the differential effect of mere measurement by socioeconomic status The previous section addressed the first research question: Has being part of BIB influenced participants' health behaviour? This provided the basis for looking at the second question: Is the effect of participation in BIB on health moderated by socioeconomic status? As many outcomes showed no significant change with exposure to BIB, just two outcomes were taken forward to test for mediation by socioeconomic status: any alcohol consumption and birth weight, as an objectively measured variable. Table 3 presents the result of two bivariate regressions each stratified by education, using pairs matched on propensity score. Although both outcomes showed a larger effect estimate for those with higher education than for those with lower education, neither was statistically significant. --- Discussion This study did not provide conclusive evidence that a mere measurement effect consistently occurred in the BIB cohort study, or that it was moderated by socioeconomic status. Although some significant findings did emerge suggesting that further exploration of this topic is merited, the majority of tests showed no statistically significant effect. --- Has being part of Born in Bradford improved participants' health behaviour? Exposure to BIB was associated with a statistically significant improvement in health behaviours in only one of six regressions carried outa significant reduction in the number of women reporting any alcohol consumption. 2. If so, has this effect been moderated by socioeconomic position? When cmstratifying by education, no significant effects were found either for any alcohol consumption, or for birth weight. However, effect estimates were larger for women of higher education, compared to women of lower education. --- Limitations --- Non-randomisation As a non-randomised natural experiment, this study design is vulnerable to confounding based on noncomparability of the control and exposed groups. Another approach to seeing if taking part in BIB affected health behaviour is a before and after study, comparing first and second BIB pregnancies for the same women. Because the same women are in the control and exposed groups, many potential cofounders are avoided. Such a before and after design would not be a good way to demonstrate whether BIB had made a difference, as there would be no way to distinguish behavioural changes that would occur anyway with increased numbers of children from changes due to participation in BIB. Nevertheless, if BIB had improved behaviour it should show up with this design. This analysis was therefore carried out as a sensitivity test and is shown in Appendix 2. No significant difference was found in any health behaviour between women's first and second pregnancy in BIB, although the effect sizes for both measures of alcohol use indicate a non-significant reduction in alcohol consumption. A statistically significant difference in birth weight was identified, however. On their second BIB pregnancy, women had babies that were an average of 137.29 g larger than their first child within the BIB cohort, and this difference was statistically significant . This finding does not rule out the possibility that there could be some modest improvement as a result of exposure to BIB, which does not reach significance in individual behaviours, but does amount to an overall improvement in mother's health and therefore birth weight. However, the fact that these effect sizes are not large or significant, whereas our primary study design did show significance on one of the behavioural outcomes does suggest that those women who had two pregnancies within the cohort period may differ in unknown variables which are not possible to account for in propensity score matching. --- The trade-off between robustness and power Rather than adjusting for differences between the two groups, propensity score matching comes closer to imitating a randomised controlled trial by creating two groups that are as closely matched as possible, thereby only comparing observations that are similar enough to be comparable, based on known covariates [26,30,31]. This approach may also limit residual confounding; by only including in the analysis controls that have a similar propensity score, bias caused by any unknown variable that is a confounder and associated with one of the known covariates should be reduced. However, the use of propensity score matching, as well as the restriction of the sample to second pregnancies involved a substantial loss of sample size and power. There are some clues that a larger sample size may have been needed. Although only the regression testing for any alcohol consumption produced a significant finding, the estimated effect of exposure for BIB on smoking, consuming five or more units of alcohol once a month or more, and birth weight showed an improvement. The effect estimates for Vitamin D consumption and caffeine were in the direction of a deterioration, though both were close to zero. When testing for moderation by socioeconomic status, significance was lost on any alcohol consumption for more and less educated women, but the effect estimates for both any alcohol and birth weight indicated a larger improvement in health behaviours for more educated, compared to less educated women. Sample size is particularly important for studies of mere measurement where effect sizes are likely to be small [4], and for exploring moderation by socioeconomic status, where a still larger sample size may be required to assess more subtle distributional effects. --- Outcome measures and fertility Both alcohol consumption [32] and smoking [33] are known to reduce fertility. This creates a potential bias if it results in women who drink and smoke with their first child being less likely to have another child within the four year BIB recruitment period. If this were the case, BIB P2s would have improved health behaviour not because of the effect of BIB, but because of selection bias between the exposed and unexposed groups, leading to a possible type 1 error. However, as identified above, selection bias cannot fully account for those significant findings which were identified as the differences at baseline were smaller than those found after the BIB P2s had participated. --- Recruitment of lower socioeconomic groups Any studies in which participation is conditional on agreeing to take part in a survey are open to participation bias. In particular, studies may struggle to recruit participants of lower socioeconomic status. Analysis comparing the BIB cohort with all other births at the maternity department in Bradford Teaching Hospital found that the cohort did have marginally lower representation from mothers living in more deprived areas [34] [Table 1]. In the case that there was a difference in take-up between those of lower and higher education, the exclusion of those with very low socio-economic status could lead to a type 2 error. --- Further research As we suggest above, there are theoretical reasons to expect that a mere measurement effect may be moderated by socioeconomic status. If this were the case, it would need to be taken into account in any long-term studies of health, and would also pose important ethical challenges for research studies, particularly those such as BIB that aim to decrease health inequalities. Although not conclusive, our findings suggest that this issue merits further study. If such a question were to be incorporated into the design of a new research project, it could be designed as a nested RCT within an existing study, as described above in McCambridge and Kypri [5]. Such a design could overcome the challenges of selection bias found in this natural experiment and could easily incorporate an analysis of moderation with socioeconomic status. Ideally, such an analysis could incorporate more sensitive measures of socioeconomic status, including for example income, and explore how these different measures affected outcomes [35]. Crucially, the study should be powered not only to detect the small effect sizes expected from mere measurement, but also to detect differences between socioeconomic groups. --- Conclusion Although there are many studies showing small effects from mere measurement, research has so far failed to explore whether these effects are moderated by socioeconomic status. Overall, this study did not find sufficient evidence to conclude that mere measurement effect had occurred in the BIB cohort study, or that it had been moderated by socioeconomic status. However, one out of six of our analyses showed significant effect, and effect estimates suggested that participation in BIB may be associated with larger positive health effects for women with higher education. Research using designs with more comparable control groups and a larger sample size and are needed in order to explore potential moderation with socioeconomic status. --- Endnotes 1 The baseline questionnaire can be found at http:// www.borninbradford.nhs.uk/uploads/documents/questionaires/Born_in_Bradford_Mothers_Baseline_Questionnaire.pdf. 2 Strictly the difference between BIB P1s and BIB P2s is a difference if gravidity rather than parity as BIB P2s have not yet had their second live birth within BIB. For simplicity however, the term parity is used throughout. 3 Current NHS advice can be found at: http://www.nhs. uk/conditions/pregnancy-and-baby/pages/pregnancy-and -baby-care.aspx#close. --- Appendix 1 Balancing for propensity score matching Table 4 presents the results of the logistic regression carried out in order to obtain the propensity scores. The outcome variable was exposure to BIB, and the odds ratios indicate the odds of being a BIB P2 according to each covariate. There were 225 BIB P2s on their second pregnancy. Only 156 of these BIB P2s were found appropriate matches due to only matching BIB P1s that were within the range of BIB P2s. Table 5 shows that after matching, the percentage bias was within 5% for all covariates except education which maintained a difference of -12.1%. Fig. 2 shows that the distribution of the propensity scores between BIB P1s and BIB P2s was adequately similar --- --- Appendix 2 Before and after sensitivity test Table 6 presents regression analyses between women who participated in BIB twice in a before and after test. Their first pregnancy in the cohort acted as a control while their second pregnancy acted as the intervention. Note that only women who had two pregnancies in the BIB cohort study were included in the control group. Abbreviations BIB: Born in Bradford cohort study; NHS: National health service; OR: Odds ratio; SES: Socio-economic status; SIDS: Sudden infant death syndrome Authors' contributions AQ participated in designing the study, carried out all analyses and drafted the manuscript. JRB provided input on statistical methods, in particular identifying the method of propensity score analysis, and helped draft the manuscript. JW helped to conceive of the study, participated in designing the study and commented on drafts of the manuscript. KEP helped to conceive of the study, participated in designing the study, identifying the appropriate analyses and helped draft the manuscript. All authors read and approved the final manuscript. --- Competing interests The authors declare that they have no competing interests. --- ---

Background: Evidence suggests that the process of taking part in health research can improve participants' health, independent of any intended intervention. However, no research has yet explored whether these effects differ across socioeconomic groups. If the effect of mere participation in health research also has a social gradient this could increase health inequalities and bias research results. This study used the Born in Bradford family cohort (BIB) to explore whether simply taking part in BIB had improved participants' health and, if so, whether this effect was mediated by socioeconomic status. Methods: Survey data on self-reported health behaviours were collected between 2007 and 2010 as part of BIB. These were augmented by clinical data on birth weight. Pregnant women on their second pregnancy, joining BIB for the first time formed the control group. Their health was compared to women on their second pregnancy who had both pregnancies within the study, who formed the exposed group. In order to limit the inherent bias in a non-randomised study, propensity score analysis was used, matching on age, ethnicity, education and date of questionnaire. The results were then compared according to mothers' education. Results: Of six outcomes tested, only alcohol consumption showed a statistically significant reduction with exposure to BIB (OR: 0.35, 95% CIs 0.13, 0.92). Although effect estimates were larger for women with higher education compared to lower education, these effects were not statistically significant. Conclusions: Despite one significant finding, these results overall are insufficient to conclude that simply taking part in BIB affected participants' health. We recommend that socioeconomic status is considered in future studies testing effects of research participation, and that randomised studies with larger sample sizes are conducted.

and survival advantage granted by group life throughout our prehistory . A corollary of this assumption is that experiencing a subjective sense of identification with groups is a precondition for good mental health . Consistent with this assumption, Sani et al. found an association between a greater number of group identifications and lower odds of being depressed in a large Scottish community sample. Furthermore, a longitudinal study of Scottish adolescents revealed that greater number of group identifications predicted better mental health over time, but also that better mental health predicted greater number of group identifications over time, suggesting reciprocal causality. However, little is known about factors that may thwart group identification. We contend that child maltreatment constitutes an experience that may seriously impair group identification across the life-span. This is because child maltreatment reduces social competence and erodes the ability to form effective relationships . The current study tested this proposition in a cross-sectional investigation of the extent to which the severity of maltreatment received during the first 14 years of age is associated with number of group identifications in young adults. It must be noted, however, that severity of child maltreatment predicts mental health problems . Consequently, since greater mental health problems predict lower number of group identifications, one could argue that any found link between severity of child maltreatment and a lower number of group identifications might be explained by the fact that greater child maltreatment Running head: CHILD MALTREATMENT AND GROUP IDENTIFICATIONS 4 leads to poorer mental health, which in turn decreases group identifications. Therefore, in the current study we investigate the effects of child maltreatment on group identifications while controlling statistically for the link between child maltreatment and mental health. --- Method --- Participants and Procedure A sample of 396 undergraduate students from a Spanish university completed a questionnaire voluntarily at the end of a class. --- Measures We assessed number of subjective group identifications using the Sense of Belonging to Groups checklist, a new instrument created for use in this study. Participants could select any number of groups from 0 to 11 from a list including, for instance, family, workplace group, and group of friends, with which they identified. To assess child maltreatment we used the scale devised by Briere and Runtz , which includes 14 items concerning psychological maltreatment and 10 items referring physical maltreatment . Because preliminary analyses revealed very similar correlational patterns for the two scales, we created an overall score for maltreatment based on 24 items. We assessed three mental health variables to be used as controls in our multivariate statistical analyses. We measured depressive symptoms, using the Major Depression Inventory , and anxiety, using the Generalised Anxiety Disorder scale , because they tend to be the most common mental health problems in the general population . We also measured borderline personality, using the McLean Screening Instrument -Borderline Personality Disorder , because it is one of the most likely mental health outcomes among people who have been maltreated as children . --- Results --- Descriptives, reliabilities, and inter-correlations Means, standard deviations, and reliabilities for all the variables as well as the inter-correlations among the variables, are reported in Table 1. Point-biserial correlations were used to calculate the relationship between gender and other variables. . --- TABLE 1 Measurement instruments had good reliabilities, with Cronbach alphas ranging from .73 to .84. Number of group identifications had a statistically significant correlation with child maltreatment , as well as with the three mental health variables, with r-values ranging from -.21 to -.27. Child maltreatment also had a statistically significant correlation with the three mental health variables, with r-values ranging from .24 to .33. The intercorrelations among the three mental health variables were all statistically significant. Gender and age were not significantly correlated with any other variable. --- Regression analysis We first ensured that data met the assumptions required for regression analysis. The tolerance statistics was > .20 and the variance inflation factors to be < 10, indicating absence of multicollinearity. The Durbin-Watson statistic was between 1 and 3, indicating independence of error. We also found the distribution of standard residuals to be normal, linear and homoscedastic, and that less than 5% of cases Running head: CHILD MALTREATMENT AND GROUP IDENTIFICATIONS 6 exceeded two standard deviations from the mean, confirming that outliers were not a cause for concern. At this point we proceeded with the analysis. In the regression we entered child maltreatment together with the three mental health variables. Age and gender were not entered as control variables because the correlation analyses revealed no links between these variables and the other variables. Number of group identifications was used as the outcome variable. Table 2 reports the results. --- TABLE 2 The overall model had a good fit with the data and explained 14% of the variability in number of group identifications. We found greater degrees of child maltreatment to be linked with lower number of group identifications: β = -.27, p < .01. Greater levels of depressive symptoms were also linked to lower number of group identifications: β = -.16, p < .05. Anxiety and borderline personality exerted a small negative, and statistically not significant influence, on number of group identifications . --- Discussion These results demonstrate that child maltreatment is associated with a thwarted ability to develop a sense of belonging to social groups. Although these findings might be partly due to the fact that a current sense of detachment from social groups will lead one to recall their childhood more negatively, they suggest that, at least in part, child maltreatment is responsible for one's difficulties to identity with social groups as a young adult. Therefore, these findings have implications for forms of intervention aimed at helping mental health patients to reconnect with social groups, such as the one advocated by the Groups4Health protocol . While helping mentally distressed people to reconnect psychologically to social groups may generally contribute to reduce distress and facilitate recovery, we must keep in mind that some people may find it especially difficult to attach to groups Running head: CHILD MALTREATMENT AND GROUP IDENTIFICATIONS 7 because they have been maltreated as children. Helping these people to reconnect to groups may need especially designed forms of social intervention. Finally, we should highlight two major limitations of this study. First, the study relies upon recollections, which may inevitably be distorted. However, the average age of our participants was just over 19 years. Therefore, they reported experiences that had taken place not a long time before. This is likely to reduce the margins of recollection error. Second, our study used university students from a specific country. Cross-national studies involving community samples should therefore be conducted in order to obtain further confirmation of the effects found in this study.

This is the peer reviewed version of the following article: "Child maltreatment is linked to difficulties in identifying with social groups as a young adult", British Journal of Development Psychology (2020), which has been published in final form at https://doi.org/10.1111/bjdp.12332. This article may be used for non-commercial purposes in accordance with Wiley Terms and Conditions for Self-Archiving.

INTRODUCTION The transition to parenthood can affect well-being in a range of domains. Most research on distress in perinatal populations has focused on psychological disorders, such as depression , anxiety disorders , obsessive-compulsive disorder , and post-traumatic stress disorder . Another important element of parental well-being is the quality of new parents' relationships with their partners and babies. In this study, we apply an attachment theory perspective to investigate whether the experience of parental divorce or separation is associated with romantic relationship satisfaction and parent-infant bonding among first-time parents during the early postpartum period. --- Relationships During the Transition to Parenthood Individuals and couples navigate many potential challenges when they become parents, including sleep disruption, introduction of novel responsibilities, re-negotiation of the division of household labor, and changes in the amount and quality of time that couples spend together . Many couples experience a small but significant decline in relationship satisfaction during the transition to parenthood . This can have important consequences; for example, postpartum relationship dissatisfaction is associated with increased risk for relationship dissolution during the first 3 years of parenthood . Relationship dissatisfaction during the perinatal period is also associated with increased risk for depression and anxiety for both men and women . New parents may also experience difficulties in their relationships with their babies. Parent-infant bonding refers to parent's affective responses and cognitive evaluations of their relationship with their infant . It is important to distinguish parent-infant bonding, which represents the parent's perception of this relationship, from infant attachment, which represents the quality of the infant's relationships with their caregivers. In community samples, the prevalence of clinically significant disturbances to the parent-infant bond approaches 10% ; among mothers referred for perinatal services, the prevalence of severe disruption to the mother-infant bond exceeds 25% . Impairments in parent-infant bonding are associated with less sensitive and appropriate parenting , later disruptions to the parent-child relationship , and increased risk for child cognitive and behavioral difficulties . Most research on parentinfant bonding has been conducted with mothers, but there is also clear evidence for the importance of the father-infant bond . Although some new parents develop problems in their relationships with their partners or infants, this experience is far from universal. There is a wide variability in the nature and degree of changes in romantic relationship satisfaction across the transition to parenthood, with approximately half of couples reporting stable or even improved relationship quality . Similarly, most new parents do not experience clinically significant impairments to the parent-infant bond, even in the context of stressors, such as maternal depression . Given the variability in new parents' experiences of relational difficulties during the transition to parenthood, it is important to identify processes that contribute to both risk and resilience as parents, couples, and families navigate the challenges associated with the early months of parenthood. --- Attachment Theory Attachment theory provides a valuable lens through which to understand new parents' relationships with their partners and infants . According to this theory, patterns of behavior and expectations that develop through early interactions with caregivers are often imposed on new relationships later in life . Early experiences with caregivers lead children to form representations, or inner working models, of the self and others, which can be used to characterize an individual's attachment style . Securely attached individuals are confident that others will be available and responsive to their needs, while insecurely attached individuals may be uncertain of others' availability or responsiveness or may not expect to receive support from others . These foundational concepts in attachment theory were later extended to adult relationships, including romantic relationships . Adult attachment is often characterized along two dimensions: attachment anxiety, which is characterized by uncertainty that one's feelings are reciprocated and concerned that a partner will leave, and attachment avoidance, which is characterized by difficulty with trust and intimacy . --- Attachment Security and Relationship Outcomes There is clear evidence that attachment style is associated with the quality of important relationships in adulthood. Meta-analyses find that, among adults, both attachment anxiety and avoidance are associated with romantic relationship dissatisfaction . Attachment style may play a particularly important role in adaptation during periods of stress or change, including the transition to parenthood . Among new parents, securely attached individuals report greater romantic relationship satisfaction; their partners also evaluate these relationships more positively . Adult attachment style is also related to new parents' relationships with their infants. Securely attached mothers report more positive mother-infant bonding during the first year postpartum . In contrast, insecure maternal attachment is associated with a range of impairments to the mother-infant bond. Mothers whose attachment styles are characterized by anxiety and ambivalence report greater anxiety in their relationships with their infants, while mothers with avoidant attachment styles report greater rejection and anger . Mothers with disorganized attachment styles, characterized by high levels of both anxiety and avoidance, also report more impaired bonding . To our knowledge, no previous studies have directly assessed the relationship between paternal attachment and the father-infant bond. --- Effects of Parental Divorce/Separation on Attachment Security and Relationship Outcomes The experience of parental divorce or separation in an individual's family of origin may affect their interactions with caregivers in ways that increase risk for insecure attachment. For example, children of divorced parents may experience reduced contact with one or both caregivers due to custody arrangements, and the quality of interactions with caregivers may be affected by increased stress or new responsibilities . There is evidence that parental divorce is associated with insecure attachment among children , adolescents , and adults . Furthermore, among young children classified as securely attached, the experience of a subsequent parental divorce is associated with greater risk for insecure attachment in adulthood . The association between parental divorce/separation and adult attachment insecurity has also been demonstrated during the transition to parenthood; expectant parents whose own parents divorced or separated are more likely to be classified as insecurely attached than those from intact families of origin . This increased risk for attachment insecurity suggests that the experience of parental divorce or separation may affect relationships throughout the lifespan. There is evidence that adults from intact families of origin experience more positive relationship outcomes in adulthood. An early meta-analysis found a small but significant increase in the prevalence of divorce/separation among adults whose own parents had divorced or separated . Adults whose parents divorced or separated also report greater dissatisfaction in their romantic relationships . Notably, no studies have directly assessed whether new parents' experiences of divorce or separation in their own families of origin are associated with parent-infant bonding. Although a history of parental divorce or separation appears to increase risk for negative relational outcomes, many adult children of divorced or separated parents do not experience these difficulties. In fact, some report relational benefits resulting from parental divorce, including closer relationships with siblings and other family members and expansion of social networks . There is also evidence that the association between parental divorce and negative outcomes may be decreasing as divorce has become more common and less stigmatized . These findings suggest a need for further research on the association between parental divorce and relational outcomes in adulthood, especially during periods of increased stress, such as the transition to parenthood. --- THE CURRENT STUDY In this study, we investigated whether new parents with a history of parental divorce or separation experience greater romantic relationship dissatisfaction or impairments in the parent-infant bond during the early postpartum period, and whether these associations are mediated by adult attachment style. Consistent with previous research, we predicted that first-time parents whose parents had divorced or separated would report lower romantic relationship satisfaction and greater impairments in the parent-infant bond. We further predicted that these parents would report higher levels of attachment anxiety and avoidance. Finally, we hypothesized that associations between parental divorce/separation and relationship outcomes would be mediated by adult attachment. --- METHOD Individuals were eligible to participate if they were a first-time parent of a child aged 6 months or younger, between the ages of 18 and 45, resided in the United States, and were currently in a relationship. Participants were recruited online and completed all study materials at a single time-point through a secure online interface. After providing informed consent and confirming eligibility, participants completed measures of adult attachment style, romantic relationship satisfaction, and parent-infant bonding in a random order. Participants then provided information about their family of origin, including a retrospective report of their parents' relationship status and demographic information. The study received approval from the Institutional Review Board of Davidson College prior to data collection. --- Measures --- Adult Attachment Style Attachment was assessed with the Experiences in Close Relationships Scale , a 36-item measure characterizing general feelings in romantic relationships that includes subscales for anxious and avoidant elements of attachment. Scores on each subscale range from 18 to 126, with higher scores indicating greater attachment insecurity. Internal reliability was excellent for both anxiety and avoidance . --- Romantic Relationship Satisfaction Relationship satisfaction was assessed with the Dyadic Adjustment Scale , a 32-item self-report measure that provides an overall indicator of romantic relationship quality. Scores on the DAS range from 0 to 151, with higher scores indicating greater relationship satisfaction. Internal reliability of the DAS was excellent . --- Parent-Infant Bonding Parent-infant bonding was assessed with the Parental Bonding Questionnaire . The original PBQ includes 25 items assessing four domains of the parentinfant relationship: impaired bonding, rejection/anger, anxiety, and risk of abuse . As previous studies have found that the risk of abuse subscale has low sensitivity and reliability , these items were excluded from the current study. Scores on our revised 23-item version of the PBQ range from 0 to 115, with higher scores indicating greater impairment of the parent-infant bond. Internal reliability for the PBQ was excellent . --- Family of Origin and Demographic Characteristics Participants were first asked if both parents were living; participants who did not report loss of a parent were asked to indicate the current status of their parents' relationship to one another. Participants who reported that their parents were divorced or separated also provided their age at the time of the divorce/separation. Following the completion of study materials, participants provided the following demographic information: age, gender, relationship status, race, ethnicity, level of education, employment status, child age, and child gender. --- --- RESULTS --- Descriptive Statistics Descriptive statistics for the primary study measures are presented in Table 2. As scores on the DAS and PBQ were skewed, these variables were winsorized prior to analyses; outliers ≥3.29 standard deviations from the mean were replaced with the value ±3.29 standard deviations from the mean . We first assessed whether demographic characteristics were associated with study outcomes. After correcting for multiple comparisons, we found no significant associations between any demographic characteristic and study outcomes and did not include these variables in subsequent analyses. The pattern of correlations among the study measures was consistent with our hypotheses . Attachment anxiety and avoidance were moderately correlated with one another. There was a marginally significant association between romantic relationship satisfaction and parent-infant bonding. Attachment anxiety and avoidance were both significantly associated with romantic relationship dissatisfaction and impairments to the parent-infant bond. Most participants' parents were married or in a committed relationship with one another ; nearly one-third of the sample reported that their parents were divorced or separated . Among participants whose parents were divorced or separated, the age at parental divorce/separation ranged from 1 to 31 years, with a mean of 12.3 years . --- Associations Between Parental Divorce/ Separation and Relationship Outcomes We first compared the two groups of participants on adult attachment, romantic relationship satisfaction, and parent-infant bonding using a series of independent-samples t-tests . There were no significant differences in any of these outcomes between participants whose parents were divorced or separated compared to participants from intact families of origin . Although we did not find differences in romantic relationship satisfaction or parent-infant bonding between the two groups of participants, it is possible to observe significant indirect effects even when the corresponding direct effect is not significant . To assess this, we conducted a series of mediation analyses investigating whether the relationships between parental divorce status and relationship outcomes were mediated by adult attachment. Attachment anxiety and avoidance were assessed as potential mediators of the relationships between parental divorce status and romantic relationship satisfaction and parent-infant bonding in separate models. Mediation analyses were conducted using version 3.4 of Hayes PROCESS macro for SPSS, using 10,000 bootstrap samples. Results of the mediation analyses indicated that parental divorce status was not significantly associated with either attachment anxiety or avoidance . Attachment anxiety and avoidance were both significantly associated with lower romantic relationship satisfaction . The 95% bias-corrected bootstrap CIs for the indirect effects of attachment on romantic relationship satisfaction ranged from -3.83 to 1.46 for anxiety and -7.57 to 1.38 for avoidance, indicating that there was not a significant indirect effect of parental divorce status on romantic relationship satisfaction via either attachment anxiety or avoidance. There was also not a significant direct effect of parental divorce status on romantic relationship satisfaction after controlling for attachment anxiety or avoidance. Results of mediation analyses revealed a comparable pattern for parent-infant bonding . Both attachment anxiety and avoidance were significantly associated with greater impairments in parent-infant bonding. The 95% bias-corrected bootstrap CIs for the indirect effects of attachment ranged from -0.97 to 2.32 for anxiety and -0.47 to 2.96 for avoidance, indicating that there was not a significant indirect effect of parental divorce status on parent-infant bonding via attachment anxiety or avoidance. There was also not a significant direct effect of parental divorce status on parent-infant bonding after controlling for attachment anxiety or avoidance. --- DISCUSSION Findings from this study did not support our hypotheses that a history of parental divorce or separation would be associated with insecure attachment, romantic relationship dissatisfaction, and impaired parent-infant bonding during the transition to parenthood. We did, however, find that adult attachment security is associated with both romantic relationship satisfaction and parent-infant bonding during the early postpartum period. These findings provide a valuable counterpoint to prevailing trends in the study of the consequences of parental divorce and new parents' romantic relationships, both of which typically emphasize potential negative outcomes . In contrast, our findings suggest that many individuals who experience parental divorce or separation in their families of origin do not experience greater attachment insecurity or more problems in their relationships with their partners or children. To our knowledge, this is the first study to directly assess whether divorce or separation in a new parent's family of origin is associated with impairments in the parentinfant bond. We found no evidence that these parents experience more difficulties in their early relationships with their babies. These results are initially surprising, given that most studies find that parental divorce is associated with a greater risk for negative relational outcomes in adulthood . One possible explanation is that parental divorce or separation may predict whether an individual pursues or achieves specific relational outcomes, such as choosing to marry or start a family, but may be less predictive of the quality of later relationships. As our sample was limited to new parents who were currently in a relationship, our findings may not be representative of single adults or non-parents. Consistent with this explanation, a similar pattern of results was observed in a longitudinal study of newlywed couples. Among these adults, who had established a successful relationship and made the decision to marry their partner, those whose own parents had divorced or separated had no greater risk of divorce during the first 6 years of marriage than newlyweds from intact families . Our findings provide clear evidence that insecure attachment is associated with relationship difficulties during the transition to parenthood. New parents with high levels of anxious and avoidant attachment were less satisfied in their romantic relationships; this finding is consistent with a large body of literature , as well as with theoretical models that emphasize the importance of attachment style during periods of transition . We also found that new parents who reported high levels of anxious and avoidant attachment experienced greater impairments in the parent-infant bond. Importantly, we demonstrated that this association is present in a sample including both mothers and fathers. Previous studies have found that securely attached mothers report more positive mother-infant bonding , and that insecurely attached mothers are at greater risk for impairments to the mother-infant bond . Our findings extend this literature and provide evidence of increased risk for impaired father-infant bonding in the context of insecure paternal attachment. Together, these findings suggest that individuals and couples may benefit from interventions addressing attachment security during the transition to parenthood. Research has identified characteristics of relationships that appear to be protective against declining relationship satisfaction among new parents, including the quality of interactions, engagement, and congruence of attitudes related to gender roles and the division of household labor . Secure attachment might enhance these factors through skills that promote adaptive processes in couple interactions, including the ability to give and receive emotional support, conflict management skills, and affective regulation abilities . --- Limitations and Future Directions A significant limitation of the current study is our relatively small sample. Our analyses, particularly those assessing mediation, were underpowered; when associations between individual variables are small, mediation analyses may require samples of 400 or more participants . Although underpowered, the effect sizes we observed suggest small or negligible differences between new parents from intact families compared to those whose parents divorced or separated. In fact, participants from intact families reported slightly more Path a represents the effect of parental divorce status on attachment, path b represents the direct effect of attachment on the relationship outcome , path c represents the total effect of parental divorce status on the relationship outcome, and path c' represents the direct effect of parental divorce status on the relationship outcome . All coefficients are presented as unstandardized estimates. To facilitate interpretability of the unstandardized coefficients, participants' average scores on the ECR subscales were used for the adult attachment variables. impaired bonding than those whose parents had divorced or separated. Despite our small sample, this pattern of findings suggests that the absence of a relationship between parental divorce/separation and relational outcomes in this study is unlikely to be accounted for by inadequate statistical power. A related limitation is that our sample included a small number of participants who had experienced parental divorce or separation; thus, we were unable to evaluate characteristics of the divorce/separation as predictors of attachment and relational outcomes. One important characteristic that should be addressed in future research is the age at which the parental divorce/ separation occurred. In our sample, participants' age at the time of divorce/separation ranged from 1 to 31 years. In an early study of the "intergenerational transmission of divorce, " Amato found that the relationship between parental divorce and marital status in adulthood was strongest when the divorce occurred when the child was 12 years old or younger. Similarly, parental divorce/separation appears to be more strongly associated with insecure attachment when it occurs earlier in childhood . We conducted exploratory analyses including only participants who were younger than 13 at the time of their parents' divorce/separation ; although our findings remained stable, the very small sample of participants who experienced a parental divorce or separation prior to adolescence remains a concern. We were also unable to assess other important characteristics of the divorce/separation, such as the presence of parental conflict or changes in the amount or quality of contact with caregivers that are associated with long-term outcomes . Future research with larger samples would allow for more robust evaluation of specific characteristics of parental divorce/separation that might influence attachment security and later relationships. In addition to evaluating larger samples, future studies investigating associations among parental divorce/separation, attachment, and relational outcomes should include more diverse and representative samples. Our sample was primarily comprised of white, non-Hispanic/Latinx women. Although we found no differences in adult attachment, romantic relationship satisfaction, or parent-infant bonding related to participants' demographic characteristics, our ability to detect potential differences was limited by inadequate representation of participants from specific demographic groups. While exploratory analyses found no evidence that gender moderated the relationships among parental divorce/separation, attachment, and relationship outcomes in our sample, other studies suggest that parental divorce/separation may be a stronger risk factor for negative relational outcomes for women . Similarly, there is evidence that associations between early parent-child relationships and later adult attachment may vary across racial and ethnic groups . Greater representation of participants from a wide range of demographic groups would allow for direct evaluation of whether associations between parental divorce/separation and relational outcomes vary according to parental characteristics, such as gender, sexual orientation, race/ethnicity, or socioeconomic status. Finally, the present study was limited by its cross-sectional design and retrospective assessment of parental relationship status. Future research using longitudinal designs would allow for the prospective evaluation of the effects of parental divorce or separation on relational outcomes. In addition to providing an opportunity to assess the effects of parental divorce or separation on romantic relationship satisfaction and the parentinfant bond, such studies would also provide opportunities to evaluate whether adults whose parents are divorced or separated are less likely to pursue specific kinds of relationships in adulthood, or to become parents. --- CONCLUSION Overall, these findings provide evidence that adult attachment is strongly associated with the quality of new parents' relationships with their partners and babies during the early postpartum period. In contrast to previous research, we did not find that attachment security or relational outcomes were associated with the experience of divorce or separation in new parents' families of origin. These findings highlight the importance of adult attachment and suggest that secure attachment may promote positive relational outcomes for parents, couples, and families during the transition to parenthood. Will Thurston for assistance with study development and participant recruitment. We also thank Dr. Jessica J. Good and Dr. Lauren A. Stutts for their valuable feedback on study design, analytic approach, and a preliminary version of this manuscript. An earlier version of this manuscript was submitted by KL as an undergraduate honors thesis to the Department of Psychology of Davidson College . Preliminary results of this study were presented at the 2018 annual meeting of the Association for Behavioral & Cognitive Therapies in Washington, DC. --- DATA AVAILABILITY STATEMENT The datasets generated for this study are available on request to the corresponding author. --- ETHICS STATEMENT The studies involving human participants were reviewed and approved by Davidson College Institutional Review Board. The patients/participants provided their written informed consent to participate in this study. --- --- Conflict of Interest: 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.

An important element of well-being during the transition to parenthood is new parents' relationships with their partners and babies. Attachment theory posits that early caregiving experiences influence close relationships throughout the lifespan. Disruptions to the parent-child relationship, such as parental divorce or separation, may therefore have intergenerational effects as adult children of divorce navigate changes in their later relationships. This study examined whether new parents who have experienced a divorce or separation in their family of origin report greater romantic relationship dissatisfaction or impairment in the parent-infant bond during the early postpartum period, and if these associations are mediated by adult attachment. First-time parents of infants through 6 months of age (N = 94) completed measures of adult attachment, romantic relationship satisfaction, and parent-infant bonding. New parents who had experienced parental divorce or separation did not differ from those from intact families with regard to romantic relationship satisfaction, parent-infant bonding, attachment anxiety, or attachment avoidance. Attachment anxiety and avoidance were both associated with romantic relationship dissatisfaction and greater impairment in the parent-infant bond. These findings suggest that the experience of parental divorce or separation, in and of itself, does not confer increased risk for negative relational outcomes among new parents. Securely attached adults, regardless of their own parents' marital status, report more positive relationships with their partners and infants during the early postpartum period.

Introduction There is evidence that living in a stepfamily is an increasingly common experience . Among representative samples of married couples with children, a recent state survey found that 40% of couples reported having one or more stepchildren . Similar proportions were found in the same study from two other states. Overall, indications are that stepfamilies are highly prevalent in local communities, particularly among low-income families and cohabiting families, and among African Americans. Based on both the common experience and the unique needs revealed through reviews of the research on stepfamily dynamics , there is a clear need for family life and couple education that is specific to stepfamilies. Assessments of couple/family life education implementation efforts of Extension agents and other community professionals exist ; however, we have no information regarding efforts to provide community-based education specifically for couples/parents in stepfamilies. The goal of the reported here study was to find current, consistent stepfamily program implementation efforts and explore the range of approaches, methods, formats, settings, delivery options, costs, and content used in selected programs. --- Methods We cast a "wide net" to identify face-to-face community educational programs for couples in stepfamilies by contacting curriculum developers experts in the stepfamily and relationship/marriage education fields, and representatives of "umbrella" and membership organizations likely to know of programs consistently offered for stepfamilies . We also conducted Internet searches using relevant key words and sent requests through faith-based and secular organization listservs. After considerable effort through electronic means and over 200 personal contact phone-calls for leads, we identified approximately 40 programs consistently serving stepfamilies and spoke with program delivery staff and conducted brief interviews that focused on the format, content, goals, design, and setting. After coding for implementation format and organizational setting, we established four categories of program types and selected 16 total program for in-depth interviewing. The 16 were randomly selected within category, proportional to the prevalence of the 40 programs in each category . Our conversations with these 16 program staff who were involved in direct program delivery lasted an average of 45 minutes. The interviews were exploratory, open-ended discussions of key aspects of the programs, including more detailed information related to program content, design and format, and outreach efforts. The following results are based on the findings obtained from the interviews conducted with individuals that represented the 16 stepfamily programs. --- Results We organized the information collected into themes that represent elements of service delivery, participant and facilitator characteristics, and program content. We present these findings descriptively. --- Format and Design The programs studied can be separated into one of four formats. Multi-session workshops: The majority of the programs typically offered instruction weekly in 2 to 2½-hour sessions for a 6-to 12-week period lead by a facilitator. Multi-session workshops for smaller groups, typically fewer than eight couples, allowed for a larger number and broader range of topics to be presented. Single-session workshops: Twenty-five percent of the programs consisted of workshops that met once and ranged in length from 90 minutes to 4 hours and were lead by a facilitator. The latter workshops were intended as complete programs, whereas the shorter 90-minute workshops were used to introduce stepfamily topics and skills and inform couples of ongoing support groups or other programs/services in the community. Participant numbers ranged from 10 -75. • Support groups: Twenty-five percent of the programs offered services in the form of support groups of 6-15 individuals that met weekly for 1-2 hours and were both faith-based and secular. Most groups were designed specifically by and for stepmothers. Topics were driven by the group membership and tended to focus on challenges faced. --- • Distance coaching: A few of the programs offered services via individual telephone or email sessions. The emphasis is on providing general information about stepfamilies and then helping couples deal with their particular challenges through skills practice. The telephone coaching sessions given by an individual experienced in work with stepfamilies run 60 to 90 minutes, and a couple usually has from 6 to 10 coaching sessions. --- • --- Outreach and Recruitment Universally, program staff stated that recruiting couples in stepfamilies is a challenge. While none reported formally collecting recruitment data, they offered possible reasons based on informal conversations and experiences in the community. Many couples may be unaware, both that stepfamilies have unique characteristics and that many of the issues and challenges they face can be addressed through education. --- • Couples in stepfamilies may be reluctant to self-identify as stepfamily couples due to a perceived cultural bias against them. Faith-based program leaders indicated that many couples did not attend church-sponsored programs because of the stigma they felt about having been divorced. --- • Stepfamilies often have more children, on average, than nuclear families and may have multiple demands on their time that may hinder their ability to attend a program even more so than nuclear families. --- • In order to address these challenges, the program staff highlighted their most effective outreach and recruitment approaches. The following categories/themes emerged. Reaching stepfamilies through word-of-mouth by program leaders and former participants seems to be more effective than brochures, newspaper advertising, and public service announcements. --- • Programs sponsored by churches find that the church leader can have a significant role in encouraging participation in stepfamily programs and in making stepfamilies feel welcome in the congregation. --- • Community Education Programs Serving Couples in Stepfamilies: A Qualitative Study of Format, Content, and Service 10/28/10 06:30:18 Several secular programs commented that most of their outreach is done through their own Web sites and the Web site of the National Stepfamily Resource Center and utilize listservs for announcing and recruiting for programs. --- • --- Program Participants Interviewees reported that the participants attending the programs were almost exclusively White, European-American, middle-and upper-middle-class stepcouples. One exception to this was a multi-session program exclusively serving South American immigrant stepfamilies in Spanish. All programs reported welcoming parents/couples, whether dating, cohabiting, or married. Program staff believed that the relationship status of stepfamilies would not influence the fundamental implications, content, and delivery planned. No studies exist that compare family functioning in married vs. cohabiting stepfamilies; therefore, it appears that program staff are using a theoretical assumption of similar functioning. Although there is some literature suggesting the benefits of including children in programs for stepfamilies , few of the programs we contacted included children. None of the programs delivered any educational content on stepfamilies for children, even though some curricula utilized includes lessons for children. This gap may partly reflect our sampling approach, which concentrated on finding programs for adults/couples in stepfamilies. Services for children and youth may be found more often in stand-alone programs that specifically target children/adolescents of divorce and remarriage. --- Setting The programs were held in a variety of settings, including religious institutions, libraries, schools, and space provided in public agency buildings. Some programs were offered by individuals without formal agency/organization affiliation. Among the programs with organizational affiliation, about half were sponsored by churches, one by a large public school district, and the rest by community agencies, including a county Extension office. --- Cost to Participants/Financial Support The costs of participation in the programs contacted varied widely. In a few programs, the cost was entirely covered by the sponsoring agency or through a grant. We noted that most did not report program-specific grant support. Fees in programs that did charge participants ranged from $15 for the book used in one multi-session program to $189 for a 10-session program. One coach reported charging $75 an hour for telephone coaching and $20 per month for email support. About half of programs charging fees offered financial assistance to couples who needed it. This took the form of reduced fees based on income level. --- Program Leaders/Staff The programs we reviewed tended to have two kinds of leaders. One type was the leader who created programs for stepfamilies after realizing that there were no programs available to help them with their own stepfamily challenges. Most of these program leaders started their programs within the past 10 years, and most did not initially have training in family life education, social work, or a related field. least one program required stepfamily experience of its leaders. Programs contacted were split about equally in terms of whether they were led by an individual or a team of two, typically a man and a woman. --- Common Topics Whereas we found the programs varied widely in terms of service delivery, we found much more consistency in the topics presented to stepfamilies. Key common topics found consistently across the programs included the following. Developing an understanding of stepfamilies' unique characteristics and how they differ from nuclear families. There was strong agreement among program contacts that an understanding of the degree to which stepfamilies differ from nuclear families could help to reduce the stress stepfamilies may experience. A key theme was the emphasis that it can take a number of years to adjust to and build relationships in stepfamilies. • Building an effective stepparent-stepchild relationship. Programs emphasized the impact the stepparent-stepchild relationship has on the couple relationship; encouraged stepparents to have realistic ideas of how long it usually takes to form a bond with a stepchild; and normalized the fact that stepparent-stepchild relationships are typically not as close as parent-child relationships. • Maintaining a cooperative co-parenting relationship. The main emphasis here was on the fact that appropriate, civil, and workable relationships with former partners are essential for children's wellbeing and can reduce stress on the couple relationship. It also is important to note that in some instancesâ such as when a prior relationship involved domestic violenceâ the most appropriate relationship with a former partner might be having no relationship at all. --- • Couple relationship skills. Most reported addressing couple communication and conflict management skills, empathy skills for understanding their partner's point of view, and negotiating skills for handling family finances. --- • --- Program Materials Program staff reported using a variety of resource materials. In many of the programs, leaders developed and assembled program materials based on their own experiences and from reading and attending stepfamily workshops. Lay leaders of some faith-based support groups reported sometimes having the group read a book on a stepfamily topic and then discuss it, instead of using a formal educational curriculum. Several respondents reported using educational programs available through Extension and the National Stepfamily Resource Center, and others reported using a commercial, video-based "curriculum in a box." --- Discussion and Recommendations This article examines current efforts to provide community education programs for couples in stepfamilies. After an extensive search for programs, we interviewed a sample of program staff who included community volunteers and professionals. Delivery methods included class series, concentrated workshops, and one-on-one coaching. Community Education Programs Serving Couples in Stepfamilies: A Qualitative Study of Format, Content, and Service 10/28/10 06:30:18 In our sample, there were nearly as many "lay" leaders, motivated by their own experiences and difficulty finding programs as professional educators offering programs. While there appears to be prevalent use of self-created materials rather than published research-based programs, key topics presented are consistent with recommendations for essential program content for couples in stepfamilies derived from research . In this sample of programs, there was no evidence of either outreach to or use by a more economically and ethnically diverse group of parents/couples. It appears that higher resource, European Americans are the predominant participants for stepfamily-specific educational programs. Among those interviewed, only one targeted ethnic minority stepfamilies, and none targeted low-income families. While we cannot draw conclusions about the population of stepfamilies being served based on the results of this initial qualitative study, it is recommended that educators and other professionals working among more diverse populations ensure that opportunities for stepfamily program participation are available. It was surprising to face such difficulty in finding community educators who consistently provide programs targeting stepfamilies. While the challenges faced in recruitment may serve to dissuade Extension and other community educators from offering stepfamily programs in favor of other family life and couple education programs, information regarding the prevalence of stepfamilies in our communities, particularly among ethnic minority and lower-resource families , and the unique family life and relationship education needs of these families, should be considered. It should be noted that current federal funding to support relationship education is being utilized by Extension and non-Extension community educators to enhance family and couple functioning in stepfamilies in at least one state , and efforts include an emphasis on ethnically and socioeconomically diverse families. These programs were not underway at the time of our search and interview with program staff. This study indicates that consistent efforts to provide targeted community education for stepfamilies are limited and continued efforts to explore barriers, from both a program and participant perspective, are warranted. We encourage Extension educators to pursue continuing education and training in stepfamily education programs in order to increase stepfamilies' access to programs. Extension educators can both provide these programs directly, as well as provide technical assistance to "lay" educators in their communities.

This article examines current efforts to provide community education programs for couples in stepfamilies. The study used qualitative interviews to explore the range of approaches, formats, settings, costs, and content used by Extension/non-Extension educators in programs for stepfamily couples. Results indicate that despite the prevalence of stepfamily couples in our communities and their unique needs in relationship programs, a dearth of current offerings exist. Those that do are as likely led by a volunteer as a trained professional; however, key concepts taught are similar across programs. Implications for educators are offered.

Introduction In many societies, people are concerned about the cultural fear of being associated with a genetic disease and the following possible stigmatization within the community [1]. Indeed, the frequency and the consequences of genetic diseases in Arab populations are high due to the degree of consanguinity between parents [2]. This could raise ethical, social, and legal issues that are related to a certain population, community, or society. Societies and cultures are extremely dynamic and changing with time, therefore, evaluating people's acceptance of sharing genetic information is of a great value to meet the public prediction about the services. Previous studies revealed different factors that affect public attitudes towards genetics, such as age, gender, and educational level among others [3,4]. The attitudes and beliefs of public toward genetics were investigated in non-Arab populations [5,6]. In Jordan, the attitudes of public toward establishing biobanks were investigated [7]; however, their knowledge and views about genetic information are largely unknown. It has been suggested that assessing the knowledge is the first step in the decision making process for adopting or rejecting any innovation [8]. Understanding public views will help to identify the potential barriers in implementing genetic services into clinical practice. The public opinion about the genetic matters might change over time with the advances in genetic research to detect many diseases, accompanied with media attention. The main goal of the current study was to explore the knowledge, factors affecting knowledge, and views of Jordanian population on genetics-related issues. In addition, evaluating the intention to change the behavior in a group of healthy adults was assessed. To the best of the authors' knowledge, this is the first study in Jordan and Middle Eastern countries that elucidated the knowledge and views of public about genetics in the medical field. --- Materials and methods --- Study design A cross-sectional questionnaire-based study was conducted in Jordan between August 2017 and June 2018. The questionnaire was administered by face-to-face interview by the researchers in Arabic language. The research assistants were trained to ensure consistent interview and avoid any missing information in answering the questionnaire. Implied consent was obtained if the participants were willing to participate in the survey and the data were completely and irreversibly anonymized. A convenience sampling approach was adopted in the current study and the participants were recruited from different public places such as malls , community pharmacies , and blood banks from north, middle, and south Jordan. The ethical approval to conduct this research was granted by the institutional review board in Jordan University of Science and Technology and has been performed in accordance with the Declaration of Helsinki and its later amendments. --- Development of the questionnaire To cover all aspects of interest in this project, a bespoke questionnaire was designed. The questions of the questionnaire were selected from the literature [3,9] and some of them were newly developed after thorough discussion with the research team. In addition to the demographic details of participants, the questionnaire consisted of four sections that addressed four different themes: knowledge, attitudes, concerns, and potential to change behavior. The latter section assessed the behavior of a group of public healthy adults to change their lifestyle utilizing a hypothetical clinical case scenario. The questions were adapted from previous study [10]. The case scenario focused on receiving the results of a genetic test that revealed their risk of developing common diseases in Jordan, such as cardiovascular disease or diabetes mellitus [11,12]. The content of the questionnaire was reviewed for face validity and content validity by the members of the research team. In addition, a pilot study was conducted prior to the study . Feedback from the pilot study was used to enhance the clarity of the questionnaire and collected data was not included in the final analysis. All sections in the questionnaire were comprised of closed-ended questions. The knowledge section consists of eight items that had three possible answers . The attitude section consists of six statements with options and the concern section consists of seven statements with a 5points Likert scale . The internal consistency of the relevant sections in the designed questionnaire was tested using the Cronbach's α measure. The results ranged between 0.552 and 0.623, which indicated good reliability [13]. --- Statistical analysis Following data collection, the responses were coded and entered into Statistical Package for Social Sciences version 20. Continuous variables were presented as median , while categorical variables were presented as numbers and percent. Univariate analysis was conducted using Mann-Whitney U test for continuous variables and Chi-square test for categorical variables. In order to determine factors that were independently associated with level of knowledge, multivariate analysis using binary logistic regression was performed including all variables with p < 0.2 on univariable analysis. Odds ratio values and their 95% confidence intervals were calculated. Statistical significance was set at p value < 0.05. The main outcome in the study was the public knowledge which was dichotomized as knowledgeable and nonknowledgeable. For this purpose, the answers to the eight different questions of knowledge for each participant were labeled as categorical variables using a cutoff point for cumulative scores of correct answers. To calculate the knowledge score, all respondents who gave the wrong answer or stated "do not know" were coded "incorrect." A participant was categorized as knowledgeable if the sum of the scores was >5 and nonknowledgeable if the sum of the scores was ≤5 . For the purpose of analysis, we restricted response options for the concern section. For example, we combined both "very nonconcerned " and "nonconcerned" as one category and both "very concerned" and "concerned" as one category. --- Results --- Demographics In the present study, a total of 5644 persons were approached and 5000 of them were enrolled with a response rate of 88.6%. The median age of participants was 32 years old and more than half of them were female . About 60% of the participants were married and the majority of those had children . Approximately half of the public had university degree . In addition, about half of the respondents were recruited from the middle area of Jordan . Demographic details of participants are presented in Table 1. Almost half of the participants reported gaining their knowledge through media rather than other sources: healthcare providers , people , or the newspaper . --- Public knowledge In the present study, 43.4% of public were knowledgeable and the mean number of correct answers was 5.02 ± 1.791 . Only 5.8% of participants answered all questions correctly and few respondents reported knowing nothing at all about genetics . An overview of the questionnaire items and answers is presented in Table 2. Question one "There is a relation between consanguinity and genetic disease" was the one that was answered correctly by most respondents . On the other hand, question eight "Genetic information will not help in predicting drug response to some therapies" had the lowest number of correct answers . As shown in Table 3, the results of univariate analysis showed that participants' gender, level of education, having children and comorbidity in first-degree relatives were significantly associated with the public's knowledge . In addition, the results of multivariate analysis identified the independent factors that predicted a higher level of knowledge. Females were knowledgeable compared with males and participants who had bachelor degree were knowledgeable compared with those who had school education . Furthermore, participants who had children and those with comorbid conditions in their first-degree relatives were more likely to be knowledgeable . The knowledge of participants was also investigated by asking them about the methods that could be used for genetic testing. The majority of respondents were aware that genetic testing can be carried out using blood samples as compared with other methods: tissue , saliva , urine , and swab . --- Public attitudes About 70% of participants showed their willingness to use their genetic data in medical research. In addition, 75.7% of participants stated that they will be happy to inform their families about the results of genetic test for a specific disease. Furthermore, 66.9% of participants were willing to store their blood samples and health-related data in a biobank for long-term use. Importantly, the majority of participants were enthusiastic for genetic neonatal screening programs and premarital genetic test to the most common genetic diseases . The analysis indicated that genetic knowledge was significantly associated with all positive public attitudes . --- Public concerns Figure 1 reveals the potential concerns to genetic service implementation in Jordan regarding different issues that are related to genetic testing. Almost 60% of respondents were concerned that genetic testing would increase the complexity of providing healthcare. In addition, a half of participants were concerned that the cost of testing would be a barrier to the genetic service implementation. Importantly, about a quarter of respondents believed that stigmatization could affect the acceptance of genetic service in Jordan. Furthermore, a third of all participants were concerned that genetic test would affect their health insurance and employment. It is worth mentioning that approximately a sixth of participants were undecided regarding all potential concerns. The analysis indicated that genetic knowledge was significantly associated with most public concerns except for stigmatization where knowledgeable --- Potential to change behavior A total of 3824 adult healthy participants from the public answered the questions related to the potential to change behavior. The majority of public participants showed their willingness to change their lifestyles if they received genetic results that indicate their potential risk to develop cardiovascular disease or diabetes mellitus. Most of the participants would increase their physical activity and change their eating behavior . In addition, more than three quarters of smoker participants would quit smoking . The univariate analysis showed that genetic knowledge was significantly associated with the motivation of participants to change their lifestyle behaviors by improving their diet, physical activity and smoking cessation . --- Discussion Genetics is a rapidly growing field of science all over the world that aims to enhance the overall human health. Nevertheless, for practical implementation of genomics in clinical fields, perspectives of public toward this issue is a major constituent. In the current study, we explored the knowledge and views of Jordanian population on geneticsrelated issues. Female gender, participants who had bachelor degree, participants who had children and those with comorbid conditions in their first-degree relatives were independently associated with public's knowledge about genetics. In spite of positive genetic attitudes addressed by the participants, there were several concerns about the application of genetics in clinical practice. The high number of participants who were recruited from different regions of the country rendered our study findings representative of the Jordanian situation. Therefore, the findings of the current national study could be generalized and give an overview about public knowledge and views. The total number of Jordanian populations is 10,248,069 with about half females and half males. The highest percentage for those aged 25-64 and the lowest for those aged between 65 and over [14]. The numbers and percentages of the current study are consistent with Jordanian statistics. In our study, 58% of participants were females and 42% of them were males. The highest percentage was for participants aged between 25 and 64 years old and only 1.3% of participants aged ≥65 years old. Overall, the findings of the current study showed that less than half of public were knowledgeable about genetics. Previous studies reported varying percentages for the knowledge about genetics [3,9]. However, it is difficult to compare the percentage in the current study with that in previous studies as other populations were studied and different questionnaires and methods of knowledge calculation were used. The knowledge of our population could be explained by several reasons. First, most participants were highly educated with more than half of them had a bachelor degree. The same finding was reported by Etchegary et al. [3]. Second, about two-thirds of the participants were females who were more knowledgeable than males as shown in previous studies [15][16][17]. Third, the knowledgeable participants were mainly young adults who are consistent with previous reports [15][16][17][18]. Although the mean number of correct answers was relatively high , only a small percentage of the participants could correctly answer all the questions. This could be explained by the fact that the lay public might have general basic information about genetics but they are unfamiliar with precise genetic details. For example, previous studies reported a misconception among public regarding the specific area of genetic testing that is related to pharmacogenomics [19][20][21][22]. In addition, there is a limited or negligible enrollment of geneticists in the healthcare team in health settings in Jordan. A study by Ahram et al. revealed a lack of knowledge about genetic counselors among Jordanian population [23]. Hence, there is an urgent need to adopt educational and training programs for healthcare professionals to enhance the clinical application of genetics in Jordan. It is noteworthy to mention that about 83% of the participants realized that there was a relationship between consanguinity and genetic diseases. A recent published study reported a declining level of consanguineous marriage in Jordan [24]. We have shown in the current study that there was a positive relationship between public's knowledge and having first-degree relatives with comorbidities. Consistent finding was reported by Calsbeek et al. where they showed a positive relationship between having a chronic disease and the patient's knowledge [18]. This could be due to the knowledge seeking behavior of participants with chronic or hereditary diseases. Moreover, participants who had children were more knowledgeable than others. This could be explained by the parents' interest to understand the health status of their children especially their susceptibility to develop a genetic disease. The knowledgeable participants in the present study reported an increased likelihood of improving their diet, physical activity, and quit smoking. This highlights the importance of educational programs to enhance public awareness, especially young adults, to motivate their health lifestyle behavior improvement for preventing the development of chronic diseases. A previous study showed that knowledge of genetic susceptibility might change the behavior of healthy adults who were at risk to develop chronic diseases [10]. Knowledge alone might not be sufficient to change the health behavior. Other factors, such cultural imperialism, powerlessness, and socioeconomic status, are also important and should be assessed in a future qualitative study. In addition, further studies are recommended to examine whether disclosing the genetic risk information would interact with their potential to improve their lifestyle behavior. In the present study, the participants showed high levels of positive genetic attitudes which are consistent with previous reports among several populations [4,9,15,16,22,25]. More than two-thirds of the participants appeared to be in favor with the importance of genetic testing for short- and long- term purposes. Importantly, more than 85% of the respondents showed interest in neonatal screening and offering premarital genetic test to the most common genetic diseases. Different neonatal genetic screening tests and premarital testing are available in Jordan upon request. However, no national genetic screening or premarital programs are currently adopted for all public in Jordan due to the limited financial resources. Allum et al. in their meta review analyzed 193 nationally representative surveys on public understanding of science and identified a positive correlation between general attitudes towards science and general knowledge of scientific facts [26], which goes in parallel with the findings of our study. Despite the positive attitudes, the participants had many concerns regarding the clinical application of genetic testing. The major concern was the increased complexity of the healthcare delivery. To overcome this concern and to ensure the delivery of high quality services, a comprehensive plan should be implemented to improve the healthcare professionals' competencies, attitude, knowledge, and skills of clinical genetics. Most participants had concerns about the cost of genetic testing, consistent with populations of other countries [22,27,28]. This could be due to the fact that patients with genetic diseases confronted more difficulty in getting health insurance [9,15,16]. Beyond the previously mentioned concerns, about half of the participants were concerned about genetic discrimination or stigmatization, a phenomenon that was widely expressed among several populations [28][29][30][31]. Furthermore, it was found in the current study that a third of the respondents were concerned that genetic testing might reveal, by chance, the susceptibility to other diseases. This additional information might be unwanted and stressful for the individual, could result in the development of discrimination in obtaining life insurance and may lead to psychological problems [32]. The knowledgeable public in the current study appeared to have more concerns compared with nonknowledgeable group. This could be explained that knowledge has the potential to affect the opinion as knowledgeable people would be more realistic about the current situation of genetic application. In addition, low level of knowledge makes it difficult for participants to comment on any genetic related issues as there would be a difficulty in opinion formulation. The current study has some limitations. First, all sections in the questionnaire were comprised of closed-ended questions. Although closed-ended questions are easier to answer, they can guide the respondents in a certain direction and hence limit the respondents' abilities to express their values and concerns. Second, there is a potential of response bias as participants interested in genetics could have participated more than others. A previous study argued that excluded participants were from under-represented backgrounds and that structural inequalities should be taken into account rather than participants' interest/ attitude as the reason for declined participation [33]. However, our study included participants from different regional areas of Jordan with a high response rate. A future qualitative study that investigates factors affecting participants' exclusion is recommended. Third, there is a gap between "knowledgeable" and "nonknowledgeable" categorization that potentially might lose a lot of nuance in the response. This could have been reduced by using different answer options, such as low, moderate, and high. However, this was performed to simplify the analysis as eight questions in knowledge section were examined. In addition, previous studies which investigated the level of knowledge on other medical topics divided the participants into two groups [34,35]. In conclusion, the current study represented a national survey that investigated the knowledge of public and views toward genetics related-issues among Jordanian population. Our results suggest that female gender, educational level, having children, and having first-degree relatives with comorbid conditions were good predictors for public's knowledge about genetics. In spite of positive genetic attitudes addressed by our participants, they were concerned about application of genetics in clinical practice. The results of the current study will help to strengthen the importance of public education regarding the value of participation in genetic research in order to establish genetic related service in Jordan. As the culture background in Jordan is close to other countries in Middle East, the present results could be generalized to other Arab countries in the region. ---

This study aimed to explore the knowledge, factors affecting knowledge, and views of the Jordanian population on geneticsrelated issues. A cross-sectional questionnaire study was conducted in Jordan. The questionnaire was administered by faceto-face interview to the participants who were recruited from different public places. In total, 5000 questionnaires were collected from public population in Jordan and 43.4% (2171/5000) of them were knowledgeable. The public's knowledge was found to be associated with female gender (OR = 1.493, 95% CI = 1.280-1.741, p < 0.001), bachelor degree (OR = 1.853, 95% CI = 1.592-2.157, p < 0.001), having children (OR = 1.433, 95% CI = 1.162-1.768, p = 0.001), and having first-degree relatives with comorbid conditions (OR = 1.669, 95% CI = 1.431-1.946, p < 0.001). Although public in Jordan had positive genetic attitudes, they raised several concerns about the applications of genetics in clinical practice. Genetic knowledge was significantly associated with all positive public attitudes and most of their concerns (p < 0.001). These findings suggested that female gender, educational level, having children, and having first-degree relatives with comorbid conditions were good predictors for public's knowledge about genetics-related issues. Public education about the value of participation in genetic research as well as educational and training programs for healthcare professionals are recommended to assist in establishing genetics-related services in Jordan.

The concept of social identity offers insights into how social processes impact positively on health and well-being . Although the mechanisms involved in the relationship between social identification and well-being are diverse, it seems they entail a common theme: identifying with social groups is associated with the belief that one can deal with stressful challenges. In what follows, we examine how and why social identity processes may be relevant to well-being and whether identifying with a large-scale social group membership is associated with such benefits. We then investigate this issue empirically with questionnaire data from rural north India. We ask if identifying as a Hindu is associated with well-being via its associations with believing that one can deal with the challenges of everyday life . Specifically, we test a model in which social identification as a Hindu has an indirect and positive association with well-being via greater stress-related self-efficacy. As we will see in the succeeding text, the fact that we investigated identification with a large-scale group, and that we did so in India, gives these questions additional interest. --- Social Identification and Well-being It is well known that the more social relationships and the more networks one has, the better one's health and well-being . The social identity perspective to group behavior enriches this literature in various ways. It argues we can think of ourselves in different ways: sometimes in terms of our personal identities and sometimes in terms of social group memberships . These latter can be small-scale face-to-face groups but also large-scale . The consequences of such social identifications are several and have various implications for well-being. Most obviously, thinking of oneself and others as sharing a group membership implies that these others are one's fellows, and this increases the degree to which people help each other . Moreover, on the basis of one's common social identification, one expects to be recognized as a fellow group member and to receive the benefits that come with such membership . In turn, the more one identifies with a group and the more one expects social support from other group members, so the better one's well-being in the face of intense stress . Research also shows that it is the level of one's subjective identification with the group rather than the actual level of contact with other group members that predicts well-being . Moreover, research shows that manipulating people's awareness of the number of their group memberships can impact upon people's resilience in the face of immediate stressors: with more group memberships psychologically salient, participants exhibit greater resilience . Again, the implication is that social identifications facilitate coping. Given such potential benefits, it is appropriate to consider the types of social identification associated with improved well-being. Not all will: most obviously, some group identifications are associated with unhealthy practices and norms that de-value health as a goal . Moreover, group identifications may have contradictory implications. For example, even if associated with unhealthy practices, a social identification may still bring benefits in the social support it allows one to access. Furthermore, with regards to the potential benefits, it is important to consider the issue of group size and scale. Thus far, most research has tended to focus on small groups whose immediate circumstances are extremely challenging , and it is not immediately obvious that larger-scale identifications would have similar implications nor that they would facilitate coping with more everyday stressors. This is the focus for our own research. --- Social Groups: Scope and Scale Large-scale identifications differ from face-to-face groups in various ways and can satisfy different identity motives . The mechanisms by which large-scale social identifications could be associated with well-being are diverse. On the negative side, some cultural norms may encourage unhealthy practices or discourage certain medical interventions . On the positive side, some cultural traditions may provide a sense of trans-historical continuity in a way that work teams may not. As work on national identifications has shown, such continuity can provide a psychologically comforting sense of meaning . Moreover, there is evidence that identifying with larger-scale groups can facilitate resilience in the face of immediate stressors: in the aforementioned work by Jones and Jetten showing that one's awareness of the number of one's group memberships affects resilience, some of the memberships involved were large-scale . The benefits of identifying with a large-scale group membership are also suggested in studies that show those who are religious tend to fare better in terms of health than their nonreligious counterparts . In part, this is because certain religious traditions discourage practices that bring health risks . It can also be because religious beliefs and cultural practices encourage ways of explaining life experiences, which limit the impact of negative events . However, it is also because being religious implies a social identification with a group . This group identification can bind one to cultural traditions . It can also bind one to fellow believers in supportive social networks-for example, church congregations . Moreover, such is the scale of the identification , that it not only connects one to an existing network, but also provides the basis for linking one to new networks if old ones should be disrupted. This is apparent in work by Ysseldyk, Haslam, and Haslam who found that among older adults undergoing significant life-course transitions , a religious identification allowed individuals to more easily join and establish new face-to-face groups . In the current research, we explored further this potential for large-scale social identifications to be associated with better well-being. On the basis of previous research, it appears that social identifications may provide individuals with a sense of continuity, purpose and meaning, social connection with others, and access to social support, and we asked if identifying with a large-scale social group may therefore predict people's beliefs about the degree to which they can cope with stressful situations . That is, we asked if large-scale social identifications have a wider association with well-being beyond the immediate and known demands of particular occupations or particular physical circumstances and whether this is bound up with people's beliefs about their general ability to deal with the various unpredictable demands thrown up in everyday life. Our research is distinctive in another way, too: it was conducted in rural north India. Thus far, most work on the association between social identification and well-being has been conducted in urban settings in the industrialized West, and the degree to which such research is relevant to other sites cannot be assumed. Accordingly, our research site has particular merits: it allows analysis of the relevance of identifying with a group for coping with the everyday stressors that are integral to rural life in a developing country. To find relationships between social identification, stress-related self-efficacy, and wellbeing in such a setting would be good evidence for the wider applicability of recent theorizing on the social bases of wellbeing . --- Community Identity in Rural North India Social identities in rural north India are complex. There are local occupation-based caste identities with their own histories and local social significance . There are also larger-scale community identities such as 'Hindu'. Although factors associated with social coordination in complex societies help explain the cultural evolution of such identities , historical research shows that this identity takes its modern form and significance because British colonial administrators sought to delineate and manage the constituent elements of Indian society . In contemporary India, this Hindu identity permeates the fabric of everyday life. It has spiritual and religious elements, yet also constitutes a cultural identity such that all aspects of life, from who lives where in a village to whether and how one relates to others, are informed by this identity. Thus, for a Hindu to self-define strongly in terms of this collective identity is not simply to subscribe to a set of spiritual beliefs but to define the nature of one's connections to others in one's everyday transactions. Investigating the association of such a social identification with well-being is complicated. There are questions concerning psychology's approach to culture and cultural differences . These can be particularly important when studying health and well-being. Stress-related symptoms can be somatised differently in different cultures , and this highlights the need for culturally-appropriate measures of wellbeing. Other issues concern the nature and significance of the social relationships found in different cultures. Researchers often differentiate between individualistic and collectivist cultures and an important aspect of this contrast concerns cultural differences in the weightings people give to information provided by knowledge of an individual's group memberships and family connections. For example, Owe et al. show such information to be more important in collectivist cultures. Moreover, in different cultures there may be different ways in which social relationships inform one's identity , and there is evidence that in some collectivist cultures , group memberships tend to be depicted in terms of networks of interrelated individual members . However, despite such potential cross-cultural complexities, there are reasons to believe the social identity concept can be useful in non-Western contexts. Although there may be different bases for group entitativity in non-Western contexts, Yuki et al. observe that this does not undermine the significance of group memberships for cognition and behavior. Yet, it remains important to recognize the cultural meanings of the relevant identities . For example, although it is common to assume that a shared social identification implies that group members are seen as interchangeable exemplars of a horizontal community , this should not be presumed to be a universal characteristic of social identities . In India, a common Hindu identification can exist alongside hierarchical caste categorizations, and although these latter remain bases for differentiation, this does not necessarily subvert a superordinate Hindu identification. Indeed, caste identities are routinely represented as complementary and as allowing the Hindu 'body' to function . Another issue relating to the application of well-being research conducted in the industrialized West to rural north India concerns the social significance of religion. Most research on religiosity and well-being has been conducted with Christian denominations in the US. This has prompted calls for a wider range of religious identifications to be studied . This requires cultural sensitivity. For example, it would be easy to assume that as public Christian prayer typically involves a set service at which a congregation gathers , then prayer at a Hindu temple has similar implications. However, this is misleading , and attendance at Hindu temples cannot be taken as a measure of network participation. It is also important to note that in religious countries , religious believers are socially valued, and such social value can impact upon the degree to which people receive various psychological benefits . In other words, some issues that are important in US Christian communities may be less important in Hindu practice, and other issues of lesser significance in more secular cultures may be more important in India. --- The Current Research Our research involved individuals who were, in formal terms, Hindus, and explored if and how their subjective identification with this large-scale social group was associated with their well-being . In particular, we addressed the question of whether this identification was indirectly associated with better well-being via its association with stress-related self-efficacy . As far as we are aware, there is no research in India exploring this relationship . Given the complex associations of a Hindu identification in India , our analyses also took into account two other factors, which could be relevant to well-being. The first concerned participants' level of engagement in religious practices. The second was participants' sense of their social standing in the community . In addition to addressing these two key alternative predictors of stress-related self-efficacy, our analyses of the relationship between social identification as a Hindu, stress-related selfefficacy, and well-being also controlled for a number of background socio-demographic variables. Specifically, we controlled for the effects of age, gender, caste, and marital status, which Indian research shows are relevant to health and well-being . We also controlled for educational-level . --- METHOD Sample Participants were recruited in the rural area within a radius of 100-120 km from Allahabad . We approached known local contacts and recruited others through a process of snowballing. All participants self-categorized as Hindu. The average age was 64.1 years . Three hundred seventy-three were female and 419 were male. Seven hundred nineteen belonged to the general caste category and 73 to the other backward caste category . These broad caste categories are used by the Indian Government to differentiate between those who are relatively privileged and those who are less so . Other caste categories below OBC exist but were not represented in our sample. Six hundred twenty-seven were married and 165 were widowed. 313 were illiterate , 380 had primaryto-intermediate education, and 99 were university educated. This is not a representative sample of the north Indian population. Rather, it is an older and higher-caste sample, and this is because we targeted villagers considering an agerelated pilgrimage . However, a relatively homogenous sample is entirely suitable for our attempt to investigate a process model of identification and well-being. --- Design and Procedure Our data derive from an orally-administered questionnaire. The scales were developed through extensive piloting and were translated and back-translated by two independent groups. Any differences in the translations were resolved by improving the items. The final items were piloted again with illiterate and literate Hindi-speakers. The questionnaires were administered by a team of 10 Hindi-speaking field investigators in participants' homes. Completion of each questionnaire took approximately 30 minutes. So as to make the concept of a five-point scale meaningful, we showed participants drawings of five glasses containing increasing levels of water and explained how participants could communicate the level to which they agreed with the statements they would hear . When approaching potential participants, the researchers identified themselves as coming from the University of Allahabad and as being interested in villagers' lives and experiences. The researchers gave an overview of the questionnaire's contents and addressed any questions that potential participants raised. After this, consent was sought . Given literacy issues, the explanation of the research, the request for consent, and the giving of consent was oral . No incentives were offered for participation. --- Measures Our questionnaire measured participants': level of social identification as a Hindu; perceptions of efficacy in dealing with everyday stressors; well-being; engagement in religious practices; and perceptions of their social standing in their community. Except where stated otherwise, all items were answered on five-point scales using the drawings of glasses of water explained earlier. The scale names and their items are as follows: --- Social Identification as Hindu --- Stress-Related Self-Efficacy Five items based on Schwarzer and Jerusalem's Generalized Self-Efficacy Scale but modified for use in India. Using the stem 'Over the last week, to what extent have you been feeling you…', the items included 'can manage all the demands on you?', 'have the capabilities to do the things that matter to you?', 'can manage your life well?', 'are in control of your life?', and 'have the skill/abilities to live your life as you want?' . --- Well-being This was measured using various scales. The first comprised three items from the internationally-used core module of the Centers for Disease Control and Prevention Health Related Quality of Life Measure . Using the stem 'Over the last week, how would you describe your…', the items included 'physical health'; 'state of mind'; and 'energy levels?' . We also included items adapted from scales developed for use in the Indian Subcontinent, which take account of the somatisation of stress-related symptoms . Using the stem 'Over the last week, to what extent have you felt…', three questions concerned psychological symptoms . Three others concerned physical symptoms . These six items were formed into two scales . --- Religious Practices Six items addressed participants' level of engagement in religious practices. Three items concerned Religious Practices at Home. Using the stem 'In the last week, how often have you…' the questions were 'performed morning pujas ?', 'performed evening pujas?', and 'chanted religious texts in your home?' Using the same stem, three concerned Religious Practices in Temples and asked how often they had 'gone to temples?', 'offered fruits/sweets/flowers in the temple?', and 'read or chanted religious texts in the temple?' For these items, participants reported how many days over the last week they had engaged in each practice . --- Perceived Standing in the Group Five items asked about the extent to which participants thought that other people in their neighborhood or village treated them as members of standing in the village community. Using the stem 'Currently, to what extent do others like your neighbours and other villagers…' the items included: 'accord you high status?', 'like you?', 'take you seriously?', 'admire you?', 'respect you?' . We assessed if these items formed constructs in the intended manner using Confirmatory Factor Analysis in AMOS 17.0 . We specified a factor structure with eight latent variables that were allowed to co-vary freely. Symptoms of ill-health were specified as two discrete factors: psychological symptoms of ill-health and physical symptoms of ill-health. Likewise, religious practices at home and religious practices in temples were specified as two separate factors. The chi-square goodness of fit statistic for the specified factor structure was significant χ 2 = 914.37, p < .001). However, with a sample of this size, this does not necessarily mean the model should be rejected . Additionally, we used the comparative fit index , the root mean squared error of approximation and the standardized root mean squared residual to evaluate model fit. It has been proposed that values of >0.95 for the CFI , <0.05 for the RMSEA , and < 0.05 for the SRMR indicate a good fit between a specified model and observed data. Given these recommendations, our specified factor structure had a good fit , SRMR = 0.03). The measurement weights for the respective latent variables specified in the CFA are displayed in Table 1. 1 Background information about participants' scores on these scales appear in Table 2. This depicts the scale means, standard deviations, Cronbach alphas, and inter-scale correlations. The reliabilities for all measures apart from two were ≥0.78. The reliabilities for the religious practices at home were rather lower than for those in the Temple -probably because home-based prayer is rather more variable . The overall reliability of the Physical Symptoms of Ill-Health scale was also low . However, the fact we have three measures of well-being means that this is less of an issue that would otherwise be. All measures were positively scored except symptoms of ill-health . The inter-scale correlations show that higher Social Identification as a Hindu was associated with better self-assessed health . Table 2 also shows that greater stress-related self-efficacy was associated with higher involvement in religious practices , higher perceived standing in the group, and higher social identification as a Hindu. 1 In order to examine the equivalence of this factor structure across the sociodemographic categories recorded in this study, we conducted a Multigroup Analysis in AMOS 17.0 . These results are presented in the supplementary materials associated with this paper. We found complete configural and metric invariance, and partial scalar and residual invariance, across the respective socio-demographic categories recorded in the study. --- RESULTS Our analyses entailed two steps. The first investigated if stressrelated self-efficacy was associated with participants' social identification as a Hindu. The second involved testing a model in which Social Identification as a Hindu was associated with well-being via stress-related self-efficacy. In both analyses, we controlled for effects associated with the socio-demographic categories recorded . Moreover, given the complex meaning of a Hindu identification, we also investigated the degree to which a Hindu social identification had associations with well-being that were distinct from those associated with participants' level of religious practices and perceived standing in the group. Both analyses employed Structural Equation Modelling . Given the equivalence of our measures across socio-demographic categories , these analyses used pooled data. --- Factors Associated with Stress-Related Self-Efficacy Our first analysis specified a model in which participants' level of social identification as a Hindu, level of religious practices , and perceived standing in the group were entered as predictors of stress-related self-efficacy. The socio-demographic categories were entered as control variables . Accordingly, stress-related self-efficacy was the only endogenous variable with the others being exogenous . The chi-square goodness of fit statistic for the model was significant = 771.57, p < .001). The remaining goodness of fit indices were satisfactory ; SRMR = 0.04). Table 3 summarizes the findings. With regards to the control variables, it shows that age, gender, and educational-level were associated with stress-related self-efficacy . With regards to social identification as a Hindu, it shows that this had associations with religious practices at home and in temples, and also with perceived standing in the group. However, and of more theoretical significance, it shows that stress-related self-efficacy had positive associations with both perceived standing in the group and social identification as Hindu. Of these associations, the former was stronger . Yet, the significance of the latter association constitutes evidence for our key prediction that a stronger social identification as a Hindu would be associated with greater stress-related self-efficacy. --- Factors Associated with Well-being In the second step of the SEM analysis, we investigated the factors associated with our three measures of well-being . Our model specified paths from our key predictor variable-social identification as a Hindu-and the two alternative predictors , to the well-being measures via stress-related self-efficacy. It also specified direct paths from these variables to the wellbeing measures. Again, the socio-demographic categories featured as controls . The exogenous variables were allowed to co-vary freely with one another. So too, the well-being measures were allowed to co-vary freely with one another. We used the bootstrapping procedure in AMOS to estimate standard errors and bias-corrected confidence intervals for the direct and indirect estimates in the model . Although the chi-square statistic was significant = 1160.20, p < .001), the remaining goodness of fit indices , SRMR = 0.04) indicated an acceptable model fit. Figure 1 depicts the significant paths in the model . Tables 4 and5 report the results relating to the model's direct and indirect paths. The co-variances for the exogenous variables are the same as reported in Table 3, whereas those for the well-being measures feature in Figure 1. Social identification and well-being in India 793 Consistent with the first step of the SEM analysis reported earlier, Figure 1 and Table 4 show direct, positive, and significant paths from perceived standing in the group and social identification as a Hindu to stress-related self-efficacy . Religious practices at home and in temples were not associated with stress-related self-efficacy. With regards to well-being, greater stress-related self-efficacy was associated with better outcomes . None of the direct or indirect paths from religious practices at home and in temples to the wellbeing measures were significant. Perceived standing in the group was not associated directly with the three well-being measures, but was associated indirectly with these measures via stress-related self efficacy. As explained earlier, we were particularly interested in the association between social identification and well-being via stress-related self efficacy. As Table 4 and Figure 1 show, we found evidence for indirect associations between social identification as a Hindu and our three well-being measures via greater stress-related self-efficacy. Taken together, these three indirect effects constitute evidence that the level of one's social identification as a Hindu predicts better well-being because this identification is associated with the belief that one can cope with life's stressors. Our analysis also revealed residual associations between social identification as a Hindu and two of the well-being measures after controlling for the pathway through stress-related selfefficacy. These residual associations were more substantial than the indirect effects and showed heightened identification was associated with more psychological and physical symptoms of ill-health. Quite why we found such residual effects indicating an association between greater social identification as a Hindu and more psychological and physical symptoms of ill-health is unclear . As explained earlier, these analyses included the sociodemographic variables as controls. The direct and indirect associations of these variables are reported in Table 5. 2 Overall, the model explained 28% of the variance in stress-related selfefficacy, 41% of the variance in self-assessed health and 28% and 44% of the variance in psychological and physical symptoms of ill-health, respectively. Social identification and well-being in India 795 --- DISCUSSION The most theoretically interesting features of these data concern the relationship between participants' social identification as a Hindu-their subjective sense of belonging to a large-scale social group-and their well-being. The correlations in Table 2 pointed to a positive relationship between social identification as a Hindu and one of the measures of well-being and to no relationships with the other well-being measures . However, once we took into account other variables, we found a more complex set of relationships . Most importantly, we found that participants' social identification as a Hindu was associated with better well-being through its association with participants' judgements of their own stress-related self-efficacy. That is, as participants' social identification as a Hindu increased they reported greater stress-related self-efficacy and better wellbeing. Interestingly, this relationship between social identification as a Hindu and better well-being via increased stress-related self-efficacy was found for all three measures of well-being . This is consistent evidence that participants' identification with a large-scale group is associated with better well-being via their sense of being able to cope in the world. What makes these associations particularly interesting is that after controlling for the pathway through stress-related selfefficacy, we found residual associations between participants' social identification as a Hindu and two of the three well-being measures . These show that a higher identification was associated with more symptoms of ill-health. What lies behind this is unclear. It is possible that a higher identification is associated with practices and routines that take their toll . However, we also know that people's interpretation of their bodily experiences as 'symptoms' and as worthy of concern is bound up with the beliefs and values associated with their social identifications . Thus, it is conceivable that a higher Hindu identification is associated with a cultural worldview in which psychological and physical experiences are given distinctive meanings and experienced as 'symptoms' . Clearly, there is a need for further work, which explores the particular meanings of Hindu identity, the way these affect health-related practices and the way in which bodily experiences are interpreted as 'symptoms'. Whatever the reason for these residual associations between participants' social identification as a Hindu and the measures of psychological and physical symptoms of ill-health, the key point to highlight is that we have clear evidence that a higher social identification as a Hindu had an indirect effect on well-being through its association with the belief that one can cope with everyday stressors . As we noted in the introduction, there are several mechanisms that might underlie this relationship. For example, social identifications can give life meaning and purpose, provide a sense of continuity, connect one with others, and provide a basis for social support. Future work should address which of these elements is particularly relevant for large-scale social groups and how this depends on their cultural contents. A further set of questions concern the relative strength of the association between social identification, stress-related self-efficacy, and well-being. As concerns the associations with stress-related self-efficacy, our own analyses show that one's perceived social standing has a stronger link than identification as a Hindu. This is unsurprising: social standing brings innumerable social and material resources, which impact upon the degree to which one may feel efficacious. With regards to the indirect association between social identification and wellbeing via stress-related self-efficacy, it is important to note that the effects are small. However, it is also appropriate to note that the type of coping we explored in our scale is rather general. That is, although our scale is similar to existing measures of stress appraisals used elsewhere , it differs in that our scale emphasizes general as opposed to domain-specific self-efficacy. This makes good sense: we were concerned with a large-scale social identification and the degree to which people felt able to cope with everyday stressors. However, such a lack of specificity is likely to introduce measurement error. Future work on largescale social identifications could employ a number of scales, each designed to tap more specific domains of life stressors. In pursuing the relationship between social identification and well-being, future work should ideally have a longitudinal dimension. As with most studies on social identity and wellbeing, our research was cross-sectional. The limits of such a design are underlined by research showing that one's level of identification with a social group can itself be a function of the degree to which it satisfies particular identity motives . Future work should also consider the manner in which stress-related self-efficacy impacts on well-being. It may be that those who feel in control of their lives flourish, obtain better resources, and hence experience better well-being . It may be they undertake less risky behaviors and thereby experience better well-being. They may also exhibit better functioning of the immune system . Although these questions concerning the further specification of the link between a Hindu identification and well-being remain, it is important not to lose sight of what we have shown. Three important conclusions can be drawn from our study. First, large-scale as well as small-scale social identities can be associated with well-being, and this is true not only when the group faces particularly demanding circumstances but also in everyday life. Second, these identifications can have complex and contradictory associations with well-being: identifying with a group can be associated with poorer outcomes as well as with better outcomes. That is, although a social identification can be associated with the sense that one is able to meet the challenges thrown up by everyday life, and thus be indirectly associated with better outcomes, the same social identification may for other reasons be associated with poorer outcomes. Third, on the importance of social identity processes and their relevance to the well-being of ordinary people, we show this is not limited to the urban western world where most work has been conducted. This provides a robust test of a social identity approach to well-being. It would certainly be plausible to suggest that in poorer countries where people are more exposed to harmful material conditions and where health systems are less organized, then social identity processes would be of less relevance to explaining well-being. However, it seems that wherever we go, whether we look at small work-groups or large social categories, at highly stressful circumstances or mundane settings, social identity cannot be discounted as a factor in well-being. --- ACKNOWLEDGEMENT This work was funded by an ESRC research grant . --- SUPPORTING INFORMATION Additional supporting information may be found in the online version of this article at the publisher's web-site.

Identifying with a group can contribute to a sense of well-being. The mechanisms involved are diverse: social identification with a group can impact individuals' beliefs about issues such as their connections with others, the availability of social support, the meaningfulness of existence, and the continuity of their identity. Yet, there seems to be a common theme to these mechanisms: identification with a group encourages the belief that one can cope with the stressors one faces (which is associated with better well-being). Our research investigated the relationship between identification, beliefs about coping, and well-being in a survey (N = 792) administered in rural North India. Using structural equation modelling, we found that social identification as a Hindu had positive and indirect associations with three measures of well-being through the belief that one can cope with everyday stressors. We also found residual associations between participants' social identification as a Hindu and two measures of well-being in which higher identification was associated with poorer well-being. We discuss these findings and their implication for understanding the relationship between social identification (especially with large-scale group memberships) and well-being. We also discuss the application of social psychological theory developed in the urban West to rural north India.

for students. After a brief description of the partnership, examples of current research that contributes to public policy at the federal, state, and local/provider levels are presented, along with students' roles in these efforts. Next, we share factors that contribute to our success and challenges . We conclude with a look to the future, specifically, the need to engage students in our academic programs, maintain our long-standing relationships, and build new, cross-organizational ties so that future gerontologists will continue our mission of doing work that positively impacts older people. --- SESSION 4440 Abstract citation ID: igad104.1777 --- PREPARING FOR ADVANCING ILLNESS AND END OF LIFE WITH DEMENTIA Chair: Jenny van der Steen Dementia involves coping with advancing illness that involves cognitive and functional decline. Also decision-making capacity diminishes, which implies that preparing for a future and taking opportunities to exercise individual and relational autonomy is particularly relevant in the case of dementia. Engaging the person with dementia and family caregivers meaningfully in formal or informal advance care planning conversations may require addressing specific challenges. Health care providers often lack time or are reluctant to broach these conversations. Often conversations are delayed even in institutional long-term care settings as residents, their family caregivers and care providers may all view advance care planning as uncomfortable and difficult to initiate. Special support for health care providers to develop skills, confidence and taking initiative is needed. This symposium presents about new, large qualitative and quantitative studies with contributions from the UK and Canada, and international work. It addresses preparing for an uncertain future through advance care planning and decision making about future care and treatment which may include conversations about the end of life. We present and evaluate tools to support the process across settings. Abstract citation ID: igad104.1778 --- ENGAGING PEOPLE WITH DEMENTIA AND THEIR FAMILY IN ADVANCE CARE PLANNING: A 33 COUNTRY DELPHI STUDY Jenny van der Steen 1 , and Ida Korfage 2 , 1. Leiden University Medical Center,Leiden,Netherlands,2. Erasmus MC,University Medical Center Rotterdam,Rotterdam,Netherlands Reaching a consensus about a definitional framework of advance care planning in dementia was a main goal of an international Delphi study conducted by the European Association for Palliative Care. Four rounds with a panel of 107 experts from 33 countries were conducted between September 2021 and June 2022. The framework comprised a definition and three issues that are of particular importance in the case of dementia, which were capacity, family, and engagement and communication. Focusing on the latter,

well-being of older adults (e.g., the role of Medicaid funding in home-and community-based care and the effects of nurse staffing levels in nursing homes). As new policy questions emerge, the foundational work of the policy center continues to guide undergraduate and graduate education at the USF School of Aging Studies. This presentation describes the multiple opportunities to incorporate policy instruction into an interdisciplinary gerontology program. These opportunities depend on having faculty who maintain ties with the organizations that provide direct care and services to the public and who connect the policy lessons they learn through these relationships to their teaching and research. Another key element is access to health services data needed to answer critical policy questions. The goal of such an approach is to enable gerontology students to engage in policy analysis and development and to understand the essential links between research, public policy, and practice that improves the lives of older adults.

869 married individuals living in Switzerland were compared, using multiple regression analyses. Results: Differential predictive patterns for social loneliness between the two divorced groups and the married group were observed. For the short-term divorced , higher extroversion and agreeableness and lower neuroticism were associated with lower levels of loneliness. For the long-term divorced and for those who remained married, extroversion was similarly important for loneliness. Additionally, higher levels of self-continuity and multiple group memberships predicted lower loneliness, but the short-term divorced did not benefit from them. Having someone to help overcome the divorce benefited members of both divorced groups. A new partner was related to less loneliness, but only in the long-term divorced group. Conclusions: Our findings demonstrate that the effects of psychological and social resources on social loneliness vary by adaptation phase. Although extroversion is beneficial for all divorced and married individuals, other personality traits play a more decisive role in the initial adaptation phase. Identity-promoting resources are beneficial only later in the adaptation process. To be successful, professional interventions must be tailored as needed. --- Introduction Losing one's spouse through divorce represents an important critical life event and its frequency is rising among older adults [1]. Social loneliness can be one of the negative consequences of divorce [2], increasing the chances of outcomes such as poorer health and higher mortality [3]. Although time heals some wounds and many divorcees can accept their new realities and recover their predivorce levels of well-being [4,5], not everyone can adapt to and accept the changes that divorce may impose on their lives, often leading to prolonged feelings of social loneliness [6]. Adaptation refers to regaining the level of well-being that one had before the occurrence of the critical life event [7]. Investigation of the factors that may affect the adaptation process is of great importance to identify why some individuals remain vulnerable while others successfully overcome divorce and move on with their lives. When people face dissolution of marriage at a later point in their lives, coming to terms and coping with divorce may be particularly challenging. For example, many of them had for much of their lives a social identity of husband or wife. Finding a new partner may also become more difficult. As only a few studies have addressed divorce in the second half of life [6], not much evidence exists about predictors that help in adaptation to divorce at that age. Divorcees may also experience various post-divorce phases, during which, depending on the time passed since this critical life event, specific resources may be particularly beneficial for adaptation. According to Amato's [4] divorce-stress-adjustment model, divorce is a threephase process. It starts with separation and/or dissolution of marriage, followed by a first post-divorce adaptation period, which lasts about 2 years and is primarily characterized by distress, and the final adaptation phase, during which the individual no longer feels divorce-related distress and returns to pre-divorce levels of well-being. Empirical evidence for these specific adaptation phases has been found in several longitudinal studies that examined adaptation to partner loss either through divorce [8,9] or bereavement [5,10]. Following this model, it is likely that specific factors are responsible for coping in the various post-divorce phases. Although some resources may be beneficial regardless of the divorce phase , other resources may be differentially important during these post-divorce phases. In the acute adaptation phase after divorce, well-established cognitive and behavioral ten-dencies may regulate how the individual adapts while staying busy reorganizing urgent practical aspects to ensure that everyday life continues as well as possible. After sorting out immediate pressing issues, the individual may have more time and energy in the later adaptation phase to consider divorce-related changes more broadly, including reevaluating one's identity and integrating aspects of the new situation into the self. Yet specific adaptation resources' time dependency has received little attention. This study aims at addressing these research gaps by investigating how various psychological and social resources are related to social loneliness as indicators of successful adaptation to divorce in later life and whether their usefulness varies across post-divorce adaptation phases. --- Later-Life Divorce and Loneliness Divorce in advanced age represents a new phenomenon associated with recent demographic changes [1]. Individuals in the second half of life experience divorce as a highly distressing event and as a crisis that is "off-time" even if the divorce is a voluntary dissolution of marriage [5]. Additionally, divorce often leads to the disruption of social relationships, as friends of the formerly married couple usually tend to feel closer to one of the partners and choose sides [11]. The resulting shift in social networks contributes to the feeling of distress that is common among divorcees [5,8]. In later life, the distress may become even greater as it is more challenging for older individuals to find new social partners. Although research has mainly focused on dissolution of marriage at younger ages, the limited findings on late-life divorce indicate poorer adjustment compared to that of younger individuals [12], suggesting that divorce is a more stressful or a more difficult experience to cope with in older age. Many individuals who go through divorce feel lonely. Although long-term married individuals can also experience social loneliness [4] and a decreasing social network over time [13], divorcees have been found to be more prone to social loneliness, particularly due to disruptions of social relationships associated with divorce [14]. Research has shown that feelings of loneliness in the second half of life remain relatively stable or even diminish in advanced age, and they are not caused by isolation but by being unable to meet one's need for socializing with valued partners [15,16]. However, the empirical evidence is still inconclusive regarding loneliness in the context of critical life events in the second half of life, such as divorce [17,18]. When people grow older, their social circles are likely to diminish due to loss of loved ones and other so- cial partners; also, health and mobility issues contribute to the reduction of social contacts [18]. However, it is not only the quantity of social contacts that affects social loneliness but also their quality [19]. Therefore, individuals may have fewer social partners with advancing age, but the remaining relations may be of higher quality [18]. In the context of divorce, losing valued social partners and the inability to replace them with others of equal importance can have long-term consequences regarding social embeddedness and well-being in later life. In his theory of loneliness, Weiss [20] suggested that social loneliness is experienced when individuals lack an engaging social environment -for instance, when they have only limited and unsatisfying contact with family, friends, or community members. Therefore, dissolution of marriage in the second half of life can result in additional loss of important social partners [14], leading more generally to maladaptation to divorce and, more specifically, to greater social loneliness. --- Factors Related to Post-Divorce Loneliness Investigation of factors related to loneliness is important, as lonely individuals have substantial mental and physical health risks ranging from depression to premature mortality [21,22]. Specifically, individuals experiencing marital instability, those with unsupportive social networks [23], those without children, and most specifically women [4] are at risk of feeling socially lonely, underscoring the importance of social resources. In countries with strong societal norms, such as in Switzerland, gender is an important factor to consider, as the dissolution of marriage is more challenging for women than for men [24]. Women are encouraged to reduce their employment rates or stop any work activity after marriage or motherhood, leading to poorer financial and social resources. Thus, in Switzerland and other conservative societies, the re-partnering rate is higher than in other countries, such as the USA or Germany [25], since being divorced may come with greater financial and social challenges. Finding a new partner may be an essential part of overcoming the divorce, as it can protect against social loneliness [4] but also against financial and societal strains. However, re-partnering in later life may be a challenge for some individuals. In divorce, other social partners, such as children or close friends, can provide more readily available social support. Hence, we hypothesize that, among both divorced groups, having a new partner, children, and someone helping to overcome divorce will be related to a lower level of social loneliness. Besides social aspects, psychological resources, such as personality, also play a role in adaptation to divorce. Personality traits influence how individuals cope with critical life events [26], and they are responsible for how a person engages in social life [27], influencing post-divorce loneliness levels. More neurotic individuals tend to experience emotional instability and relational deficits [28]. Higher levels of neuroticism are then likely to lead to more emotional vulnerability, unsatisfied needs for socialization, and social loneliness. Individuals who are more extroverted and more agreeable feel less socially lonely, as, for them, it is easier to approach compatible social partners and create meaningful relationships [27]. Being conscientious entails some personal qualities that are appreciated and valued by others, such as being hardworking, reliable, and self-disciplined [29], making conscientious individuals more likely to be surrounded by social partners or embedded in groups. Finally, being more open to new experiences offers more possibilities to meet new people [27] and therefore to satisfy the need for social connectedness. Regarding personality and divorce, middle-aged women undergoing divorce showed higher adaptability when they had higher levels of extroversion and openness and low levels of neuroticism [5]. Besides being linked to enhanced coping with critical events, extroversion and neuroticism have also been found to be related to lower social loneliness in the general population [28]. Regarding personality, we therefore expect that higher extroversion, openness, agreeableness, and conscientiousness, as well as lower neuroticism, will be related to lower social loneliness. Other psychological resources, such as identity mechanisms, may promote adaptation to critical events. Continuity theory suggests that during life changes, a person seeks to maintain or protect a sense of identity [30]. To do so, individuals engage in cognitive strategies to experience continuity, remembering persistent inner-psychological aspects and continuous social-environmental aspects [30]. Experiencing high self-continuity has been found to contribute to adaptation after critical life events [31] and may therefore prove similarly important in the context of divorce. Divorce may raise identity questions such as, "Who am I now?" or, "Am I the same person as before the divorce?" Chandler and Proulx [32] suggest that self-continuity enables individuals to connect the various pieces of their past, present, and future into a coherent story that reflects a sense of identity stability. As being a spouse represents a central element of many older adults' self-defi- nitions, particularly when favoring traditional life forms as is common in Switzerland [33], the need for self-continuity in divorce can be high. Feeling like the same person as before the divorce may be important for maintaining a clear sense of who one is, which represents a central prerequisite for social interactions and, specifically, for the development and the maintenance of a supportive social network of family and friends. Hence, in line with this reasoning, we hypothesize that high perceived selfcontinuity contributes to feeling less social loneliness after divorce. Individuals who interact with familiar people and groups, and engage in well-known environments [30], are more likely to perceive social continuity, the second cognitive mechanism proposed by continuity theory. Perceived membership in important social groups is a psychological resource that reflects social continuity and is associated with various positive factors such as wellbeing, health promotion, and mental and physical health [30,34]. Additionally, perceived membership in important social groups has been found to lead to positive outcomes such as well-being over and above social interactions outside of valued social groups [35]. Although belonging to such groups may also encourage social engagement, facilitate shared social activities, and provide the individual with access to multiple social partners who could provide support, being a member of valued social groups may also provide the "space" for individuals to create positive social identities and "merge" them into their sense of self [36]. According to the social cure theory [36], it is not the activities and tasks performed in the context of the group that strengthen social identity in times of transitions, but the symbolic relationship with that group of high value. People tend to assimilate the characteristics of social groups that are important to them . These valued social groups act as anchors for identity and may have particular beneficial effects in times of life transitions, such as divorce, during which individuals lose their selfdefinition of being a wife or a husband. Therefore, in contrast to other nonimportant social groups, the important groups may allow people to experience parts of their social identities as stable and feel socially embedded, despite divorce-related changes. We hypothesize that more important group memberships will be related to feeling less socially lonely in divorced individuals, but that having valued social groups, given their nature, will also benefit married individuals. Last, as these two mechanisms promote identity stability [37], feeling a greater continuity of self and having more valued social groups may indicate a specific resilient profile of less lonely divorcees. Additionally, an excess in one of the mechanisms may compensate for a lack in the other one . For instance, one may not feel like the same person as before the divorce, but by being embedded in many social groups, one may feel less lonely. However, the concurrent absence of valued social groups and self-continuity may suggest higher loneliness of divorcees, as they cannot benefit from any of the two identity mechanisms. A lack of valued social groups and self-continuity, may, in addition, be a risk factor for married individuals in terms of loneliness. --- What Helps When? Depending on the adaptation phase, different factors may facilitate the process of coping with divorce [6,38]. Personality aspects, for instance, may be of utmost importance during early phases of adaptation, and having social resources in any adverse circumstances has been found to be beneficial for mental health [4,23]. Caspi and Moffitt [26] suggest that in new and ambiguous situations, the individual seems more likely to rely on well-established cognitive and behavioral tendencies captured by personality traits. High stress levels may emerge more often during the initial adaptation phase, as divorcees have to deal with immediate and pressing demands regarding their novel life conditions, yet they do not know how [26]. As the individual tries to cope with the new reality, automatic and well-known behavioral tendencies may emerge more frequently, rather than the person employing more cognitively demanding processes that require excessive mental energy. For instance, more neurotic divorcees may more often reject social interactions, which will not satisfy their need for relatedness with others and will increase the feelings of social loneliness. An extrovert may, instead, interact more often with others in that phase, with an immediate positive effect in return. Nevertheless, the social support associated with the bonds created in this interaction may become beneficial at a later stage, as time is needed to develop supportive relationships. Therefore, the more prototypical personality-associated behavior will not only be more frequent; its explanatory value may be accentuated under more challenging times, such as those closer to divorce, but the benefits of other factors, such as social participation, may become more pronounced later [26]. These trait-related behaviors and cognitions may be more strongly associated with well-being than other available resources, particularly when the person is still in the acute stress phase following the event, when struggling with adaptation is more likely. Findings are, however, still inconclusive regarding which specific personality traits may be accentuated in divorce [27]. Hence, aiming to close this research gap, we extend the more specific hypothesis that personality plays a role in adaptation , by assuming a time dependency of this effect: we expect that when people are closer to divorce, high levels of neuroticism and lower levels of extroversion, agreeableness, openness, and conscientiousness will be related to high levels of social loneliness, following previous research on the association of specific personality traits with loneliness [27][28][29]. After the initial stressful phase, and when individuals have dealt with the actual and emotional loss of the partner and the immediate consequences of divorce , they may be ready to advance to the next phase of adaptation. In this later phase, divorcees may work on restructuring their social lives and identities. Therefore, psychological resources such as self-continuity, representing inner-psychological continuity, and membership in valued social groups representing social continuity may become more important for adaptation. Although individual-based interactions may be beneficial regardless of the adaptation phase, as they are more direct in nature and easier to achieve, social groups may require a certain level of connectedness with other group members, which is accomplished with time. For self-continuity, the adaptation time is very relevant, as individuals may need time to determine whether the divorce was a truly disruptive event, and whether they may ever accept it as part of who they are. Hence, it is likely that self-continuity and important social group memberships, as well as their interaction, may explain more variance in social loneliness at a later stage, helping individuals return to pre-divorce levels of social loneliness that more closely resemble those of married individuals [7,8]. To our knowledge, the beneficial effects of multiple important group memberships and self-continuity on well-being outcomes have not yet been investigated in the context of divorce. --- The Present Study This study investigates the importance of psychological and social resources for adaptation to divorce, as indicated by the experience of social loneliness. As time since divorce is associated with adaptation progress [9], we compared individuals who were in an earlier postdivorce phase, coping with the new reality after divorce , individuals who were in a later post-divorce phase, when adaptation should have ad-vanced , and married individuals who had never experienced a divorce, serving as a control group. In particular, we examined the role of social resources , personality , and identity-promoting mechanisms for social loneliness, expecting differential predictive patterns across groups while controlling for subjective health and sociodemographic aspects that have been found to be associated with adaptation to divorce in prior studies. --- Methods --- Sample and Procedure The present study included a total of 1,719 individuals aged 40-92 years who were either married or had been divorced or separated within the past 5 years. The data derived from the LIVES Intimate Partner Loss Study. The sample was stratified by age, gender, and marital status. Participants were selected by the Federal Office of Statistics [for details, see 39]. Divorced and separated individuals were combined into one group in line with previous research [5]. These 850 divorced individuals were split into two subgroups according to the adaptation phase: the short-term divorced group consisted of individuals who had experienced divorce up to 2 years prior to study participation, and the long-term divorced group consisted of individuals who had experienced divorce 2-5 years prior to study participation. We compared the divorced groups with an age-matched group of married people who had never experienced a divorce. --- Measures The participants filled out a paper-and-pencil or online questionnaire, including the measures described below, and they received no compensation. --- Grouping Variable To separate married and short-term and long-term divorced individuals, the participants indicated whether they had ever experienced a separation or divorce, and if so, when this had happened. Years since the divorce were then calculated to create the following categories: 0 = married, 1 = up to 2 years after the event, and 2 = 2-5 years since the event. The 2-year cutoff was used based on the findings regarding adaptation to divorce by Booth and Amato [8] Outcome Social loneliness was measured using the corresponding items of the short De Jong Gierveld Loneliness Scale [40]. It consists of three items that are answered on a 5-point scale . A mean score was built to represent social loneliness, with higher values indicating higher loneliness . --- Independent Variables Independent variables included demographic variables, health, social resources, personality, MIGM, and perceived self-continuity. Demographic variables included respondents' age and gender, as well as income adequacy . Subjective health was assessed with one item asking for the current health status . The availability of social resources was measured with three single items: children yes/no , new partner , and someone to count on . Personality traits were measured with the short version of the Big Five Inventory . The items were evaluated on a 5-point scale , and responses were combined into a mean score for each personality trait: neuroticism, extroversion, conscientiousness, openness, and agreeableness [6]. Higher scores indicate higher levels of the personality traits. MIGM and self-continuity were measured with the Exeter Identity Transition Scales [42]. Individuals were asked to report up to six different social groups that they belonged to and to rate how important these groups were to them using a 5-point scale . An MIGM sum score was calculated, using only the groups that were rated as important or very important , with higher values indicating a higher number of valued social groups. Although, previously, Jetten and colleagues [36] had created an indicator for MIGM by multiplying average importance with the number of groups mentioned, for the present paper we used specifically those groups evaluated as important or very important to ensure that this construct reflected membership in highly valued groups only. Self-continuity was measured with the following three items: "I am the same person as I always was," "With time a lot of things have changed, but I'm still the same person," and "I'm a different person than I was in the past." These items were evaluated on a 5-point scale . The mean score of the three items was computed, with higher values indicating higher perceived self-continuity . --- Analytical Strategy We conducted between-group analyses of variance with Scheffé's post hoc tests to examine mean-level differences between married and divorced groups in social loneliness and its predictors. Regression analyses were then conducted separately for each of the three groups using demographic variables, health, social resources, personality, selfcontinuity, MIGM, and the interaction between self-continuity and MIGM as predictors, and social loneliness as the outcome. The data were examined for univariate and multivariate outliers and multicollinearity. Bootstrapping was used to test the robustness of the models. In order to test whether the standardized regression coefficients, examined in separate analyses, were significantly different across groups and, therefore, confirm our theoretical assumptions about differences and similarities between the divorced groups and the married control group, we conducted follow-up regression analyses with the whole sample to test interaction effects . For the interpretation of the results, we used standardized coefficients , F values, and R 2 values. All statistical analyses were conducted with SPSS version 23. --- Results --- Descriptive Statistics Mean levels and standard deviations are presented in Table 1. Divorced individuals felt more lonely than married individuals, regardless of the time passed since divorce = 20.95, p < 0.001). Long-term divorced individuals had the lowest MIGM score = 14.66, p < 0.001). In terms of self-continuity, the three groups were significantly different from each other, with the married having the highest and the short-term divorced the lowest levels = 108.34, p < 0.001). The correlational analyses revealed moderate associations in the expected directions. Expected correlations with loneliness were found for the short-and long-term divorced groups . However, age and children in both groups, and gender, conscientiousness, and self-continuity only in the shortterm divorced group, were not significantly associated with loneliness. Furthermore, for the married individuals, greater social loneliness was significantly negatively associated with all other variables, except for neuroticism, for which the association was positive. --- Factors Associated with Social Loneliness Multiple regression analyses were performed to investigate the predictors of social loneliness separately for the short-term divorced, long-term divorced, and married individuals . For the short-term divorced group, the model explained 34% of the individual differences in loneliness. In this group, individuals with fewer financial resources and poorer subjective health felt lonelier, as well as those who were less extroverted , less agreeable , and more neurotic . Having someone to count on in overcoming the divorce was also associated with lower social loneliness . Importantly, MIGM and self-continuity did not explain any individual differences in loneliness. For the long-term divorced, the model explained 39% of the total variance in loneliness. Men felt lonelier than women in this group , as well as those who did not have a new partner . Similar to the Table 2. Correlations of the study variables for the short-term divorced and long-term divorced individuals . For the married individuals, the amount of total variance in social loneliness explained by the regression model was substantially smaller . In this group, age was a significant predictor: younger married individuals felt lonelier . Similar to the long-term divorced group, men were also lonelier than women . Lower income adequacy and poorer health were linked to higher loneliness, similar to the divorced groups. Aside from the beneficial effect of extroversion , which was also present in both divorced groups, being less agreeable and less conscientious were associated with higher loneliness. Similar to the long-term divorced group, having less MIGM and a lower sense of self-continuity were related to higher loneliness among the married individuals. Additionally, the interaction between self-continuity and MIGM was significant in this group . The three-way interaction between the grouping variable , MIGM, and self-continuity was also tested , suggesting that the interaction was positively associated with The "new partner" and "someone to count on" variables do not apply to the married. MIGM, multiple important group memberships. + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. social loneliness only in the married group . Indeed, as presented in Figure 1, the levels of social loneliness were lower for the married than for the divorced individuals in any combination of MIGM with self-continuity , except when having more group memberships and high selfcontinuity, which was particularly beneficial. With this combination, the long-term divorced resembled the married in levels of social loneliness. However, in all three groups, the combination that best protected against social loneliness was being a member of multiple important groups and perceiving high levels of self-continuity, while the lack of those factors was associated with higher levels of loneliness in all three groups. Regarding the other two combinations, namely, having high MIGM with low self-continuity or low MIGM with high self-continuity, the levels of social loneliness differed significantly only between the married and the two divorced groups. These findings indicate that having high levels in either self-continuity or MIGM can compensate for the lack of the other in all groups, but having high levels in both is most beneficial. To further confirm these findings, the analyses were replicated with bootstrapping to check for the robustness of the model, producing an average bias estimation of < 0.007, leading to the same results as reported above. The reliability of the differential predictive patterns across groups was tested with additional regression analyses conducted for the whole sample, including the group variable and interaction effects . The results confirmed the findings reported above and our hypothesis regarding the different predictive patterns across groups . --- Discussion This study investigated the importance of psychological and social resources as predictors of social loneliness in the context of later-life divorce, with a particular focus on time-dependent differences and a special interest in identity-promoting aspects. Our study contributes to existing research on adaptation to divorce with the following findings: we confirmed differential time-dependent associations of personality , identitypromoting mechanisms , and social resources with social loneliness, comparing two post-divorce groups and married individuals. --- The Central Role of Personality in the First Post-Divorce Phase Early after divorce, personality factors had a particularly important role in explaining individual differences in social loneliness. For the recently divorced individuals, extroversion had the strongest effect among the investigated personality aspects, with higher extroversion being associated with lower loneliness. Also, being more agreeable and less neurotic was related to less social loneliness. These findings replicate Pudrovska and Carr's [5] results, showing that more extroverted individuals cope better with divorce, and partly confirm our hypothesis regarding the effect of different personality traits on social loneliness in the context of divorce . Although extroversion was also associated with lower loneliness in the longterm divorced, suggesting that being more open to social contact and enjoying social interactions are also beneficial later, agreeableness and neuroticism were not linked to loneliness in this group. These results confirm our hypothesis that during acute transitions, such as right after the divorce, the contribution of personality in overcoming difficulties is more important than other resources, supporting the accentuation model of Caspi and Moffitt [26]. Identity-Promoting Mechanisms Are Beneficial, but Only for Long-Term Divorced and Married Individuals Confirming our hypothesis that identity-promoting resources may explain interindividual differences in a later adaptation phase of divorce, we found that being a member of multiple highly valued social groups was associated with lower social loneliness in long-term divorced individuals. These findings support the assumption that belonging to such social groups may be beneficial due to not only accessing social partners but also a context that promotes a person's identity through the importance that they place on those groups. Married individuals also benefitted from multiple important groups, which could indicate their protective nature against consequences of age-related social losses, as well as potential issues occurring within their long-term marriage . These findings also seem to be in line with Weiss [20], who argued that individuals lacking an engaging social context are at risk of experiencing social loneliness. Perceiving higher self-continuity was also associated with experiencing less social loneliness in the long-term divorced, confirming theoretical assumptions about the importance of self-continuity as a crucial identity process and as a means of coping with adversity . As indicated by Atchley [30], self-continuity may enable adaptation in times of change, when previously important identities can no longer be maintained. However, our findings expand the existing knowledge about self-continuity by indicating the time frame in which it becomes particularly important in the divorce process, namely, after the first adaptation phase has passed . In our study, a positive effect also became apparent in married individuals, suggesting that self-continuity may be beneficial even without specific critical life events. In line with Chandler and Proulx [32], who argued for the importance of identity stabilization, self-continuity was associated with lower loneliness. The similar predictive patterns among the long-term divorced and married groups are in line with Amato's theory [4], providing new insights into how the process of adaptation to divorce is associated with social engagement and identity, as well as adding to the literature regarding the protective role of self-continuity, which seems beneficial regardless of having a life crisis or not. For those individuals who had more recently been divorced, the positive impact of perceived self-continuity was not confirmed . In line with Amato [4], the results indicate that time is an important factor to consider: in the first post-divorce phase, individuals may experi-ence higher psychological distress because they have to deal with the loss and their new life circumstances. During the early post-divorce phase, individuals try to evaluate the new situation, making them less likely to perceive any self-continuity, which is supported by the fact that the short-term divorcees reported the lowest level of selfcontinuity. Similarly, MIGM did not prove beneficial for the recently divorced , which partly stands in contrast to the findings by Haslam et al. [36], who showed that multiple group memberships have a positive impact on well-being in times of life transitions, such as in poststroke rehabilitation. The contrasting findings may suggest that divorce, as a life event, has its particular challenges that are not only bound to the availability of psychological and social resources but also closely related to the time frame. In the context of divorce, it seems to be the case that the "social cure" effects [36] only occur later in the adaptation process. Thus, paralleling Amato's divorce theory [4], MIGM and self-continuity may not be helpful during divorce in the short run, but they become important after some time. Regarding the interplay between self-continuity and multiple group memberships, individuals who had MIGM and high levels of self-continuity were less lonely across all groups . However, there were differential associative patterns for each group. In the married group, individuals with concurrent low self-continuity and few group memberships had a significantly higher risk of experiencing loneliness. Being prone to loneliness, even in the context of marriage, is in line with prior studies [2,4]. Married individuals may experience other transitions, such as loss of shared interests, or personal or partner health issues, which could be additional risk factors for loneliness. Previous research [36] has suggested that individuals who have MIGM in times of transitions are better able to find new social roles and adjust them to their identity, which may, according to our findings, also be important for older married adults. Additionally, self-continuity allows these changes to be perceived as additions to their life story rather than disruptions [30]. Thus, our findings make an important contribution to the understanding of the interplay of social group memberships and self-continuity, as well as how they are associated individually with adverse outcomes, such as social loneliness, in post-divorce phases in later life , as well as in the context of marriage. This is the first study to address both variables together; future studies may replicate the interplay between MIGM and self-continuity in different contexts with or without critical life events. Having someone to help deal with the divorce was important for short-and long-term divorced individualsas was having a new partner, although only for the longterm divorced group -and was associated with lower social loneliness levels, partially confirming our hypothesis , because children did not explain any variance in social loneliness. For the recently divorced in particular, having a person to help them overcome the divorce was the most powerful factor examined in this study. These findings indicate the beneficial role that the availability of social partners can play after divorce, especially in the recently divorced group. Haslam et al. [35] previously reported that only group social engagement and not individual social engagement was important as a longitudinal predictor in the context of age-associated cognitive decline. However, individual and group social engagement may both have independent beneficial effects in the context of loneliness. Thus, our results add to the literature regarding the importance of individual ties and group engagement during a life crisis such as divorce. --- Age and Social Loneliness Being older and married seems to be beneficial in terms of social loneliness, in line with previous research indicating that with advancing age, individuals generally feel as lonely as or less lonely than they did at younger ages [15,16,44]. The fact that age was not negatively related to social loneliness in any of the two divorced groups suggests that individuals, independent of their age, may experience loneliness due to the challenges they face in adapting to divorce. These findings indicate that individuals who have experienced a critical life event, such as divorce, in later life may be at risk of not experiencing the same normative reduction in social loneliness levels as their married counterparts do with advancing age. They may also experience stability in social loneliness, but at significantly higher levels than the married individuals do, as indicated by our results. Nevertheless, the relationship between age and social loneliness may be stronger when examined with prospective longitudinal data, where the pre-divorce levels of social loneliness could also be considered. --- Limitations Despite the various strengths of this study , various limitations deserve mention. One limita-tion is that we used the moment when separation or divorce was declared as the main marker of the transition. However, separation or divorce is a long-lasting process that may begin well before it is publicly declared [4]. Defining the start of a transition is always difficult, which may particularly be true for divorce. Thus, the definition of the three groups, despite the clear-cut results, may hide some heterogeneity. Another limitation is that we used cross-sectional data, meaning that changes in predictors and outcome before and after divorce could not be examined. Causal inferences were not possible, and only longitudinal data would help disentangle the dynamics of the coping process. Nevertheless, the results provide an important starting point for future prospective longitudinal research and the development of interventions. Lastly, we were interested in performing additional gender analyses for each group because the literature indicates that further research would contribute to a better understanding of the differences and similarities between men and women in how they experience changes to identity [45,46]. However, this was not possible because the subsamples of men and women were too small for the number of predictors entered in the regression, leading to unreliable findings. --- Conclusions In conclusion, the presence of social and psychological resources, including valued social groups and self-continuity, seems to be beneficial for both divorced and married individuals, in order to prevent social loneliness. Although some resources have similarly positive effects in all groups, such as health and financial means, other resources were more important under certain circumstances -such as personality right after divorce, and group memberships and self-continuity only at a later post-divorce phase. These differential effects highlight the necessity of carefully considering time frames when studying adaptation and creating divorce interventions that take life circumstances and the adaptation phase into account. Furthermore, interventions that target social loneliness in later life should focus on the beneficial effects of important social group memberships and perceived self-continuity for married and long-term divorced individuals. Strengthening the person through the development of self-continuity and group social engagement , with and without divorce experience, seems an effective way to prevent social loneliness in the second half of life. --- Statement of Ethics The authors declare that no statement of ethics is needed. --- Disclosure Statement The authors declare that there are no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. ---

Background: Critical events in the second half of life, such as divorce, pose a significant threat to well-being. Individuals undergoing divorce often experience feelings of social loneliness and may benefit differently from available resources depending on how much time has passed since the event. Personality traits have been found to be related to adaptation, with particularly strong effects immediately after the critical event. Other resources, such as identity-stabilizing mechanisms (i.e., valued social groups and self-continuity), may play a role only later in adaptation. However, little is known about the benefits of these resources and their potentially time-dependent effects on social loneliness when one is overcoming later-life divorce. Objectives: This study investigates the role of psychological (e.g., personality, selfcontinuity, multiple important group memberships) and social resources (e.g., new partner, having someone to help deal with divorce) for social loneliness in two post-divorce phases, using a married group as the reference, controlling for sociodemographic aspects and health. Methods: A representative sample of 850 divorced (aged 40-79 years) and

Findings There was no significant association with employment status . Men with hearing loss were more likely to be married, whereas women with hearing loss were less likely to be married . Although hearing loss was not associated with a current smoking status in men, women with hearing loss were more likely to be current smokers . Moreover, hearing loss was associated with psychological distress in men and women . --- Conclusion These findings suggest that hearing loss is related to social and health issues in daily life, including a lower likelihood of marriage, more frequent smoking, and poorer mental health, especially in women. These issues may reflect a gap between the actual needs of women --- Introduction Hearing loss is one of the most frequent types of disability [1]. Currently, more than 360 million people (5.3% of the worldwide population are estimated to have some degree of hearing loss [2]. Evidence suggests that hearing loss leads to a poorer quality of life, greater difficulty in social activities, and a higher level of social isolation [3]. Evidence also suggests that hearing loss is associated with lower household income, poor occupation status and lower educational attainment [4,5]. It is also known that there is an association with several health conditions, including smoking [6,7] and psychological distress [8][9][10]. The health and wellbeing of people with disabilities, including hearing loss, remains one of the top priorities in public and health policies [11]. The United Nations adopted the Convention on the Rights of Persons with Disabilities in 2006, which came into effect in 2008 [12]. The convention affirms that persons with disabilities have "the right to work on an equal basis with others" , "the right to marry on the basis of free and full consent of the intending spouses" , and "the right to the enjoyment of the highest attainable standard of health without discrimination" . It also places emphasis on the importance of statistics and data collection in formulating and implementing necessary policies . Thus, policy makers need to consider formulating and implementing evidence-based public and health policies. Japan was the 140th country worldwide to ratify the CRPD in January 2014 [13][14][15]. Although Japan may have made substantial progress in changing its social norms and policies against people with disabilities because of the disability rights movements in the 1990s and 2000s [16], it still took Japan seven years to amend its domestic laws, including the Services and Supports for Persons with Disabilities Act, to follow the convention [13,17]. This delayed ratification may have had negative impacts on the living conditions of people with disabilities, which need to be examined carefully. However, evidence on the living and health conditions of people with hearing loss in Japan is scarce. A nationwide survey of people with disabilities, conducted in 2006, reported that 0.27% of the Japanese population had some degree of hearing loss [18]; however, this number was restricted to those having a physical disability certificate . Thus, many more people with mild to moderate hearing loss were not included in the target population of the survey or covered by the social welfare program under the Law for the Welfare of People with Physical Disabilities [19]. Indeed, about 6,000,000 people are estimated to receive hearing aid, however can't receive a physical disability certificate in Japan [20] and the health needs and living arrangements of those who do not have the physical disability certificate have rarely been focused on or investigated [21][22][23]. Additionally, women with disabilities need special attention, as recognized the Convention: "women and girls with disabilities are subject to multiple discrimination" [12]. Several qualitative studies have reported that Japanese women with disabilities might be disadvantaged compared with men with disabilities [24]. However, there has been no reported quantitative study examining gender differences in the associations between hearing loss and social and health conditions. Thus, the purposes of this study were to examine the association of early-onset hearing loss with several social and health measures and to examine any gender differences in these associations, using a larger sample of residents in Japan. We aimed particularly to investigate the conditions of early-onset hearing loss among young working-age adults. Here, within this age range, people are usually faced with various important choices for their lives [25]. For example, they conduct a job search, enter the world of work, find a marriage partner, and discuss family planning with the partner. All the decisions here are usually the lifetime ones and require intensive communications, and thus usually these issues can be the source of daily hassles, stresses, and health-related behaviors [26]. Therefore, we examined the influence of earlyonset hearing loss on these broad ranges of living conditions: employment status , marital status , current smoking , and psychological distress . --- Methods --- Study population We used the Comprehensive Survey of the Living Conditions of People on Health and Welfare , which was conducted by Japan's Ministry of Health, Labour and Welfare in June 2007. The LCPHW is a population-based cross-sectional health survey known to be the best nationally representative dataset for various socioeconomic and health conditions. Among the 624,178 respondents who answered the questionnaire, we restricted the study population to those who answered a series of questions on subjective symptoms . We also restricted the study population to those aged 20 to 39 years of age. In the 2007 LCPHW, the question on hearing loss, one of the subjective symptoms, did not distinguish early-onset hearing loss from age-related hearing loss . A previous study showed that the prevalence of overall "hearing loss" increased markedly at 40 to 50 years of age, due to the increased number of cases of age-related hearing loss [27]. Because the initial manifestations of age-related hearing loss are not likely to be seen before 40 years of age [28] study participants younger than 40 years are expected to be at lowest risk of age-related hearing loss. Indeed, the 2007 LCPHW showed that the prevalence of those reporting hearing loss remained at the same level among those aged 20 to 39 years, and then increased after age 40 . Thus, we used 136,849 study participants whose ages ranged from 20 to 39 years in the 2007 LCPHW for further analyses. We obtained permission for secondary use of the 2007 LCPHW data. This study was approved by the Ethical Committee of the University of Tsukuba . --- Measures We used four outcome measures , which did not substantially overlap with the state of hearing loss itself and which covered various aspects of social and health conditions in young working-age adults. Socio-demographic factors. Regarding socio-demographic factors, age, gender, employment status, and marital status were determined. Employment status, which reliability and validity were examined [29] was measured by asking the study participants if they are currently working or not. Marital status, representing living arrangements, was measured by asking if they are currently married, never married, widowed, or separated, which reliability and validity was discussed and has been widely used [30][31][32]. We created a dichotomous indicator variable for marital status: married and unmarried . Smoking behavior. Smoking behavior was measured by asking the study participants if they do not smoke, smoke every day, smoke sometimes, or smoke but not within the past one month. The reliability and validity of self-reported smoking behavior were examined and grossly assured in multiple studies . We then dichotomized these into currently smoking and not currently smoking [35]. Mental health status. As a psychological distress measure, we used the Kessler-6 scale . K6 has been used widely around the world [36]. A Japanese version of K6 has also been validated [37,38] and been widely used [32,39]. K6 is based on answers to six-item psychological distress questions, and we calculated the sum of the reported scores, ranging from 0 to 24. We created a dichotomous indicator variable representing psychological distress, based on a previously proposed threshold of K6 scores 5 [37]. Hearing loss. Hearing status was assessed on the basis of subjective symptoms. The study participants were asked whether they had each of 41 items-subjective symptoms . One of the items is hearing loss . Self-reported hearing loss is a widely used measure in epidemiological research, and validation has been reported in multiple studies [40,41]. Previous studies indicated that self-reported hearing loss was grossly equivalent to the averaged pure-tone thresholds > 25 decibel hearing level [42]. Although standard audiometric testing is an objective technique used to examine hearing ability, a subjective measure of hearing loss is also important for understanding how study participants feel their hearing difficulty affects their daily life activities [43]. Statistical analysis. First, the associations between hearing loss status and other variables were assessed separately using χ 2 tests and Student's t-tests. First, χ 2 tests were used to determine whether hearing loss was associated with the dichotomous variables . Additionally, Student's t-tests were used for comparing the continuous variables . Furthermore, we examined the associations of hearing loss with these variables separately in men and women. Second, generalized estimating equations were used to calculate odds ratios and standard errors, which accounted for potentially correlated observations within the same households [32]. In the models, the associations of hearing loss with each of the four outcome variables were examined by controlling for age and gender. Furthermore, we examined these associations separately in men and women. We also calculated p-values for the interaction term to formally test the gender difference in the degrees of associations of hearing loss with the outcome variables. Odds ratios with 95% confidence intervals are presented. All analyses were performed using the SPSS software . married , while being widowed or separated was relatively rare. Among the study participants, 35.0% were current smokers, and 30.9% reported mild or severe psychological distress . In total, 1,012 individuals reported a symptom of hearing loss. Given that a recent national survey reported that 0.27% of the Japanese population had "certified" hearing loss [18], these results indicated that we had identified a larger proportion of individuals with mild or moderate hearing loss. The results of statistical tests on bivariate associations are also shown in Table 1. --- Results --- Sample characteristics --- Multivariate analyses Fig. 1 shows the associations between hearing loss and each of the four outcome measures in the multivariate analyses. Fig. 1A shows that hearing loss was not associated with being employed . There was no gender difference in the association . Fig. 1B shows that hearing loss was not associated with being married . However, after stratification by gender, in men, hearing loss was negatively associated with being unmarried , while in women, hearing loss showed a positive association with being unmarried . After excluding the rare cases of widowed status and separated status from the data, the regression results did not change substantially in men or women , among men , or among women . This indicates that the results for an unmarried status reflect those for a never married status in our young working-age study population. Fig. 1C shows that hearing loss was positively associated with current smoking . However, after stratification, this association was found among women , but not men . These results reflect that men had a high prevalence of being current smokers regardless of hearing condition , while women with hearing loss had a higher prevalence of being current smokers compared with women without hearing loss . Fig. 1D shows that hearing loss was associated with psychological distress . After stratification by gender, the association was significant in men and women . --- Discussion To our knowledge, this is the first reported study providing evidence of the social and health conditions among people living with hearing loss in Japan. We found that hearing loss was associated with several social and health measures. First, there was no association between hearing loss and employment status. Second, women with hearing loss were more likely to be unmarried, while men with hearing loss were more likely to be married. Third, women with hearing loss were more likely to be current smokers, but this was not observed in men. Finally, those with hearing loss were more likely to have psychological distress in both men and women. The most importantly, the significant effect modification by gender on the hearingloss-marriage association and the hearing-loss-smoking association has never reported previously; our study first reports this potential pitfall in public and health policies, as far as we know. Regarding employment status, we saw no significant association. Our findings thus differ from those of previous studies in the U.S. [44,45] in which individuals with hearing loss were more likely to be unemployed than those with no hearing loss. These results may reflect the Handicapped Person's Employment Promotion Act in Japan , which requires companies to hire a certain percentage of persons with disabilities. A previous study from Japan indicated that individuals with hearing loss were more likely to leave and change their jobs [46], and thus we need to interpret these results carefully given the simple measure of employment status used , which may mistakenly overstate the effects of the Handicapped Person's Employment Promotion Act. Indeed, an additional analysis indicates that having hearing loss is positively associated with being employed among unmarried women , while having hearing loss is not associated with being employed among married women , which interaction was not significant . This simply implies a possibility that women with hearing loss who have not married yet are motivated or forced to stay in employment, which we also need to interpret carefully. Regarding marital status, the results revealed that young working-age men with hearing loss were more likely to be married, while young working-age women with hearing loss were more likely to have never been married. Although a previous study from the U.S. indicated that individuals with hearing loss were more likely to be unmarried [44], gender differences were not assessed. Thus, there is a lack of literature suggesting potential explanations for the gender differences that we found in the present study. Here, we propose three potential explanations in the Japanese context. First, the gender difference in the prevalence of early-onset hearing loss and assortative mating patterns in those with hearing loss may explain the results. Our data showed a higher prevalence of hearing loss among women. Moreover, because those with hearing loss were reported to more frequently marry partners with hearing loss [47], the excess number of women with hearing loss may make it difficult for them to find a partner in Japan. Second, men with hearing loss may find a marriage partner more easily than women with hearing loss, given the current situation surrounding people living with disabilities in Japan. A larger number of women are engaged in social welfare activities, such as supporting people living with disabilities. For example, 80% of employees in welfare industries in Japan are women [48], and 90% of sign language interpreters in Japan are women [49,50]. Thus, men with hearing loss arguably may have a better opportunity to meet a woman who understands the situation of those living with hearing loss. This may also explain why men with hearing loss were more likely to be married in the present study. Third, the stigma towards women with hearing loss may still persist in relation to the ideology of eugenics in Japan [51]. The Japanese government approved the legalization of abortion and the Eugenic Protection Law in 1948 , which permitted induced abortion and sterilization for health reasons, including hearing loss [52]. It had been rumored that Japanese women with hearing loss were discouraged from getting married or forced to not get pregnant . There is little literature describing the historical perspectives on how women with hearing loss suffered under this legislation, and thus caution and more information are needed regarding this explanation. Regarding smoking behavior, the excess risk by hearing loss was found among women , but not among men in 2007 . In contrast, a previous study using a roughly comparable population in 1997 shows that the excess risk by hearing loss on smoking behavior was not found among women or among men [53]. Since this previous study used a small sample size [53], the comparison between the two studies requires a careful operation. Nevertheless, a potential rise of the smoking prevalence among individuals with hearing loss during the decade may be an unfound public health concern in Japan. This can relate to the psychosocial pathways among them , and relate to the other possibility that smoking cessation and prevention programs are difficult to reach those with hearing loss [7]. Psychological distress was found to be associated with hearing loss, consistent with previous studies [8][9][10]54]. Previous studies proposed several potential mechanisms to explain the progression from hearing loss to mental illness [8,10]: 1) access to effective communication, 2) socioeconomic environment in relation to stigma and discrimination, and 3) less access to mental health services. First, individuals with hearing loss often do not have access to medical care with effective communication [55,56]. Second, it has been reported that working adults with hearing loss frequently face negative emotional situations , leading to a negative physical and mental health status, including social isolation, depression, irritability, and feelings of inferiority [57,58]. Third, individuals with hearing loss often do not have sufficient knowledge about mental illnesses, so they may come to distrust the mental health system, resulting in less frequent access to mental healthcare [59]. Moreover and more generally, individuals with hearing loss can face various kinds of issues and hassles in their daily life [4,8]. Although the gender interaction was not statistically detected in psychological distress, men with hearing loss may have an additional source of psychological distress . Although we used a large sample from a nationally representative survey, the present study has several limitations. First, our measurement of hearing loss was self-reported, and thus, potential bias in the measurements could generate false-positive associations between hearing loss and the outcome variables. Although the implications from previous reports are not consistent, they grossly suggest that self-reported hearing loss may tend to underestimate the state of hearing loss [60,61]. Second, the 2007 LCPHW is a cross-sectional study, and therefore we need to be careful of any causal interpretation of the reported associations. However, the argument induced by the "reverse causality" is not persuasive clinically: being unmarried, unemployed, or a smoker causes hearing loss. In the 2007 LCPHW, the prevalence of hearing loss from 20 to 39 years of age was almost at a consistent level , and early-onset hearing loss is unlikely to be cured later, suggesting that hearing loss occurred before the four social and health measures. Indeed, lifestyle stress related to these social conditions may induce hearing loss , but this would not likely explain the reported gender differences. Third, several other related measures such as self-rated health and the frequency of physician visits were available in the 2007 LCPHW. However, we didn't analyze these measures because hearing loss could be an aspect of self-rated health, and hearing loss might require scheduled hospital visits for follow-up at an otolaryngology department. Moreover, the physical disability certificate under services and supports by the Services and Supports for Persons with Disabilities Act allows individuals with severe/profound hearing loss to gain easier access to medical care . Therefore, the impact of early-onset hearing loss on other health conditions and health care use can be the direction of the future research, in which the data to distinguish hearing-loss-related health loss or physician visits from non-hearing-loss-related health loss or physician visits. Fourth, potential confounding factors, which were not measured in the 2007 LCPHW, may at least partially explain the association of early-onset hearing loss for each of the four outcome variables. Such factors include poor maternal diet, maternal alcohol use during pregnancy, premature birth, and noise exposure in childhood, all of which were reported to be associated with early-onset hearing loss in prior literature [62][63][64]. Fifth and lastly, since status of the physically disabled person's certificate was not available in the 2007 LCPHW, we could not directly distinguish individuals with severer hearing loss from those with mild or moderate hearing loss, or examine the effect of the formal program. Further more precise original survey will be needed to cover these limitations. In summary, this study revealed that social and health burdens among individuals living with hearing loss appeared particularly in women. Young working-age women with earlyonset hearing loss in Japan are more likely to be unmarried , smokers, and psychologically distressed than women without early-onset hearing loss. This may be because they have a larger number of unmet needs in their daily lives, and because those with mild to moderate hearing loss have been overlooked in public and health policies and even in public health research. Although the implications of this study cannot be easily stated, the general aims of the CRPD may not be achieved only by legislation. Future initiatives can include recruitment of professional female counseling staffs in the local government offices, training of women with hearing loss themselves as counselors , and construction of strong networks of young working-age women with/without hearing loss to get together and share their experiences . This study illustrates that many aspects of people with disabilities-job, marriage, smoking, and mental healthrequire special attention and multidimensional approach by policy makers and all involved. Such attention and approach will accelerate the social change of Japan, and also contribute to realizing the aims of CRPD. --- Data are available through the Ministry of Public Management, Home Affairs, Post, and Telecommunications, interested readers can contact via http://www.mhlw.go.jp/toukei/ sonota/chousahyo.html. ---

To examine the consequences of early-onset hearing loss on several social and health measures and any related gender differences in Japanese populations.Data from a 2007 nationally representative cross-sectional household survey of 136,849 men and women aged 20 to 39 years were obtained (prevalence of self-reported hearing loss: 0.74%). We focused particularly on four social and health measures: employment status (employed/unemployed), marital status (married/unmarried), smoking behavior (yes/ no), and psychological distress (K6 instrument: 5 or not). We examined the association of hearing loss for each measure using generalized estimating equations to account for correlated individuals within households.

Introduction Numerous fundamental problems of the economy and society -such as educational inequalities, unemployment, migration or deprived areas -require long-term solutions that call for innovative co-operation between social actors, citizens' direct and voluntary participation in decision-making processes and the implementation of social innovation efforts. Social innovation initiatives are solutions for a sustainable future, and may be identified as one of the possible means to reduce regional disparities, increase territorial competitiveness and close the gap. In measuring the impact of good practices in education based on social innovation, particular attention should be paid to issues of sustainability and the relationship between the social and institutional dimensions of sustainability. Sustainability is not merely an environmental concept: it is also a key element in defining solutions that focus on social sustainability and reduce educational inequalities. This study pays particular attention to the relationship between social innovation and social learning, as well as to measuring the impact of social innovation initiatives and assessing their sustainability. Accepting that as a new means of addressing the challenges facing each community social innovation is best understood in innovative partnerships between academia, business and civil society, the study focuses on two good practices that pay special attention to knowledge management and learning in disadvantaged areas and may be defined as part of a specific value-driven training model. Our study included 38 good practices that enabled the creation of our knowledge base. The collection is a database that also contains structured good practices. The good practices included in the knowledge base were classified according to a certain set of criteria using a rules-based system, and two practices were identified, one of which is already being implemented effectively in the region under study, and another which may be considered as a new methodological adaptation and is in the process of being implemented . --- Social innovation and social sustainability Technological and economic innovations cannot respond to all societal challenges. Natural and material resources are becoming increasingly scarce, making it necessary to use the goods available for investment as reasonably as possible, and to achieve the greatest possible social and economic efficiency. Sustainability is not merely about the environment; a sustainability-based approach needs to take into account economic, environmental and social issues simultaneously. The social challenges that require long-term solutions call for new forms of social collaborations. According to the approach to social innovation [1], social innovation is the generation and implementation of new ideas, in the course of which people organise their social interactions to achieve a common goal. A further trend in social innovation theories [2] focuses on solving social problems, emphasising the significance of stakeholder collaboration. In a different approach, the concept emphasises the value created by social innovation [3], and highlights the satisfaction of social demand as the main goal. Uncertainty, crises, unforeseen technological changes and globalisation make the future unpredictable [4]. The process of social innovation makes societies more sustainable and cohesive through inclusive solutions, collaborations and proactive, grassroots initiatives [5]. However, it does not only mean a process based on grassroots efforts and citizen involvement, since social innovations, which can also be seen in new approaches to co-operation and structural transformation in society, often come from above, as a result of macro-level measures. The concept of social innovation focuses on meeting the needs of the community, and through this process, the quality of life and well-being are also improved [2,3,6,7,8]. When studying initiatives related to social innovation, the social benefits of innovative ideas that can be interpreted at the local, community level in problem solving, and the role of proactive community participation in raising living standards are emphatic. Social innovation is the process of change that responds to social challenges through a creative and reconsidered combination of the available tools and solutions. The analysis of the sustainability of this social change is a key focus. This study examines good practices that focus on disadvantaged groups and that offer opportunities and pathways for students in primary, secondary and tertiary education. --- Analysis One of the Hungarian units of our action research is the Northern Hungary region. The rationale for its selection and the way it strengthens justification for social innovation are described in the next section. --- 3.1 --- Factors affecting and reflecting the social innovation capacity in the Northern Hungary region The Northern Hungary region covers 610 municipalities and is characterised by a structure of small villages. Ten of the 29 districts of Borsod-Abaúj-Zemplén, Heves and Nógrád counties are to be developed in the framework of a complex programme ). A total of 269 municipalities in the region are social, economic and infrastructural beneficiaries, while 221 municipalities are affected by significant unemployment . The Northern Hungary region is located in the north-eastern part of Hungary, on the periphery of the country. The region is a mix of marginalised areas and fast-growing regions. The needs and challenges of peripheral regions may be grouped according to three factors : Source: Authors' own elaboration based on [9]. Due to their peripheral location, disadvantaged localities face numerous problems. The challenges of the region constitute a double stigma for municipalities: these include, on the one hand, a negative external perception, and on the other hand, internal stigmatisation processes of the local residents, which require the accurate identification of local needs and the involvement of the population in decisions. The aim of this qualitative study is to identify good practices that play a key role in addressing the region's educational disadvantages, as their adaptation as good practices may support the process of closing the gap. In describing the two good practices, the study also identifies the main sets of criteria that, in addition to providing a structured record of each case study, help to make the good practices comparable and analyse their impact. The demographic characteristics of the region are influenced by its ethnic specificities, i.e. the high ratio of Roma. The distribution of the Roma population by region in Hungary is shown in Figure 1 . In addition to the high ratio of the Roma population, a critical factor is the accelerating outflow of the non-Roma population. Emphatic factors related to ethnic specificities, such as low levels of education, low employment rate, low incomes and out-migration are also observed in the Northern Hungary region. The educational attainment of the region's population is typically lower than in more developed areas. According to the 2011 Census , 20% of the region's population aged 7 and over had completed grade 8 or less of primary school . The ratio of people with upper secondary education is 19%, close to the national average . In the region the ratio of secondary school graduates also remains below the national average . The ratio of people with tertiary education is 10.2% in the region, just over 2/3 of the national average . Different figures may be observed for the three counties: 20% of the population in Borsod-Abaúj-Zemplén County, 18.5% in Heves County and 21.7% in Nógrád County have completed at most grade 8 of primary school. The ratio of people without a general certificate of secondary education is 18.4% in Borsod-Abaúj-Zemplén County, 19.3% in Heves County and 19.4% in Nógrád County. The ratio of those with a secondary school leaving certificate is 23.9% in Borsod-Abaúj-Zemplén County, 24.4% in Heves County and slightly under 23% in Nógrád County. There is a significant difference in the ratio of people with completed tertiary education: 10.5% in Borsod-Abaúj-Zemplén County, 11.2% in Heves County and merely 8.8% in Nógrád County. --- Paper-Social Another factor contributing to the lower education levels of the region's population is the outward migration of skilled labour from the region . Out-migration is most common among the middle-aged population in deprived municipalities. The Northern Hungary region lost nearly 8.5% of its population between 2000 and 2011, with the loss of jobs caused by the global economic crisis of 2008-2009 playing a role. There was a minor increase in 2012, but since then the region's population has steadily declined . The less favourable labour market and income situation of the population compared to the rest of the country plays an important role in out-migration. In 2019, the unemployment rate was 3.5% nationally and 4.5% in the Northern Hungary region . The rate is 4.5% in Borsod-Abaúj-Zemplén County, significantly lower in Heves County , while in Nógrád County the rate is 55% higher than the regional rate . --- 3.2 Good practices in social innovation in the Northern Hungary region A basic criterion for the social innovation process is that the endeavour must be a novel initiative. This does not necessarily mean a completely new solution, but a novel combination of schemes that have worked well in the past and meet emerging societal needs. Social innovation is not linear, but a spiral process with feedbacks. Following successful implementation, new needs emerge, which can be met by the socially innovative community through the above process. Social innovation is therefore a dynamic process that leads to social learning through certain feedback loops and constant risk assessment. These efforts fundamentally respond to local needs, addressing the challenges facing the population through creative problem-solving and innovative partnerships, and leading to higher living standards and well-being. --- 3.3 Participatory study to explore social innovation efforts As an initial step in the research, we sought to articulate the research problem in collaboration with local communities, and local resources and problems were mapped. After the analysis of the Northern Hungary region, it can be established that regional differences in development are closely related to the size of the social innovation potential. During the identification of good practices, the main challenges of each municipality -out-migration, unemployment and educational inequalities -provided a kind of categorisation opportunity to present cases. In the course of the research, we kept a research diary from 2016 , recorded the interviews , and took more than 2000 photographs. We used specific, literal quotations and documented the changes created during each action . During this documentation, we made and analysed video recordings, audio recordings, and materials documenting various kinds of group work. In the framework of semi-structured and unstructured interviews, during the research the members and leaders of various groups were interviewed. In addition to the interviews, we also analysed each endeavour as observers of the forums and analysts of the related documents. At the end of each interview, we asked the respondents to recommend another respondent and/or an event for each stakeholder where we could participate as observers to follow the evolution of social aspirations. In the first part of the interview, we asked our interlocutors to formulate the concept of social innovation in their own words. After recording the answers, we continued the investigation with the following groups of questions: mapping general information on the municipalities and/or organisations, presentation of the implemented social innovations, outlines of the planned social innovation efforts, describing successes and obstacles, and drawing lessons, assessment of regional co-operation . To answer the research questions, we analysed our notes taken at the events, the interview summaries, written summaries of the forums, our personal notes, and audio and video materials from each forum. We paid special attention to repetitive elements, individual opinions and possible contradictions. Based on the interviews, we could identify the relations and mergers within the region, and the factors that make co-operation difficult or impossible. We identified the factors that can be identified as key challenges for the region and identified areas where endeavours can be identified for the region based on the contribution of social innovation to well-being. The knowledge base defined according to the above criteria is described in the Appendix 1 . --- 3.4 Selection of good practices using an expert-system In the course of the research, special attention was paid to the role of social innovation processes in the convergence processes of peripheral areas. In addition, these studies have highlighted the need to identify factors that focus on knowledge management and learning in the selection process [10,11,12]. In a full description of this process, a specific model interpreting a value-driven training solution is presented . An economic and social database provided the framework for the value-driven training model, which, together with the qualitative data revealed by the participatory action research, enabled the development of a knowledge base of good practices. The repository of good practices is a database of structured good practices and their evaluation according to specific criteria. The database should be continuously updated with new good practices and statistics. Since we considered the establishment and operation of social innovation as a change management process, we found that all its features were identifiable. In this approach, change management aims at recognising, acknowledging, planning, implementing and sustaining the changes needed for environmental adaptation and municipal/organisational renewal. In line with the process of change, the training model pays attention to the types, structure and target groups of training, to the definition of measurement levels for social innovation initiatives and to attitude formation . In our value-driven training model, the focus is on knowledge transfer in collaboration with disadvantaged students in higher, secondary and primary education. We undertake to transfer the knowledge required for the successful implementation of social innovations, e.g. about entrepreneurship and the institutional system, to selected university students, who, with our mentoring, then pass on the acquired knowledge to secondary-school students coming from their municipalities and being in similar social statuses, who then do the same with primary school students involved in the experimental project on a voluntary basis. Thus, through a multi-level training structure, learning through engaged participation, and with the involvement of innovators and sponsors, economic development innovations and social programmes are implemented. The various phases of this process are made novel by the special attention paid to the activities of the value chain of knowledge management, which may contribute to the formation of a new community in the locality and to the retention of young people in the region. In order to shape up and operate community building and well-being initiatives as social innovation solutions, the programme participants need additional training, coaching, counselling and sensitisation, which is also a relevant element of the initiative. Social innovation takes the form of innovative partnerships involving local residents, local authorities, businesses and social enterprises, research institutes, universities and civil society. --- Comparative analysis of selected good practices Within the framework of our research, we have identified two practices that, based on a conscious selection, provide the greatest added value among initiatives aimed at eliminating educational inequalities . --- General information --- Contact person Hungarian Maltese Charity Service, Symphonia Foundation --- University of Miskolc --- Purpose The main aim of the programme is to reduce early school leaving and early school drop-out, and to help children succeed in school. The aim of the programme is to enable disadvantaged children who are talented in individual sports to achieve better integration into society, to make good progress in their studies and to reduce environmental disadvantage. Target group disadvantaged students in primary and secondary education and the wider community disadvantaged students in primary, secondary and tertiary education, as well as the wider community --- Target region --- Northern Hungary region --- Human resources The programme employs 16 social workers and 25 music teachers. The professional management team consists of 5 people and the project management team of 3 people. A total of 49 people are working on the programme, but they also aim to involve as many volunteers as possible. Physical education teachers, coaches, school principals, teachers involved in improving learning outcomes. --- Financing funds from foundations, voluntary donations, tender applications local and central funding, tender applications, voluntary donations Required infrastructure internet, technical equipment, musical instruments, real estate gyms in schools that are in operation, possibly closed but not vacated; swimming pools --- Description of the practice Brief description A social programme in which music is a means of creating opportunities and a way out for disadvantaged students. Sport as a socialisation factor has a positive impact on talent management and better integration into society. --- Presentation of implementation The idea of the Symphony Programme was invented in Venezuela in 1975 by pianist, conductor and economist José Antonio Abreu. The concept was brought to Hungary by the Symphonia Foundation. The children living in the village were lured to the school with violin music, and the programme quickly became very popular with them. The Symphony Programme brings together pedagogy, music education, social and community development work. It is a systemic approach, based on the premise that children's school careers cannot be understood in isolation, but only within the context of the family, local social, child-welfare and healthcare institutions that surround them. Developing a scholarship scheme to support young people who are good or outstanding in sport, but also have good academic results, at all levels of education. A scholarship for good performance in both academics and sport will provide the student with the support he/she needs to enable him/her to continue his/her studies and further education. The personality traits and characteristics developed through sport will help candidates to perform well in other areas of life. Results, outcomes and ideas about the future An important result identified was a significant change in the children's behaviour, perseverance and concentration. The Symphony Programme offers an innovative tool for children living with insecurity, severe deprivation and difficult family circumstances, with key elements of a sense of achievement, selfconfidence and a sense of belonging to a community. Group music sessions are the backbone of the Symphony Programme, developing children's intellectual, emotional and social skills through an experiential approach. They also support the participating children and their families with social and community development activities. Currently 700 children participate in the programme. To create a life situation for the students involved in the programme that will enable them to integrate successfully into society, in sport and in their profession of study, and to become active players in the labour market. The scholarship is a tool that helps the beneficiary to develop and maintain a higher quality of life. An important outcome of mentoring is a sense of achievement, self-confidence and a stronger sense of belonging to the community. --- Problems encountered and lessons learned Lack of resources and volunteerism, individual case management for families in need, the channelling of various problems into the local care system. In addition to providing resources, running the system and network that has been set up, individual case management for those concerned, and strengthening an inclusive attitude. iJET -Vol. 17, No. 17 Why can it become good practice? Territorial approach For disadvantaged children in the region, neither the family nor the school alone can create the appropriate conditions for success in school and further education, and this is where the programme can help. --- Grassroots approach The practice clearly focuses on local needs, and involves the local population in implementation. --- Partnership approach The exercise has a clear focus on local social challenges, but it is primarily implemented through the efforts of local civil society organisations, educators and support professionals. Parents' associations to support programmes, sports clubs to create conditions for participation, foundations to support disadvantaged children, local governments, local entrepreneurs, minority organisations, churches. --- Innovation The resulting youth orchestras have become models of the social functioning for children, where everyone has an equally important role to play, and where a common goal can be achieved through co-operation and solidarity. By implementing the adaptation of the US draft system, it offers disadvantaged young people a way out and supports their integration into society on the basis of performance. --- Integrated approach As a result of the programme, the incidence of juvenile deviance resulting from social tensions and a lack of purpose has decreased significantly. --- Publicity/networking The programme places a strong emphasis on networking and knowledge sharing, i.e. the continuous exchange of experience between programme sites. The project has developed an educational programme and an accredited teacher training programme, which will enable the Symphony Programme to be extended and introduced in other schools. The programme also provides an opportunity to expand beyond the school walls and enrich the activities of those working with children and young people in community centres and other community spaces. The implementation of the Sport Socialisation Scholarship Programme in a network requires the development and continuous provision of network competences, where, in addition to funding, the availability of human resources is, in our opinion, the biggest risk. The promotion of the programme is an essential condition for its success. Local initiatives and results should be reported as widely as possible, involving as wide an audience as possible. --- Sustainability Hungary has a long tradition of instrumental music education and choral singing. However, the Symphony Programme does not compete with the traditional music school network, as it uses a different methodology and targets children who are typically excluded from traditional music education. Music teachers typically visit schools in groups, in other words, several music teachers give instrumental lessons at the same time. The following aspects of sustainability are relevant: social sustainability can include the inclusion of a disadvantaged group, -it contributes to economic development by bringing out those concerned from the labour market. --- Adaptability This good practice can be adapted to other regions and target groups, taking into account local conditions. In schools, the programme is embedded in the all-day school system, with morning and afternoon music sessions forming the backbone. They also organise a number of school and municipal performances, as well as inter-school, regional and national meetings and concerts. Their main goal is to create orchestras at regional and subsequently national level, and to provide their students with as many opportunities as possible to perform and experience success. This good practice can be adapted to other municipalities, regions and target groups, taking into account local conditions. The scholarship scheme can be integrated into primary, secondary and tertiary education through a specific mentoring activity, which ensures both its specialisation and its scalability. Source: Authors' own elaboration based on the participatory action research. The good practices will be presented in a structured way, outlining the background to the initiative, its main objectives, local conditions and circumstances, the implementation process, the results and impacts to date, and the adaptability and sustainability assessment. In the selection process, we made a conscious effort to present a pilot programme , in addition to a successful example that has been operating well for many years , and to actively contribute to the development of the framework conditions for this programme. While the Symphony Programme focuses primarily on students in primary and secondary education, Sport as a Socialisation Programme primarily targets students wishing to enter higher education and then continue their education. --- Conclusion Innovation is always present, it is created by constraints, but whether there will be a long-term solution to a problem depends on the institutional framework. Initiating social innovation is based on expertise, willingness and trust, which are key elements of co-operation. Adapting good practices supports the generation of further innovation. The question is, however, whether the practice in other municipalities can be imitated and embedded in the life of a given community. Imitation in itself is only a transfer of good practice, but transferring the process, encouraging co-operation, supporting collaboration between organisations and municipalities, and setting the framework conditions for networking will lead to effective and sustainable solutions to key problems in disadvantaged areas. The good practices examined have responded to the local needs, and the participants have co-operated and consulted with institutions, civil society organisations, entrepreneurs and local residents in the municipalities concerned to implement the programme. In an innovative way, they have initiated genuine dialogue and innovative solutions, while encouraging the public to take action. This in itself is a social innovation. The impact of good practices in social innovation can be measured by the analysis of the social, economic and environmental impacts that result from the initiatives. The changes that are taking place are creating new needs, which also support the generation of further social innovation practices. Considering social innovation as a process of change management also allows for the analysis of good practices that aim to identify, recognise, plan, implement and sustain the changes required for environmental adaptation and regional/organisational renewal. Learning and understanding good practices in social innovation, embracing change and committing to the initiative will lead to change in the behaviour and this will measure the impact of social innovation and the added value of the overall transformation. In our study, we set out to present good practices selected on the basis of this impact assessment, which have the greatest added value in the field of community development and well-being. Based on the research, it can be concluded that the value-driven social innovations observed in the Northern Hungary region and integrating young people of different age groups and disadvantaged groups provide useful methodological solutions to community problems. Social aspirations have a powerful role to play in reducing the outflow of skilled young people, supporting disadvantaged groups and meeting a higher demand for public culture. --- Appendix Dr. Mariann Veresné Somosi is the Dean of the Faculty of Economics and Director of the Institute of Management Sciences at the University of Miskolc, Hungary. She is a mechanical engineer and her research interests are focused on project management, social innovation, organisational change, competence management, performance management. Author and co-author of nearly 40 journal articles and more books on social innovation, performance management, competence management. She has been an active researcher for more than 30 years, with a special focus on economic development and networking. Her research group is working on the development of neuro-fuzzy solutions for generating social innovation. In the faculty's doctoral school, she aims to sensitise students to social innovation by announcing topics and the social innovation subject. She is the active author and editor of the White Paper on Social Innovation. Dr. Krisztina Varga is an assistant professor at the Institute of Management Sciences at the Faculty of Economics of the University of Miskolc. Her research interests are the measurement challenges of social innovation, competitiveness, social good practices. Co-author of 15 journal articles and one book in English on social innovation. She has been involved for 5 years in participatory action research in the field of inclusion programmes in disadvantaged municipalities. Within social innovation, the social innovation potential and methodological developments related to educational equity are more prominent, and her co-authoring model was recognised at a conference in Orlando. In addition, the applicability of the expert system in generating social innovation is also part of her research. She is an active participant in conferences and workshops. Author and editor of the White Paper on Social Innovation.

Social innovation initiatives are solutions for a sustainable future that can help reduce regional disparities, increase territorial competitiveness and promote closing the gap. Reducing and then eliminating inequalities in education requires long-term solutions that call for new forms of co-operation between social actors. Our participatory action research focused on disadvantaged groups in the Northern Hungary region. The methodological basis of our study is a value-driven training model, built on a process approach capture of and support to social innovation, as one of its pillars, through elements such as change, measurement level and training forms. The other pillar focuses on value drive and knowledge management in its innovative combination with focus on participation, integrating primary, secondary and tertiary training solutions, thus reinforcing the potential interdependencies. In this study, a structured presentation is provided of two good practices that offer the greatest added value among the initiatives we have collected and developed in an effort to eliminate educational inequalities. Our research methodology also integrates elements suitable for measuring the impact of good practices under study in social innovation. As a result of the research, it can be conceived that the value-driven social innovations observed in the Northern Hungary region and integrating young people of different ages and disadvantaged groups contribute to the settlement of community problems by useful methodological solutions.

and non-severe injuries [1.61,]. Relative to 60-69 year olds non-serial users, the oldest cohort demonstrated lower odds for psychiatric/alcohol and drug-related diagnoses [OR=0.63, CI:0.42, 0.95]. Our findings highlight the need to design age-group specific interventions to reduce the frequent ED utilization among the elderly in Maryland. --- SESSION 7220 --- MINORITY AND DIVERSE POPULATIONS Abstract citation ID: igad104.2337 --- CHALLENGES AND OPPORTUNITIES IN RECRUITING DIVERSE OLDER ADULTS WHO ARE FRAIL FROM THE MAPS-B STUDY Suleman Tariq 1 , Alexa Kouroukis 1 , Courtney Kennedy 1 , Jonathan Adachi 1 , Carolyn Leckie 2 , Alexandra Papaioannou 1 , and Isabel Rodrigues 1 , 1. McMaster University,Hamilton,Ontario,Canada,2. St. Joseph's Healthcare Hamilton,Hamilton,Ontario,Canada Including diverse individuals at the research and participant levels are essential to improve the effectiveness of real-world interventions; however, there are challenges when including such individuals. Our study purpose was to report the challenges of recruiting diverse older adults for the Mapping Sedentary Behaviour study. Our methods were guided by Step 1 in the Knowledge-to-Action-Ethics Framework. We assembled a diverse team of eleven researchers, clinicians, and patient partners. To recruit a broad group of participants, we partnered with City Housing Hamilton, which provides subsidized housing for older adults. We met with the organization's partnership development advisor who organized two recruitment orientations; 80 potential participants and returning attendees were present for both sessions. The organization provided coffee and donuts. Most attendees were from visible minorities and had visible disabilities . To build rapport, we met with attendees in groups of 5 to 6 to introduce the research team and explain the study. We recruited 13 participants . Before their scheduled study visit, twelve participants dropped out citing medical mistrust . The last participant dropped out after the initial study visit due to their family's skepticism in research. Additionally, some individuals may have enrolled for financial incentives as they were interested in receiving immediate monetary compensation. We faced challenges when recruiting frail older adults from diverse backgrounds. Future studies should focus on developing methods to target medical mistrust with older adults and their families.

the years. Therefore, nursing homes have invested significant resources into maintaining or improving the ratings awarded to them by CMS. This study utilizes statistical models to analyze the past rating histories to predict future evolution of ratings of nursing homes. Data on all U.S. nursing homes for the years 2017 -2021 was derived from CMS Nursing Home Compare. Primary variable represents the overall five-star quality rating of nursing homes. A Markov chain model and a Mover Stayer model are used to predict changes in ratings of nursing homes. The one-year rating transition probabilities are estimated using maximum likelihood estimation. The key findings indicate the likelihoods of future upgrades and downgrades in nursing homes' ratings. Nursing home administrators are under constant pressure to strategically allocate resources for maintaining or improving the five-star ratings awarded to them by CMS. The findings of this study could help nursing home administrators in forecasting future change in their rating and making better resource allocation decisions such as those related to preventing a rating downgrade. The study findings could also be used by policymakers to focus on nursing homes facing downgrades and assisting them in improving their quality of service based on the potential change in their rating. The same results could also be utilized to assess the stability of ratings given to nursing homes.

Introduction --- M ortality patterns vary among ethnic minority and migrant groups compared with their host populations. [1][2][3][4] Various studies had reported low mortality 2,3,5 , while other studies, in contrast, had reported high mortality 3,6,7 among different migrant groups. The explanations for high or low mortality among migrants compared with the host population vary. Differences in mortality may be driven by access to care, migration related lifestyle changes and pre-migration among migrant groups. The development of mortality patterns among migrants in a host country overtime is still uncertain, particularly among different generations of migrants. There is a need to investigate mortality patterns across time between different generations of migrants to get more insight into the health risks of younger migrant groups in host countries. Analyses of migrant mortality patterns by generation are increasingly gaining interest as this may give insight into the various underlying influencing factors, such as cultural-, 8,9 environmental-10 and lifestyle factors. 8,11 It has been suggested that mortality differences between migrant groups and the host populations might be more profound in the first generation than in the second generation migrants due to acculturation towards the host population. Various studies have shown to some extent that migrant mortality patterns move towards the host populations levels by the second generation. 9,11,12 The development of migrant mortality patterns in the third generation in host countries remains unclear mainly due to lack of data. Furthermore, the younger generation ethnic minorities with mixed ethnic background in host countries are growing. Multiple studies have investigated all-cause mortality among persons with mixed ethnic background. [13][14][15] All these studies show largely a lower mortality among persons with a mixed ethnic background compared with host populations. It remains uncertain by which generation of migrants with a mixed ethnic background mortality rates move towards the rates of the host local populations. Three generations of the Moluccan population in the Netherlands provide an unique opportunity to study this. A group of $3000 Moluccan soldiers with their families were transferred to the Netherlands in 1951. 16 These Moluccan soldiers served in the Royal Dutch East-Indies Army during the Second World War. After arrival, the Moluccans were placed in several barrack camps throughout the Netherlands. These barrack camps were not suitable for long term housing. A decade after arrival, the Moluccans were housed in Moluccan residential areas, spread over approximately 60 so-called 'Moluccan districts'. Nearly all these stem from the Moluccans who arrived in 1951; currently one third is third generation. 16 In general compared with the Dutch, Moluccans are lower educated and have lower-ranking occupations. 16 The aim of this study was therefore to examine all-cause mortality among three generations of Moluccans compared with the Dutch. Additionally, we examined all-cause mortality among Moluccans with a mixed and non-mixed ethnic background. Our prior hypothesis was that the mortality pattern of Moluccans would move towards Dutch national level by the third generation. In addition, we hypothesized that the mortality pattern among Moluccans with a mixed ethnic background would move more rapidly towards Dutch national level than Moluccans with a non-mixed ethnic background, assumingly due to a strong influence of the Dutch culture among Moluccans with a mixed ethnic background. --- Methods We used data from the death and municipality registry for the years 2000 through 2013, including all registered Dutch inhabitants. The data included all persons who lived sometime within the study period . All registered deaths in the period were included, regardless of the country of death. An additional registry was used, which holds pseudonymised information on the person and the respectively parents. Each person in this database was given a 'personal identification code' by which both registries could be linked. According to the Ethics Medical Committee of the Academic Medical Centre , no approval was needed considering the Dutch law on medical research . --- Selection and generation A list of Moluccan surnames was used to select the Moluccan population. This list was derived from Museum Maluku Netherlands who created the list by combining the passenger lists of the various boats on which the Moluccans came to the Netherlands in 1951. Persons of Moluccan origin were selected based on their full surname. Notation differences of family names were included in the selection of Moluccan persons. No cases are known of Moluccans changing their family name after arriving to the Netherlands. In this study, the first generation was defined as the group of persons, the former soldiers and their family members, who came to the Netherlands in 1951. We made a distinction between the first generation who came to the Netherlands as an adult and as a child . The offspring of the first generation was considered the second generation . We considered the third generation those who were born after 1 January 1970. --- Variables The cause of death was identified and divided into categories according to the ICD-10. From the population registry, we derived information on sex, age-at-baseline, four-digit postal code and country of birth of the subject and both parents. The area-level socio-economic status was based upon the neighbourhood income estimations of the Netherlands Institute for Social Research via the four-digit postal code to control for uneven distribution of the Moluccans across the Dutch territory that resulted from historical events . These estimations were divided into 10 deciles, whereas decile one indicated a low SES area and decile ten a high SES area. By linking the death-and relatives registry through the personal identification codes, we could identify parents and assess whether they were of Moluccan origin. A person with two parents of Moluccan origin was considered 'non-mixed ethnic background' and persons with one parent of Moluccan origin was considered 'mixed-ethnic background'. For both Moluccan and Dutch, the total amount of person time and deaths was calculated. For each study population, age-standardized mortality rate per 100 000 persons was calculated based upon the age-distribution of the Moluccan population. All-cause and cause-specific mortality rate ratios with corresponding 95% CIs were calculated by Poisson regression, controlling for gender, age-at-baseline and district SES. The reference group was persons who were registered with a Dutch nationality. The statistical analyses were performed using SPSS version 22.0. --- Results Table 1 shows the background characteristics, person years at risk , number of deaths, the ASMR and all-cause MRR for each population investigated by sex. The majority of the Moluccan men and women lived in low SES areas unlike their Dutch counterparts. The all-cause mortality of Moluccan men and women was significant higher compared with the Dutch men and women. Table 2 shows the cause-specific MRRs for Moluccans compared with the Dutch population by gender. For almost all disease categories, high mortality rates were found among Moluccans compared with the Dutch population. Among Moluccans, high mortality rates were found for infectious diseases , cardiovascular diseases especially for cerebrovascular accident , ill-defined conditions and diabetes . For neoplasms and external causes, no differences were found among the Moluccans compared with the Dutch population. The mortality patterns of Moluccan men and women were very similar. A significant lower mortality was found for suicide among Moluccan women , whereas no difference was found among Moluccan men compared with the Dutch. Table 3 shows the all-cause MRRs for Moluccans compared with the Dutch population by generation and gender. In each generation, all-cause mortality was significantly higher for the total group of Moluccans compared with the Dutch population. This excess was larger for the first generation adult than for the other generations . Moluccan men and women showed similar mortality patterns of generation-specific mortality. Table 4 shows the all-cause MRRs for Moluccans with a nonmixed and mixed ethnic background compared with the Dutch by generation. For the non-mixed Moluccans, higher mortality was observed in all generations, with the largest difference among the first generation adult , compared with the Dutch. A similar pattern was observed for non-mixed Moluccans, although the differences were smaller in general and not statistically significant in two of the generations. --- Discussion This is the first study which examines the mortality pattern among three generations of migrants with a Moluccan background. High all-cause mortality was found among Moluccans compared with the Dutch. High mortality for most causes of death was found among Moluccans compared with the Dutch. All-cause mortality differences were observed in all Moluccan generations although the differences were smaller in second and third generation relative to the first adult generation. Similar all-cause mortality differences were found for non-mixed and mixed Moluccans compared with the Dutch by generation. --- Evaluation of data limitations A major strength of this study is the use of complete registry mortality data in combination with data of relatives. This gave us the unique opportunity to select and investigate Moluccans with a mixed background. Furthermore, this is the first study to give a total overview of mortality among Moluccans who are among the longest ICD-10 codes: D50-H95, K00-Q99, S00-V00. -Expected number of deaths is too small for accurate estimates. residing migrant group in the Netherlands. Till now, no mortality information was available among this group. Some limitations in this study should be considered when interpreting the results. Even though the Moluccan surnames are unique and can be distinguished from Dutch and Indonesian surnames, there may be some persons excluded due to the selection method used in this study, particularly Moluccan families with Dutch surnames. This may have led to some extent to selection bias among the group of mixed Moluccans. However, the number of persons excluded for this reason is small, and therefore unlikely to influence our results in a significant way. Second, we used neighbourhood income estimations as a measure for SES. We did not have individual income estimations. We therefore cannot exclude the possibility that the observed mortality differences may be partly attributable to the lower occupational class and income position of Moluccans. Furthermore, we did not have information regarding lifestyle, nutrition, physical activity and education at individual level. We cannot exclude the possibility that the observed mortality differences may be partly attributable to these factors among Moluccans. There is limited data regarding all-cause mortality among the younger generations of migrants in host countries. 17 Our current findings of high mortality across generations of Moluccans corroborate with earlier findings on other minority groups in Europe. 8,17,18 For example, the study of Liem et al. 17 found that younger generations of Asian-Indians had persistent high cardiovascular risk compared with Europeans. The high all-cause mortality among Moluccans may be influenced by several factors. The high all-cause mortality among Moluccans may be influenced by factors such as obesity. In van der Wal et al., 19 the rate of overweight/obesity was slightly higher than the Dutch general population. Moluccans seem to be even as likely to smoke and consume less alcohol compared with the Dutch. Factors such as cultural identity, geographic separation and social and economic position may play a role. Such factors may prevent Moluccans to adopt preventive behavioural and health measures to mortality. First, cultural identity may play a role. The Moluccan culture is preserved within local communities. 16 The traditional norms and values, social hierarchy, Malay language and family bonds are important within Moluccan communities. 20 The cultural influences go beyond the Netherlands as Moluccans have a strong bond with the country of origin and the families living there. Also in mixed marriages the Moluccan culture may be passed on as Moluccans have a strong sense of identity, feeling of pride in relation to their origin. 21 A strong urge for cultural perseverance may cause Moluccans to maintain a distinct lifestyle 19 that prevents the uptake of healthier lifestyles as promoted within Dutch society. Second, geographic separation may contribute to the differences. Segregated neighbourhoods, also known as Moluccan districts, were created in the 1950s to host the Moluccan migrant, and still exist today. 16,20 Approximately 40% of the Moluccan community lives in one of these districts and only Moluccans families are allowed to occupy the houses. The majority of these districts still have their own community centre and churches. The existence of these districts may strengthen the perseverance of the Moluccan culture and lifestyle, including among those who live elsewhere but who return to these districts at specific occasions. 22 Last, social and economic disadvantage may also contribute to the differences. The integration of Moluccans to the Dutch society has been slow and partial. In the decades after their 1951 arrival to The Netherlands, the first generation Moluccan men or women have been allowed to get a permanent position on the Dutch labour market. The second generation Moluccans had received primary education separately from the Dutch children, for some time in different barrack camps. Nowadays, Moluccans still experience ethnic discrimination 23 and have a lower socio-economic position within the Dutch society. Including those of the second and third generation, have lower-ranking occupations and are lower educated compared with Dutch. 16 The lower socio-economic position of Moluccans, for which could adjust only partially in our analyses, may have contributed to their unfavourable all-cause mortality over generations. --- Moluccans with a non-mixed and mixed ethnic background We found all-cause mortality rates to move towards Dutch levels among the non-mixed as well as the mixed Moluccans, with no major differences between the two subgroups. Multiple studies have observed a lower mortality among persons with a mixed ethnic background compared with the local host populations. [13][14][15] The Moluccans do not exhibit a similar pattern as we found a high all-cause mortality among Moluccans with a mixed ethnic background compared with the Dutch. This particular pattern might be due to the persistent influence of Moluccan culture, norms and values even among Moluccans with a mixed ethnic background. In addition, it may represent the long-term influence of the environmental and socioeconomic disadvantages that affected the segregated Moluccan district in which they grew up. --- First generation adult vs. child Although all-cause mortality was higher among all who came to the Netherlands in 1951, mortality rates were lower among those who came as child instead of adults. The mortality advantage among the first generation child may be due to the short exposure time to environmental, material and psychosocial disadvantages in the country of origin. 24 During the 30s and 40s, the living conditions in the former Dutch East-Indies Company were extremely poor by European standards. Also, experiences during the world war, when these men acted as soldiers, may have left it scars among the first generation adults more so than their children. Finally, many first generation adults were traumatised by the fact that the promise of returning to Moluccans was broken by the Dutch government after arrival in the Netherlands. The adult first generation the soldiers at that time were discharged at arrival in the Netherlands. 25 This has led to deep-seated distrust among the adult first generation. --- Conclusion More than half a century after their involuntary migration to the Netherlands, the Moluccan first generation adult still have considerably higher mortality compared with the Dutch population. Their mortality disadvantage is reproduced, albeit in attenuated form, in their offspring, including those who married with the Dutch. More fundamental and behavioural research is needed to identify the contributing factors of these persistent intergenerational mortality disadvantages among Moluccans. This mortality disadvantage is likely to reflect the cumulative effect of adverse events in early life and poor social position in the Netherlands. Their experience shows that mortality inequalities may persist, though in an attenuated form, over generations among ethnic minorities. --- Conflicts of interest: None declared. --- Key points High all-cause mortality is reflected in the higher mortality for most causes of death found among Moluccans compared with the Dutch; All three generations of Moluccans show high all-cause mortality although the extent of the differences are smaller in second and third generation compared with the first generation; Mixed as well as non-mixed Moluccans show high all-cause mortality compared with the Dutch.

Background: Mortality variations have been reported for ethnic minority groups compared with their host populations. It is uncertain how this mortality variations change over time as younger generations become older and mix with the host population. This study aimed to examine all-cause mortality among three generations of Moluccans in the Netherlands, and to compare Moluccans with a mixed and non-mixed ethnic background. Methods: We used data from the death and municipality registry for the years 2000 through 2013, including all registered Dutch inhabitants. A list of Moluccan surnames was used to select the Moluccan population. Mortality differences were calculated by Poisson regression, controlling for sex, age and district socio-economic status. Results: High all-cause mortality rates were observed in all generations of Moluccans although the extent of the differences between Moluccans and the Dutch were smaller in second (1.15, 1.07-1.23) and third generation (1.14, 1.00-1.29) compared with the first generation (1.55, 1.49-1.60). Higher all-cause mortality is also reflected in the higher mortality from most causes of death except neoplasms and external causes. Both mixed and non-mixed Moluccans showed high all-cause mortality among the first (child) and second generation compared with the Dutch. Conclusion: Our findings show a higher all-cause mortality in three generations of Moluccans compared with the Dutch. The results show that mortality inequalities may persist, though in an attenuated form, over generations among ethnic minorities.

INTRODUCTION Globally, children and young people are expected to experience three to four times as many extreme climate events in their lifetime compared to older generations according to the 2023 Intergovernmental Panel on Climate Change synthesis report 1 . Like preceding reports, this latest report provides yet another warning that urgent climate action must be taken by governments to safeguard the future of today's youth. With the planet 1.1 °C warmer than in 1850-1900, we are now experiencing increased climate related extreme events including more frequent heatwaves, floods, droughts and bushfires 1 . If no action is taken, the 2023 IPCC report 1 projects we are on track to a gain of 3.2 °C of warming by 2100, posing extreme risk to human life. As young people face the most vulnerability to climate change impacts, they are now appearing at the fore of the climate justice movement and are mobilising through structured youth-led networks such as the famous 'School Strike 4 Climate' 2 . While the authors acknowledge there are other forms of climate activism, this perspective will focus on describing the collective, non-violent youth climate activism research that has been the predominate focus in Australia. It will identify the climate justice narratives used by three key activism groups: the Australian Youth Climate Coalition , the Seed Indigenous Youth Climate Network , the 'School Strike 4 Climate' network and two litigation cases; Sharma v Minister for the Environment and Youth v Waratah. This perspective will first introduce the Australian context in which climate activism takes place, it will then focus on identifying the climate justice narrative within the youth climate movement to explore the practices taken by young people. --- THEORETICAL FRAMING AND APPROACH With much of the current discourse in youth climate activism situated around the concept of climate justice, it is important to define it and explain the theoretical framework that will be applied throughout this perspective. The climate justice movement evolved out of environmental justice, a concept that emerged in the late 20th century that exposed the racially unjust impacts of environmental damage on minority groups [3][4][5] . Climate justice concepts have been addressed locally and globally through large organisations, academia and grassroots activism 4,6 . Climate justice in grassroots activism evolved around the early 1990's with a focus on addressing the root causes of climate change, and an emphasis on the role of the production and distribution of fossil fuels in causing harm to future generations 3 . This perspective will frame contemporary youth climate justice using Newell's 4 four pillars of climate justice described as procedural, distributive, intergenerational and recognition justice to explore the various practices and discourse undertaken by young people to drive a climate justice agenda. Procedural climate justice refers to having fair, accountable, and transparent decision-making processes about the impacts of and responses to climate change. It also involves giving access to information, participation, and redress to those who are affected by climate change, especially the marginalised and Indigenous communities who often face exclusion and exploitation 4 . Distributive climate justice refers to how the costs and benefits of climate change and its solutions are shared among different groups of people. It also considers how factors such as race, gender, class, and ethnicity affect people's access to resources and their exposure to environmental injustice. Recognition justice is defined as respecting and representing the diverse views and rights of marginalised groups who face discrimination and exclusion. It also exposes and challenges the historical and current injustices that result from exploiting energy resources and people 4 . Finally, intergenerational climate justice is a concept that recognises the responsibility within current generations to ensure that future generations are not harmed by a lack of responsible decision making on pollution and other climate risk management 4 . Respect for and protection of human rights of future generations living in Australia are the core concepts of intergenerational justice in this perspective. Although these pillars often overlap and are not mutually exclusive, this perspective will focus on the intergenerational and recognition pillars to frame Australian youth climate activism. This perspective has been informed by a literature review that was undertaken by the first author as part of her PhD project. Using the genre outlined by npj, this perspective provided an opportunity to draw from the existing research to contextualise Australian youth climate activism and present suggestions for future research that can support young people. This paper will focus on Newell's notions of intergenerational and recognition justice. This is because these concepts highlight the growing realisation among young climate activists, including the first author, that political authorities and major polluters in Australia are failing to adequately address and take responsibility for climate change that will disproportionately affect young people and future generations. --- THE AUSTRALIAN LANDSCAPE: DISASTERS, MENTAL HEALTH AND YOUTH CLIMATE ACTIVISM Natural disasters are a normal occurrence in Australia, however, frequency, predictability and intensity of these events are increasing and are attributed to climate change 7 . In recent years, climate related bushfires and floods have devastated communities and caused widespread damage to the ecosystem and wildlife 8 . The 2019-2020 bushfires in New South Wales caused widespread destruction, burning 46 million acres of land, resulting in more than one billion animals perishing and the deaths of 34 people 9 . The ongoing and increased frequency of disasters in Australia has shown that current disaster preparedness processes cannot keep up, with slow recovery time periods increasing the vulnerability of communities for future disasters 10 . The vulnerabilities of Australian communities and lack of adequate disaster preparedness at the national level was highlighted again in March 2022. Less than two years after the bushfires, regions in Queensland and New South Wales experienced the most severe flooding event since 2011 11 . The floods affecting southeast Queensland and northern New South Wales led to the deaths of 22 people with more than 10,000 homes lost or damaged and 7000 people displaced 12,13 . Hence, it is understandable that young people, who will become the most impacted demographic, are actively pursuing structured modes of collective climate action. The severity, frequency and diversity of such disasters in Australia over the past 15 years provides a snapshot into the future for its young people. Increasing climate disasters are causing concern for young Australians, which is substantiated by a growing body of literature indicating increasing mental health issues in this population group [14][15][16] . For example, research by van Nieuwenhuizen et al. 17 and Gunasiri et al. 18 have identified how climate-related disaster events in adolescents' lives are affecting their mental health, with young people experiencing increased anxiety, grief, anger, powerlessness, fear and hopelessness. In the wake of the 2020 bushfires, a survey of young people found 90% of respondents aged 16-25 had experienced a natural disaster in the past 3 years 19 . Further, this survey also found that 88% of respondents felt that they are not being taught enough disaster resilience to protect themselves 19 . Climate related disasters in Australia will continue to increase in both severity and frequency, requiring a focus on development and implementation of national adaptation policies within affected communities and the ecosystems that they rely on 20 . In the absence of effective climate leadership, Australians and future generations will continue to be exposed to ongoing disasters that have already been linked to social, physical and mental health issues [21][22][23][24][25] . As the first generation that will be impacted by climate change throughout the entirety of their lifetimes, young people have emerged as leaders in advocating for climate action 26 . As a result of the increasing prevalence of climate risks, youth-led climate activism in Australia has gained momentum and evolved into a structured movement, most famously through the 'School Strikes 4 Climate'. Youth climate activism plays a pivotal role in driving a climate justice agenda using both institutional and noninstitutional pathways including litigation, protest, education, awareness-raising and personal carbon reduction practices [27][28][29][30] in activists attempts to hold governments and large corporations accountable for their role in the climate crisis. --- COLLECTIVE CLIMATE ACTION FOR INTERGENERATIONAL AND RECOGNITION CLIMATE JUSTICE: AN AUSTRALIAN PERSPECTIVE It is clear more than ever before that young Australians are concerned about climate change and are mobilising to take direct action. This is reflected by the emergence of large coalition groups such as the Australian Youth Climate Coalition , Indigenous youth-led Seed Mob and the 'School Strike 4 Climate' network and through litigation cases. These groups undertake various direct action, in the form of decentralised, non-violent grassroots activism primarily targeting governments, banks, and the fossil fuel industry to pressure decision makers to take responsibility for their actions. Each of these groups reinforce a climate justice narrative and leverage this through various climate action approaches both within and outside institutional systems. For example, SS4C primarily uses approaches that work 'outside' of institutional systems through protesting and the young people involved in the Sharma v Minister for the Environment litigation case have utilised working 'inside' institutional systems to take climate action. It is important to understand how climate activism is practiced in youth-led organisations, which then can be applied to other personal and collective action in the Australian context. AYCC, for example, was formed in 2006 by 27 youth organisations to encourage young people to become leaders in climate advocacy and remains Australia's largest youth run grassroots organisation. With over 15 years of experience, the coalition demonstrates the diversity of their activism to run efficient campaigns that adopted an 'educative-movement-building' approach. This approach is grounded by a broad climate justice narrative frequently citing the social, economic, Indigenous and environmental issues associated with climate change 31 . Furthermore, AYCC's approach to activism utilises hybrid organisational structures to create a space for young people to act as capable political climate advocates and simultaneously provide education to enhance their leadership and agency. AYCC illustrates this by targeting schools and organisations and uses storytelling strategies that frame 'youthfulness' as a strength to connect, motivate, educate and mobilise young people to engage in climate action 31,32 . AYCC also runs 'Youth Decide" campaigns during federal elections. In the 2022 federal election, the campaign focussed on the importance of opening up the conversation about the impacts of climate change with family and friends in order to encourage more votes for political parties and independents with strong climate policies 33 . AYCC recorded over 3000 conversations with youth volunteers during the election period, demonstrating themselves as strong advocates for climate justice. These approaches to collective climate activism in Australian youth has been immensely successful at engaging young people to participate in taking climate action. It appears that the emphasis of personal growth through education and leadership within a collective climate action movement is an effective model that could be adapted locally and globally. Another youth-led network in Australia is the 'School Strike 4 Climate' network . The network took the world by storm in 2018 after climate activist pioneer Greta Thunberg skipped school to sit outside parliament house in her home country of Sweden. Two years later in 2021, the SS4C network continued to strike with over four million people globally participating, including 350,000 students and workers in Australia 34 . In 2023, SS4C describes themselves as an established student-led, decentralised, nonviolent grassroots network 2 . Their primary target is the fossil fuel industry, with a focus on a just transition and job creation for communities and people reliant on the industry. Most recently, they continue to drive an intergenerational climate justice agenda by collaborating with adult-led climate justice organisations, including Extinction Rebellion, with joint campaigns including #DontNabOurFuture, targeting Australian bank NAB who currently support gas and coal companies Whitehaven and Santos 2 . Whilst the network is widely known for its strikes, SS4C also provides training workshops and summits to educate budding activists on social movement strategies 35,36 . Much like AYCC, SS4C uses peer-to-peer education and leadership development opportunities to engage collective action and build political and social agency in young people 35,36 . It appears that these organisations have been effective in mobilising young people to advance the intergenerational climate justice agenda. However, a critical component to the youth climate justice movement in Australia is the way is it being enacted. In addition to focusing on holding decision makers and major polluters accountable for their actions, AYCC and SS4C build youth empowerment through peer-led education. By becoming agents of change through actively educating themselves and others, they are able to utilise their personal and collective resources, knowledge and critical thinking skills to navigate their own future to protect themselves and the next generation from climate change related harm. The Australian youth climate movement also intersects recognition and intergenerational justice concepts. For example, young people of colour and First Nations youth have brought recognition justice to the fore of their campaigns-where a focus is given to Indigenous or marginalised groups who face discrimination or exclusion concerned with the unequal impacts of environmental issues 4 . AYCC recognised the need for authentic diversity and inclusion, and in response launched the People of Colour Climate Network to include a more focused multicultural voice into their campaigns 33 . The network now has 200 young climate activists, who regularly take direct action to include and educate marginalised communities about climate change and its impact on future generation of Australians. For example, during the 2022 Federal election, the network distributed flyers, posters and social media content in five languages, other than English, to engage migrant and religious relatives in the climate conversation and to encourage them to vote for climate action supportive politicians 33 . Also originating within AYCC, the Seed Indigenous Youth Climate Network Programme , was established in 2020 as an independent organisation. It is the first and only group of First Nation's youth-led climate organisation, made up of 250 Aboriginal and Torres Strait Islander youth volunteers and 35,000 supporters. The group aims to empower young Indigenous people through fostering leadership and Caring for Country 37 . Their primary campaigns have been focussed on fracking in the Northern Territory, Queensland and Western Australia by working alongside local Aboriginal communities 37,38 . The campaign is embedded in a long history of Indigenous land rights 38,39 and draws attention to the damage caused to First Nations communities by coal and gas mining. Their campaign, that focuses on the damage fracking has on human health, Country and water, is linked to climate justice through recognition of land rights and self-determination 38 . In addition, Seed held the HEAL Country summit in 2022 that aimed to provide a space for First Nations young people to share, discuss and inspire each other with their ideas for future networks and campaigns 37 . Like AYCC, Seed is also grounded in a climate justice agenda and peer-learning with a focus on addressing systemic inequality and working alongside marginalised communities. This is an important Australian perspective, as it demonstrates how Seed Mob is also a learning organisation, that can support the diversity of Australian youth climate activists' needs both formally and informally within their organisation. As illustrated above, perspective lies at the crossroads of recognition justice and intergenerational justice within the organised collective youth climate movement. This becomes evident through Seed Mob's campaign strategies, which emphasise Newell's recognition justice pillar through the unjust and disparate consequences of fossil fuel production on marginalised communities. In the context of intergenerational justice, young First Nations peoples are also championing the cause, addressing both the aforementioned inequities and the need for a secure future for their communities 38 . --- YOUTH-LED CLIMATE LITIGATION: AN INTERGENERATIONAL JUSTICE PERSPECTIVE Intergenerational climate justice has also been a key component in emerging litigation cases led by young people throughout the world including in Germany, the Netherlands and Canada 40,41 . Following suit in Australia, one group of eight teenagers argued that the Australian government had a legal duty of care to protect them and future generations from the harmful effects of climate change. The plaintiffs in Sharma v The Minister for the Environment argued that the Australian Federal government's approval of Whitehaven coal mine in New South Wales was inconsistent with its duty to protect the environment and the health and wellbeing of its citizens 42 . The case was ultimately unsuccessful, as the Federal Court ruled that the government did not have a duty of care towards its citizens in relation to climate change. This underscores a significant acknowledgement of how the existing political decision-makers and polluters receive support from the broader environmental legislative framework in Australia. However, the case sparked important discussions about the legal responsibilities of governments towards their citizens in the face of climate change and has inspired further legal action and activism around the world 42 . In another landmark climate related human rights case bought on by First Nations young people, known as the Youth Verdict v Waratah Mine case. The case was centred around a proposed coal mine that was expected to cause significant environmental damage and contribute to impacts of climate change in an already highly affected area of Australia. Despite the Queensland Human Rights Act 43 not specifying the right to a healthy environment or climate, the activists argued that the Galilee Coal Project, owned by Waratah Coal, would impact sections 6 and 23 of the Human Rights Act 41,43 . This included property rights, the right to life, right to privacy, rights of the child, the right to freedom from discrimination and cultural rights of Aboriginal peoples and Torres Strait Islander peoples 43 . The case was successful in blocking the development of the mine, demonstrating the power of human rights law in protecting vulnerable populations and preserving the environment for future generations 41 . The landmark cases of Youth Verdict v Waratah Coal and Sharma v Minister for the Environment have had a significant impact on the climate justice movement in Australia, demonstrating the potential for a legal pathway as a tool for young activists. These cases illustrate that young individuals possess the capability to demand accountability from governments and corporations regarding their choices concerning the human rights of future generations in the context of climate change 41 . Similarly, Peel and Markey-Towler's 42 case study explored a 'recipe for success' in youth climate activism. They specifically identified an intergenerational justice focus as a powerful tool and demonstrated how making innovative legal arguments that emphasise duties of protection is an important component of such litigation cases. As demonstrated in the Sharma case, young people have used a duty of protection claim as a legitimate argument for demanding intergenerational justice. This intergenerational justice perspective can be more widely adopted in Australia and globally as an effective campaign tool. In summary, the above snapshot of Australian youth climate activism demonstrates the sophistication of grassroots action and the power of educational approaches individuals and organisations have taken in Australia. Their request for a just and equitable future has often become the core of their campaigns, reinforcing the validity of Newell's 4 intergenerational and recognition justice framing of youth climate activism in Australia. --- BARRIERS TO CLIMATE ACTIVISM IN AUSTRALIA In this perspective, it is crucial to emphasise that Australian youth climate activists frequently encounter a significant obstacle in the form of conservative media outlets that consistently undermine and obstruct young people's endeavours to initiate a meaningful dialogue about their own future. These organisations can create a highly critical and hostile environment when discussions about climate justice arise 44 . The conservative media often portrays young climate activists as naive and misguided, dismissing their concerns about climate change as alarmist and unfounded. For example, Mayes et al. 45 conducted a media analysis on Australian newspapers between August 2018 and December 2019, which analysed the dominant narratives presented by the conservative media during the mobilisation of SS4C. The study found that youth climate activists were painted as politically inexperienced, vulnerable, and as "kids", or overly emotional and therefore "in a phase" 45 . This kind of ageism and dismissal of youth was found to be expressed in widespread journalistic reporting on climate activism across the mainstream media, acting perhaps as one of the largest barriers to legitimising youth climate activism. Moreover, the former Australian Prime Minister Scott Morrison fuelled much of this discourse in the media by condemning the strikes and berating the students for missing school 36 . The "school-or-strike" moral debate inundated the media and diverted the focus from why young people were protesting and the intergenerational justice claims they made. Similarly, Alexander et al. 44 in their analysis of mainstream media representations of adult responses to the SS4C protests also identified how adults reinforced the notion that education in school should be the priority if young people want to create change. However, climate change education in Australian schools is still heavily focused on the science and not adequately explored across humanities or governance, failing to acknowledge climate change as a wicked problem that needs to be addressed through a multidisciplinary and holistic approach 46 . This highlights a gap for providing adequate climate education in formal schooling that is integrated across the curriculum 47,48 . This perspective demonstrated evidence of how young climate activists in Australia are "learning on the job" to become effective advocates for intergenerational climate justice. This presents another opportunity for researchers and educators to reimagine climate education and use a multidisciplinary approach that reflects climate justice and further supports young people in their advocacy 36,49,50 . In addition to media adversaries, young people also face threats to their right to protest in various Australian States. There has been a concerted effort to criminalise climate protests and to limit the Australian public's right to peaceful assembly. New legislations brought in by several states in Australia including Queensland 51 , New South Wales 52 and Tasmania 53 aim to undermine the fundamental principles of democracy and limit young climate activists' ability to undertake direct climate action. These laws are not congruent with human rights under international law. Article 12 of the Convention on the Rights of the Child 54 , states the following: "Parties shall assure to the child who is capable of forming his or her own views the right to express those views freely in all matters affecting the child, the views of the child being given due weight in accordance with the age and maturity of the child". The Convention on the Rights of the Child clearly implies that in cases where there are concerns that societal issues that have a direct impact on the child, in this case, climate change, the "best interests of the child" 54 must be determined by listening to the collective voices of the children. These international laws can provide a useful platform for Australian youth climate activists to reclaim their rights in Australia. Despite this, there has been efforts to curb human rights of Australians through the introduction of legislation by the 2019-2023 NSW Liberal state government , who introduced a new anti-protest law. This was rushed through by the Legislative Council amending the Roads and Crimes Legislation Amendment Bill 2022. The new bill introduced measures that deliberately repress protestors, with provisions for up to 2 years imprisonment and a $22,000 fine 55 . The ambiguous wording of the bill makes it unclear whether young climate strikers are legally allowed to advocate for climate justice in this manner without fear of arrest. In order to achieve meaningful climate justice and support the voices of young people in Australia, it is crucial that such laws are revisited. The increased repression of protests and direct action through legislation in Australia is of concern to youth climate activists. These laws prevent youth climate activists from expressing their political agency to drive societal change outside of institutional pathways 47,56 . This ultimately excludes young people's voices in climate change discussions and decisions, limiting their ability to advocate for a sustainable future in Australia. While the Australian states' governments have worked to silence youth voices, young climate activists have successfully advocated for political change. A month prior to the last Federal elections in Australia in 2022, 80,000 young people between the ages of 18 and 24 enrolled to vote within a week 57 . This was a major win for young Australian activists. In May 2022, the more socially progressive Labour government was selected, with a substantial increase in votes for the Australian Greens party 58 , and historically low votes for the Liberal-National Coalition in Generation Z and Millennial age groups 58,59 . This result was welcomed by youth climate activists, after an almost decade long inaction on climate change policy from the centre-right Liberal-National Coalition in government, that largely supported fossil fuel extraction 60 . With a more climate-friendly government, this has created a more progressive socio-political space for intergenerational justice conversations. Although the Labour Party's stance on climate change is a step in a more carbon neutral direction with changes to the Climate Change Act for more aggressive emissions reductions, with the approval of new oil and gas exploration projects 61 Australia continues to fail to transition away from fossil fuels quickly enough, this is compounded by a notable deficiency in adequately addressing a just transition for communities. The next step for the Australian youth climate movement is to push for the acceleration of a just transition, however, there is extremely limited research on young people's climate action in regional communities that will be most affected. In mining, Indigenous and other small communities, it is unknown what a just transition will look like. This perspective paper illustrates that whilst SS4C works at lobbying fossil-fuel industries and governments to move towards a carbon-free future, the voices of the youth who will be impacted by this transition have not yet been included in the climate conversation. A report by Edwards et al. 62 suggested that the SS4C campaign towards a 'just transition' is not always accepted by coal communities, with miners and community members seeing environmentalists as having very little understanding of the reality of living in a regional, coal-reliant community. The idea of a 'just transition' is often taken negatively and associated with threats of loss of high paying jobs and the future risk of not being able to support their families 62 . Curran 63 also points out this disconnect, where the culture associated with employment and a prospering economy associated with coal mining in regional Australia is something that is deeply embedded into people's identities 63 . This perspective and understanding of what mining means to regional Australians has not been actively acknowledged in the climate movement. Given the lack of research with youth in these communities, the question remains: How can an intergenerational climate justice agenda be driven at the local level when we don't know what young people in regional areas want climate justice to look like? Future research should aim to explore such gaps in our knowledge. Indeed, previous research tended to focus on youth climate activism in metropolitan areas in Australia. It is essential to recognise this bias in climate activism research, as emphasised in this perspective paper. There is a pressing need for a greater representation of culturally and geographically diverse voices among Australian youth in climate activism research, as well as within the broader climate justice movement in Australia. We cannot assume homogeneity of young people's needs, particularly those who live in regional Australia, who will be the most affected by climate related disasters and whose voices are currently missing. It is where direct action is needed by governments to ensure survival of these communities. Future research must take into consideration localised, place-based climate justice concerns of young people, recognising that community contexts go beyond the rural/urban divide. Research with young people should also utilise participatory and collaborative approaches to address power dynamics often inherent in traditional research methods. Inclusivity in research is exemplified by White et al. 36 , Godden et al. 64 , Sanson and Bellemo 16 who actively involve young people as co-authors in their research. In conclusion, examining the youth climate movement with a climate justice lens in Australia highlights the ways young people leverage and hone their leadership and advocacy skills to drive climate justice to the fore of the youth climate movement. This perspective also identified the barriers associated with engaging in climate activism and pinpointed various opportunities for young people to be supported-namely through holistic climate justice education and furthering research into youth climate activism in regional Australia. After all, it is regional Australia that will shoulder the lion's share of the 'just transition' to renewable energy, all the while enduring the escalating and compounding impacts of climate change. Received: 14 April 2023; Accepted: 16 November 2023; --- --- COMPETING INTERESTS The authors declare no competing interests. --- ADDITIONAL INFORMATION Correspondence and requests for materials should be addressed to M. Hohenhaus. Reprints and permission information is available at http://www.nature.com/ reprints Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

This perspective brings together published peer reviewed primary research on youth climate activism in Australia and provides context of the political and social landscapes in which young people are taking climate action. As the generation most vulnerable to the impacts of climate change, young people all over the world have mobilised to drive a climate justice narrative to the fore of the climate movement. Climate justice framing will be applied to contextualise youth climate activism in Australia. This perspective also addresses the context-specific challenges faced by youth, including the media's role in shaping public perceptions and, anti-protest laws that restrict the right to protest. Finally, this perspective highlights the opportunities for how to support youth climate activism in the future.

However, currently, there is no available vaccine and the efficacy and safety of potential treatments remains unproven [5]. Implementation of contact and respiratory precautions to prevent the spread COVID-19 is essential [6][7][8][9]. Although cases of this emerging disease will likely overwhelm health systems worldwide [10], transmission prevention is particularly important for the Global South, specially most of sub-Saharan Africa [11,12] and Latin America [13,14], where health systems were already under significant strain before the discovery of SARS-CoV-2 [15]. Ecuador's healthcare system has been among the hardest hit [16,17], with more COVID-19 cases per capita than nearly any other country in the Global South [18] and one of the highest rates of mortality in Latin America [19]. SARS-CoV-2 transmission prevention strategies include low-cost, high-impact behaviors such as hand hygiene, face mask use, and social contact avoidance [20]. Unfortunately, implementation of these measures is not sufficiently widespread. To encourage adoption of these behaviors, calls have been issued for governments and health agencies to disseminate timely, accurate, and science-based information to the public, assuming this will result in behavioral change [21][22][23]. Effectively changing human behavior, however, often requires going beyond providing information [24][25][26]. Adoption of preventive measures, particularly in the context of infectious disease, is largely determined by KAPs [27][28][29][30][31]. In this study, we explored the understanding of COVID-19 by the Ecuadorian public, in order to inform the design of effective prevention and control strategies in the context of the current pandemic. --- Methods --- --- Measures The KAP questionnaire employed in this study was based on a previous one developed by Zhong and colleagues in Hubei province, People's Republic of China [32]. Zhong et al.'s questionnaire was translated into Spanish, adapted to the Ecuadorian context, and updated for recommended prevention practices. The first author, an Anglophone researcher, reviewed the English-version constructs, anchored their conceptual expressions in Spanish and translated the instrument. The fourth author, a Hispanophone researcher reviewed the Spanish-version and made minor edits. The questionnaire was then sent to two additional Hispanophone researchers, one informed of the research purpose and the other not, to account for linguistic and conceptual equivalences. The first author then synthesized the translations in reference to their equivalence to the English questionnaire. The first part of the questionnaire consisted of demographic variables. These included age group , gender, marital status, education, occupation, and place of current residence . The second part of the questionnaire consisted of KAP items. Following Zhong et al., the knowledge component had 12 questions: 4 discussing the clinical presentation of COVID-19, 3 regarding transmission pathways, and 5 regarding prevention and control. Participants could answer "true", "false", or "do not know." The term "COVID-19 virus" was employed, following Zhong et al., because it is more easily understood by non-specialists. Each correct answer was assigned 1 point; incorrect answers or unknown answers were assigned 0 points. A participant could receive a total knowledge score ranging from 0 to 12; higher scores indicate better knowledge of COVID-19. Additionally, following Zhong et al. attitudes towards COVID-19 were measured by 2 questions. The first question assessed agreement or disagreement on whether COVID-19 would ultimately be controlled. The second question assessed confidence in Ecuador's ability to win the battle against COVID-19. Finally, practices were assessed using 3 items regarding behaviors. In addition to Zhong et al.'s item asking participants to report whether they had gone to crowded places and whether they wore a face mask when outside their home during the week before taking the survey, we also assessed whether participants reported washing their hands each time they returned home or were in contact with another person. In addition to Zhong's assessment, if a participant indicated that they had ventured out of their home in the week before taking the survey, they were also requested to specify how many times they had done it, the reason, and whether they maintained a distance of at least 2 m from other people while out. --- Statistical Analysis Frequencies of correct answers and various attitudes and practices were described . Independent-samples t-test, one-way analysis of variance , and Chi-Square tests, as appropriate, were used to compare members of different demographic grouping's knowledge scores, attitudes, and self-reported practices. We used binary logistic regression analyses to identify demographic factors associated with each attitude and practice. Factors were selected with an enter method to avoid both inflationary and deflationary effects. Unstandardized regression coefficients and odds ratios , respectively, and their 95% confidence intervals were used to assess the associations between demographic variables and KAP. Since only some of the individuals reported leaving their homes during the week before the survey, the follow-up questions were applied only to this subset of the study population, and the resulting data did not meet assumptions for regression Therefore, we only report frequencies for this subset. All data analyses were conducted with SPSS version 26.0. The statistical significance level was set at p < 0.05. --- Results A total of 2491 individuals consented to participate in the survey. Participants who reported either being under 18 years of age or not being Ecuadorian on the screening questions were excluded, and no further data were recorded for them. After removing participants who skipped all substantive questions, 2399 individuals were retained. This final sample was majority female . About half of the participants were between 30 and 49 years of age, married , and college educated . Participants represented all labor sectors and nearly all major population centers in Ecuador. A large proportion of participants were from the national capital, Quito. Full demographic characteristics are shown in Table 2. The correct answer rates for the 12 questions on the COVID-19 knowledge questionnaire ranged from very low to near universal knowledge . 2). Gender, age, marital status, and place of residence were not associated with differences in levels of knowledge. Respondents displayed mixed opinions regarding whether COVID-19 will be successfully controlled. Almost one-half of the participants agreed that COVID-19 will eventually be successfully controlled, while about one-fifth disagreed. The remainder stated that they did not know. Respondent's attitudes toward final success differed significantly by all demographic variables . Overall, the binary logistic regression model successfully classified 69.9% of cases and explained about 9% of all variance in attitudes to toward final successful control of COVID-19 . Binary logistic regression revealed that women were less likely to predict successful control than men were p < 0.01). Additionally, older participants were more likely to believe successful control will eventually be attained . Participants holding Bachelor's or Master's degrees were more optimistic as compared to those with less schooling , p < 0.01 and OR 1.27 , p < 0.05, respectively). Participants who predicted unsuccessful control displayed lower levels of knowledge than those who predicted successful final control , p < 0.05). Although preliminary Chi-square analyses indicated significant differences by occupation type, place of residence and marital status, they did not emerge as significant predictors in the regression model. Similar to their view regarding the global situation, participants reported mixed opinion as to whether Ecuador will successfully control COVID-19. About two-thirds of participants were confident that Ecuador would succeed, while the remainder disagreed . Respondent's attitudes toward Ecuador's success differed significantly by several demographic variables: gender, age, marital status, city of residence, as well as knowledge . Overall, the binary logistic regression model successfully classified 64.0% of cases and explained about 8% of all variance in attitudes to toward Ecuador's winning the battle against COVID-19 . Binary logistic regression revealed that, as participants were older, they were more likely to believe successful control will eventually be attained , p < 0.01; oldest group as compared to younger groups, OR 2.03 , p < 0.01). Divorced persons were more optimistic as compared to all other groups ; p < 0.05). Although preliminary t-test analysis indicates significant differences by knowledge, and although Chi-square analyses indicated significant differences by place of residence and gender, they did not emerge as significant predictors in the regression model. In the realm of practices, participants overwhelmingly stated that they followed approved practices. Most participants stated that they have not gone to a crowded place in the prior week. Respondent's self-report that they had avoided crowded places differed significantly by gender, age grouping, marital status, and place of residence . Overall, the binary logistic regression model successfully classified 88.7% of cases and explained about 9% of all variance in approaching or avoiding crowded places . Binary logistic regression revealed that men were less likely to have avoided crowded places than women ; p < 0.01). Furthermore, as participants were younger, they were less likely to have avoided going to crowded places , p < 0.01; younger groups as compared to oldest groups, OR 0.70 , p < 0.01). Persons who were single-never married were more likely to have gone to crowded places than all other groups , p < 0.05). Residents of some cities, specifically Guayaquil , p < 0.01), Machala , p < 0.05), Manta , p < 0.05), or Santo Domingo de los Tsáchilas , p < 0.05) were less likely to have avoided crowded places. Although preliminary t-test analyses indicated significant differences by knowledge and chi-square analyses by occupation type, they did not emerge as significant predictors in the regression model. Similarly, most participants reported that they had worn a face mask when they had left the home in the prior week. Respondent's self-report that they had worn a mask whenever they leave the home differed significantly by all demographic variables except marital status . Overall, the binary logistic regression model successfully classified 93.4% of cases and explained about 9% of all variance in mask wearing . Binary logistic regression revealed that women were less likely to report wearing masks than men ; p < 0.01). City dwellers were more likely to wear masks compared participants who reported living outside of big cities , p < 0.01). Although preliminary Chi-square analyses indicated significant differences by educational level, occupation type, marital status, and t-test indicated significant differences by knowledge they did not emerge as significant predictors in the regression model. Nearly all participants stated that they had washed their hands for at least 20 s every time after returning home or touching another person. Respondent's self-reports of handwashing significantly by age, marital status, and knowledge . Overall, the binary logistic regression model successfully classified 69.9% of cases and explained about 8% of all variance in handwashing . Binary logistic regression revealed that persons who were unemployed were most likely to report always handwashing as compared to all other employment groups , p < 0.05). Persons who reported not always handwashing held lower levels of knowledge than those who reported always handwashing , p < 0.01). Although preliminary Chi-square analyses indicated significant differences by age and marital status, they did not emerge as significant predictors in the regression model. Most participants stated that they had not left home in the prior week . Among those who had left the home the week before the survey, a majority had only left the home once . Seventy-eight had left home two or three times, and only 40 people had been outside their home four or more times. Most people who had left the home reported maintaining a distance of at least two meters from other people , with the rest admitting that they had not maintained distance. Reasons reported for leaving home included performing essential work , purchasing food for self, family, or pets , purchasing medicine , or visits to physicians or veterinarians . Indeed, of the 269 people who had ventured outside, only one had left for a reason other than essential travel; he reported leaving home to purchase marijuana. --- Discussion This study was conducted during the COVID-19 outbreak in Ecuador. These findings outline areas that should be addressed by the Ministry of Health in Ecuador, the private sector, and non-profit actors to help prevent the spread of SARS-CoV-2. In comparison to Zhong et al.'s Chinese sample, the Ecuadorian participants reported a substantially lower rate of knowledge about COVID-19. Among the Chinese sample, participants scored 90% on the knowledge questionnaire; however, Ecuadorian participants answered 82.3% questions correctly. Transformed into school marking terms, their grades were a letter grade lower, indicating substantial room to educate the population about SARS-CoV-2/COVID-19. Our data indicate that four items should be specifically addressed. First, there is substantial misconception about potential zoonotic transmission that should be countered. Second, the role of medical masks in limiting the spread of the virus that causes COVID-19 could be promoted. Third, differences between symptoms of the common cold and of COVID-19 are often confused in our sample. Finally, the public may wish to be aware of the populations that are most vulnerable to COVID-19-older people, people with chronic illnesses, and persons who are obese-so that they may know who in their social networks are most at risk. Promoting improved understanding of COVID-19 may also help promote confidence that COVID-19 will eventually be controlled. Less than half of the participants were confident that COVID-19 would ultimately be controlled, and only two-thirds had confidence in Ecuador's ability to defeat COVID-19. This stands in sharp contrast with Zhong et al.'s findings where over 90% of the Chinese participants believed that COVID-19 would eventually be controlled and over 97% believed that China would be successful in its battle against COVID-19. Ecuadorians, at least in this sample, are much less optimistic. Women, older people, and more highly educated people are the least optimistic about succeeding against COVID-19, representing primary targets for persuasion. This persuasion will not be a problem of simple education. Our data indicate that people with higher knowledge of COVID-19 had fewer "I don't know" answers about eventual success, but they were not more likely to believe in eventual success. Moreover, greater knowledge had no association with optimism about Ecuador's ability to win this battle. These findings indicate that greater knowledge is insufficient to change attitudes by itself. Attitudes are a primary motivator for action on threats to health [25,33,34]. That is, when people believe that success is likely, they are also more likely to act. Therefore, it is essential that future studies determine why people are not optimistic about the world's and Ecuador's ability to overcome COVID-19. Reasons for this could include distrust in governing institutions, pessimism about science and medicine, or other potential explanations. Overall, participants in this study reported adopting the practices recommended to limit the spread of COVID-19. A supermajority of the participants reported avoiding crowded places, always wearing a mask when outside the home, and washing their hands for at least 20 s upon returning home. When leaving their home, usually once a week, participants reported having done so for essential purposes, ‡ For A1, F value for ANOVA, t value for t-test for A2 significant at p < 0.01 and maintaining the recommended 2-m distance from other people. Men, young people, single people, and unemployed people engaged in the riskiest behaviors. Even among these populations, however, reported compliance with recommended behaviors was extremely high. Importantly, knowledge was not a predictor for avoiding crowded places or wearing a mask. Although handwashing was associated with knowledge, it only explained part of the variance. Rather than repeating information about the importance of these practices, it may be more important to look at why men, younger people, single people, and unemployed people are less likely to adopt the recommended behaviors. Expected gender roles may be encouraging men to take more risks of going out in public. Younger people may believe they are less vulnerable to COVID-19 or its complications. Single persons may have no one else in the household to purchase food and medicine or have less access to protective resources, and unemployed persons may be unable to afford to stockpile food or to purchase face masks. Additionally, many unemployed individuals in Ecuador subsist through informally selling products or services in the streets [35]. Certainly, the informal sector is larger in Guayaquil , the most populous city in Ecuador, than in Quito . These people depend on their daily earnings to survive and, therefore, may be forced to leave their homes on a daily basis. Therefore, public messaging campaigns should be accompanied by even stronger efforts to provide goods and resources to the most in need segments of the population to encourage these groups to adopt the actions recommended to prevent the spread of COVID-19. --- Limitations Although this study is the first, to our knowledge, to examine knowledge, attitudes, and practices related to COVID-19 in a South American population, there are some limitations. In our sample women, people living in urban areas and people with high levels of education were overrepresented. Greater participation of individuals from rural areas or with lower educational levels could have reduced the measured knowledge about COVID-19. Therefore, KAP towards COVID-19 among these populations deserves attention. However, in the present context, face-to-face methods of recruitment would place both researchers and participants at greater risk of COVID-19. Therefore, studies like this one should carefully weigh the potential benefits and risks of face to face data collection methods. --- Conclusion In summary, our findings indicate that people in Ecuador are likely to have moderate to high levels of knowledge related to COVID-19 and are likely to practice recommended behaviors to prevent the further spread of COVID-19. The greatest area for improvement is regarding pessimistic attitudes on the eventual control of COVID-19, which suggests that health education and outreach should not only focus on knowledge and prevention practices, but should also promote optimistic attitudes. --- Compliance with Ethical Standards Conflict of interest The authors declare that they have no conflict of interests. --- Informed Consent The study protocol was approved by the Institutional Review Board at Ohio University as an exempt study with a waiver of signed informed consent. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Preventing the transmission of SARS-CoV-2 (causative agent for COVID-19) requires implementing contact and respiratory precautions. Modifying human behavior is challenging and requires understanding knowledge, attitudes, and practices (KAPs) regarding health threats. This study explored KAPs among people in Ecuador. A cross-sectional, internet-based questionnaire was used to assess knowledge about COVID-19, attitudes toward ability to control COVID-19, self-reported practices related to COVID-19, and demographics. A total of 2399 individuals participated. Participants had moderate to high levels of knowledge. Participants expressed mixed attitudes about the eventual control of COVID-19 in Ecuador. Participants reported high levels of adoption of preventive practices. Binomial regression analysis suggests unemployed individuals, househusbands/housewives, or manual laborers, as well as those with an elementary school education, have lower levels of knowledge. Women, people over 50 years of age, and those with higher levels of schooling were the most optimistic. Men, individuals 18-29, single, and unemployed people took the riskiest behaviors. Generally, knowledge was not associated with optimism or with practices. Our findings indicate knowledge about COVID-19 is insufficient to prompt behavioral change among Ecuadorians. Since current COVID-19 control campaigns seek to educate the public, these efforts' impacts are likely to be limited. Given attitudes determine people's actions, further investigation into the factors underlying the lack of confidence in the ability of the world, and of Ecuador, to overcome COVID-19, is warranted. Edu-communicational campaigns should be accompanied by efforts to provide economically disadvantaged populations resources to facilitate adherence to recommendations to prevent the spread of the virus. The world is rapidly learning about the SARS-CoV-2 virus, including its origin, transmission, and progression of the condition it causes, . Characterization of clinical manifestations associated to COVID-19 and diagnosis of SARS-CoV-2 infection are also improving [1,4].

INTRODUCTION As of March 2010, the undocumented population in the United States was estimated at 11.2 million . Mexican migrants make up the largest portion of the U.S. undocumented population , of which youth are a fast growing segment . The immigrant flow from Mexico to the U.S. has a high proportion of youth; 40% of the immigrant flow in 2001 and 2002 was made up of youth ages 12 to 24 . Similarly, for return migrants to Mexico, a Mexican Population Council survey indicated that approximately 30% were between ages 15 and 24 . Young temporary migrants have less formal education, are less likely to have migrated before, and are more likely to enter the U.S. without legal authorization compared to older adult temporary migrants . Due to their young age, undocumented youth are more vulnerable to dangers along the journey and to a host of negative outcomes in the U.S. including low educational attainment and living in poverty . Significantly less is known about the trajectories of migrant youth once they return to Mexico after having resided in the U.S. Despite the large number of youth migrating between the U.S. and Mexico, we know little about this understudied population. Most research on immigrant youth has focused on immigration and assimilation processes in the U.S., while largely ignoring the reasons that brought them to the U.S. in the first place. Contextual factors in youths' communities of origin, aspects of their family life, and individual characteristics and attitudes toward migration likely influence youths' decision to migrate and may have a lasting impact on youths' immigration and adaptation experiences in the U.S. . Likewise, these same factors may influence youths' decisions to return to Mexico and their reintegration into their home communities ; . Yet, we are aware of no studies that have examined these issues among youth. In the present study we aim to fill this gap by examining migration decision-making among Mexican youth. Our main objective was to explore qualitatively the role of individual and family characteristics in the migration decision among youth ages 15 to 24 by comparing youth who have migrated to the U.S. and returned to Mexico to youth in Mexico who have not migrated. A second aim was to contrast these findings between two Mexican communities with differing migration densities to explore the influence of community-level factors on youth migration decisions. Comparing decision-making between return migrants and non-migrants and across two communities enabled us to evaluate individual, family, and community factors that may influence migration behavior. A better understanding of what differentiates migrant youth from non-migrants in their country of origin is an important first step in assessing the impact of migration on youth. Findings from this exploratory qualitative study can inform how youths' adaptive experiences in the U.S. and their decisions to return home might be shaped by factors prior to migration . --- Migration decision-making Migration theory has focused predominantly on economic, social, and cultural motivations for international migration . Microeconomic models state that migrants weigh their education, experience, and expectations of income differentials in both countries against the perceived costs of migration and their decisions often reflect a larger household strategy to minimize risk . Implicit in the microeconomic model is a level of uncertainty in the perceived costs and benefits of migration . In the face of this uncertainty, individuals who choose to migrate likely have a higher risk-taking propensity than those who, in similar contexts, choose not to migrate . Risk preferences may be particularly relevant for youth migration given that adolescence is a time of increased participation in risky behaviors . Social theories on migration point to the important role of family and community norms on migration decisions . Kinship and family ties form an important source of social capital, providing information and support to prospective migrants that reduce the costs of migration and settlement. As the prevalence of migration increases, the population with access to networks becomes less selective and the migration process becomes self-perpetuating. In communities with high rates and a long tradition of migration, a "culture of migration" emerges in which international migration becomes normative . The decision to migrate then becomes an accepted path toward economic well-being and the values associated with migration become ingrained in the sending community . These same economic and social theories also have been applied to the phenomenon of return migration. These theories suggest that individuals with fewer economic investments in the U.S. or those that are less successful may be more likely to return home . Other migrants may plan to return home once a target amount of remittances or savings are earned . Individuals who are less socially integrated in the receiving country while maintaining strong ties to kin in the sending country may be more likely to migrate home . Though these models have been developed and applied to the experiences of unmarried males and families primarily, they suggest factors that may be relevant to youth migration decisions, namely educational and occupational expectations, attitudes toward risk and uncertainty, family and kinship migration experience, and community migration norms . The present study attempts to enhance our understanding of the factors influencing youths' migration by comparing migration decisions between return migrant and non-migrant youth. Following a classic study design by , we also compared findings in two sending towns in Mexico to gain insight into the role of the community context on youth migration decisions. --- METHODS This analysis was part of a larger qualitative study on migration and reproductive health among Mexican adolescents. The target population for this qualitative study was male and female youth ages 15 to 24 who had migrated to the U.S. and returned to Mexico, and, for comparison, an equal number of youth who had no migration history but lived in households in which a family member had migrated to the U.S. . We recruited only non-migrant youth who had at least one family member in the U.S. to control for lack of family ties as a reason for not migrating. In order to capture a broad range of migration experiences among youth, we recruited youth who self-identified as having migrated to the U.S. without any restriction based on length of stay in the U.S., reasons for migration, years passed since having migrated or returned to Mexico, or legal documentation status. Study participants from two communities in Mexico, one with a high and one with a low density of U.S. migration, were identified from community venues and through snowball sampling. These methods of recruitment enabled us to reach a broad range of youth, including both in-and out-of-school youth. To permit comparison between communities of differing migration densities, we chose communities from two states: Puebla and Michoacán. These two communities provide for a strong comparison because of their similarities on a number of economic indicators including per capita income, unemployment, and percent of youth attending school . At the same time, these regions have differing migration densities and histories. Puebla has a more recent history of migration whereas Michoacán is considered a traditional sending state . Tehuacán, Puebla, is a low-density migration municipality in which 1.9% of families receive remittances . Jiquilpan, Michoacán, is a high migration density municipality with 16.6% of households receiving remittances . The majority of empirical studies on migration in Mexico occur in traditional sending states, thus the inclusion of Puebla allows for a more diverse view of how these factors may vary by community-level migration context and tradition. Formative work was conducted to pilot-test study instruments and to interview key informants about community-specific migration dynamics. Community informants also guided our selection of recruitment locations, which included schools, church youth groups, stores and plazas. Data collection took place in October through November 2006 and was conducted by two bilingual females, a medical anthropologist from the National Institute of Public Health, Mexico and a public health professional from the University of California San Francisco, USA . We conducted purposive sampling to maximize variation according to high and low density migration areas and by the gender, age, and migration history of participants. Half of the participants were migrants and half were non-migrants with relatives in the U.S. . Sampling continued until informational redundancy and representation from each age, gender, community and migration history categories were reached. All youth participants provided written informed consent; parental consent was obtained for minors. Interviews took place in a private location of the participant's choice, including homes, plazas, cafes, and schools and lasted one hour. Semi-structured interviews elicited information on youths' personal and family histories, their decision to migrate or not migrate, migration experiences, attitudes toward risk and uncertainty, return migration, and future expectations. Interviews were digitally recorded and participants were compensated with a gift equivalent to $10. All study procedures received approval from the institutional review boards at both UCSF and INSP. Interviews were transcribed and analyzed in Spanish using ATLAS.ti . Transcripts were coded thematically by the two data collectors who developed a codebook based on themes that emerged from independently coding the first three interviews. All interviews were coded independently by the two researchers and regular meetings were held to discuss any new codes or discrepancies and changes to existing codes and meanings. For this analysis, quotes coded under migration decision, personal preferences, reasons for return, and future expectations, were outputted and stratified by migration history and community of origin to compare between return migrants and non-migrants. Memos and field observations were recorded and shared between the two data analysts and included in the analysis. --- RESULTS A demographic profile of the sample comparing return migrants to non-migrants is available in Table 1. The majority of youth interviewed was single and studying in high school at the time of the interview. Non-migrants were slightly younger on average and fewer had partners or children than return migrants. Non-migrants were typically among the younger siblings in their families and all were in school at the time of the interview. Among the 24 return migrants, age at migration ranged from 5 to 21, with a median of 14.4 years . Only a few had legal permission to enter and work in the U.S. One-third of migrants were considered involuntary migrants, meaning that their parents made the decision to migrate for their children. Nearly half of all migrant youth travelled with their parents. The majority of return migrants had migrated one time with the exception of six youth from Jiquilpan who made multiple trips to the U.S. . Time spent in the U.S. ranged from as little for two months for one youth who was caught by border patrol and admitted into a detention facility until a relative paid her bail so she could return to Mexico. The longest stay in the U.S. reported was 13 years for one participant who was taken to the U.S. by her parents when she was 7 years old. She was so accustomed to life in the U.S. that she did not realize that she was not a U.S. citizen until she applied for college. Almost all migrants traveled to California, though some resided in Arizona, New York, and Nevada. Time since returning from the U.S. varied from one month to six years. More than half of return migrants had been back in Mexico less than one year. A third of return migrants were studying at the time of the interview . The remaining youth reported that they were working or looking for work, caring for children or grandparents, visiting relatives, "hanging out", and/or awaiting legal documents. Seven of the return migrant youth had plans to migrate to the U.S. in the future, while two said they would probably return, though their plans were less definitive. Reasons for and against migration, as well as reasons for returning to Mexico, are listed in Table 3. Below we present in more detail the reasons for and against migration according to individual, family, and community level factors. We then discuss the primary reasons for return migration. --- Individual level factors The most common reason for migration echoed amongst almost all migrant youth was financial need and lack of local employment opportunities in their home communities. Migrant youth viewed the higher wages in the U.S. as a means to finance future schooling, build a house, start a business, and support their parents and younger siblings. As one male migrant from Tehuacán explained: The problem is that here you cannot do anything, buy a house or anything, and there [in the U.S.] you can. Why? Because … here you go to work a whole day and they give you 100 pesos [$10], which is what you would make over there in 1 hour. For males particularly, migration was a way to finance their own house in order for them to become independent from their parents and to form their own families. Migrant youth also were influenced by their peers who had returned to Mexico with tangible successes from abroad, prompting them to want the same material items and experiences. As one 17 year old migrant male from Jiquilpan expressed, "many here live with few resources and things, and so, seeing those people return in December with their new cars and all that, I think it impels them to go." Non-migrants, on the other hand, often did not face as much need to migrate; their jobs or remittances from family enabled them to continue living in Mexico. Unlike their migrant siblings, they envisioned opportunities for themselves in Mexico. As expressed by one female non-migrant from Jiquilpan: I did not leave because I was interested in continuing high school … I like my job and I like being in my country, and I'm not doing too poorly with my salary, and so for me, it is not this grand thing to go to the U.S. For non-migrants, migration decisions were intertwined with educational goals, especially for younger participants who expressed a strong preference to postpone migration until they completed school. Though non-migrants sometimes expressed that they desired similar material items as their migrant relatives, they expressed a willingness to wait. When asked under which circumstances they would migrate, non-migrants sometimes replied only if they were no longer attending school or if they performed poorly. Instead, they preferred to study and work in Mexico or wait until they could migrate legally. Comments from a non-migrant in Tehuacán clearly captured the differences between migrant and non-migrant siblings: I do not know what [my brother's] intentions were in going. He said he wanted money, a car, like he saw some of his friends with cars … I also [want a car], but its tranquilo, everything comes in time. A less common motivation for migration was curiosity; though this was more common for males than females. Stories from peers and family members abroad created a sense of curiosity among migrants to see the U.S. and to verify firsthand whether stories they heard were true. A 17 year old male migrant youth described his town in Mexico as dull compared to the U.S. where he imagined he would have a car and could drive around having fun. A female migrant from the high-density migration community, though she expressed fear about migrating without papers, explained how she decided to join her uncle in the U.S. because all her cousins were there and she wanted to experience what her cousins had seen. In contrast, fear was a frequently named reason for not migrating, especially for females. A common saying among non-migrants was that they would rather risk staying in their country even if it meant eating beans everyday than risking their lives to go north. Non-migrant youth sometimes gave the impression that their migrant siblings were different from themselves. For instance, one female non-migrant youth described how her brother "thinks extravagantly" and how he "just does things" whereas she thinks about things before acting. Another 18 year old non-migrant male from the high density migration community remarked about his migrant brother, "my brother comes and goes and many people ask him if he has papers and he doesn't; he says he likes to take risks." Many non-migrants intentionally chose to stay expressing little or no desire to 'conocer' the U.S. They expressed a deep connection to their towns and indifference, and in a few cases, distaste for what they imagined or had heard life was like in the U.S. A female non-migrant who had five brothers working in the U.S. said if she migrated someday, she would only go for vacation but not to live because of the way they discriminate against Mexicans in the U.S. Several other non-migrant females echoed this, saying that they would only go for vacation or to study, but not to work or live permanently in the U.S. --- Family influence For almost one-third of the migrant youth, the decision to migrate was not their own and was made by their parents. Younger participants were especially likely to have migrated with their parents who, from the youths' perspective, were seeking a better life and better educational opportunities for their children. Migrating as a family unit was more common in Jiquilpan where there was more access to fake or borrowed documents and legal means of crossing. The remaining two-thirds of youth made the migration decision on their own or did so in conjunction or with permission from their parents. One female migrated to be reunited with her husband who was working in the U.S.; however, nearly one-third of female migrants decided to migrate on their own. None had children at the time of migration. Having relatives in the U.S. was another draw for youth to migrate. Once one family member migrated, it became easier for others as a non-migrant from Jiquilpan explained: "My older sister marries her husband who resides in the U.S. and he takes her. Two years after, she takes my brother. Then another brother doesn't want to study anymore, and so he says, « Well, I'm going to the U.S. »." Many youth decided to migrate with short notice upon invitation from an uncle. A male migrant from the high density migration community captured this phenomenon in his remark: "The majority that go, they go because, it is tradition that one goes, their cousins go, their uncles go, and then the uncles pull the nephews and the nephews pull their friends." Though less commonly mentioned, changes in family circumstances or escaping family problems resulted in the migration of a few youth. One female migrant from Tehuacán never planned to migrate but, at the urging of her mother following her parents' divorce, dropped out of school and migrated eight days later with her uncle to finance her university studies. A male migrant, whose uncle came through on his way north, decided to join him because he was angry with his father. Another youth from Jiquilpan described how her dad never sent remittances. With four siblings in school and tired of seeing her mom suffer, she migrated with her uncle. Having relatives in the U.S. dissuaded some, especially those that were the younger siblings in their family, from migrating. Parents and siblings' migration sometimes made youths' own migration unnecessary, especially those who were the younger siblings in the family and received financial support from older siblings in the U.S. to continue their schooling. Others recalled their siblings' struggles in the U.S., which made them less apt to migrate. Several non-migrant youth shared that their parents and older siblings convinced them to stay in Mexico. They heeded advice from their parents to wait until they were older because it was too dangerous or to remain in Mexico to finish school. One male non-migrant's mother explained that he would make money once he finished school. His impression of migrant youth was that "many do not understand this, they just want money already and they think that over there they can get it in an instant." Several youth stayed behind to care for family members including their parents. Though they acknowledged that they could make more money in the U.S., it would not be the same to live without their family. While it was common for male youth to migrate to form their own families, there was a parallel norm for females with children to stay behind since migration was perceived to be too dangerous for young children. --- Community level factors Though the reasons for and against migrating were similar across study sites , youth's decisions reflected their communities' migration dynamics. In the high migration density community, Jiquilpan, all migrants had parents and/or grandparents who had migrated before them and benefited from their remittances . Jiquilpan youth had higher levels of education, more possibilities of migrating legally, and more social networks in the U.S. due to Michoacán's longer migratory history. Youth in Jiquilpan had easier access to 'coyotes' , who were often relatives or well-known and trusted in their communities. Many youth in Jiquilpan had access to fake or borrowed documents and were thus able to cross the border more easily through Tijuana. These decreased costs of migration in Jiquilpan enabled some youth to migrate for reasons unrelated to economic gains including for vacation, to visit relatives, to process immigration documents or because they had nothing better to do in Mexico. A few youth in the highdensity migration community migrated for more capricious reasons, such as attending a cousin's wedding in the U.S., as a quinciñera gift from relatives residing in the U.S., and to spend Christmas with relatives. The expansiveness of migrant networks in Jiquilpan made migration all the more inevitable. One youth who had never imagined migrating explained that she and many of her friends fell in love and married migrants whose lives were in the U.S., so they had no choice but to migrate. A few youth were expected by their parents to migrate in order to visit relatives. As one female youth from Jiquilpan explained, her mother did not know how to read or write so she dropped out of school to guide her mother to the U.S. In the high-density migration community of Jiquilpan, both migrant and non-migrant youth expressed a less romanticized impression of life in the U.S. Their reasons for not migrating focused on not wanting to live in the U.S., rather than on the dangers of the journey. For example, one male migrant from Jiquilpan stated that he preferred Mexico where he had his own bedroom than the U.S. where he had to share a room with his mother. Though he migrated regularly for work, he contended that he never wants to live in the U.S. permanently. In the low migration density community of Tehuacán, only half of the migrants had either a parent or grandparent who had ever migrated . They did not benefit from remittances of earlier generations and faced more pressure to migrate themselves to support single mothers and younger siblings. Fewer resources were available, thereby increasing the uncertainty and perceived risk of migration. All but two youth in Tehuacán crossed through more dangerous routes in the desert or river. Fear of the journey itself, including death, concerned youth in Tehuacán whereas fear of physical and emotional suffering was absent in discourses from Jiquilpan. Tehuacán nonmigrants had a saying that migrants went to "suffer on the other side." One male nonmigrant from Tehuacán shared, "I wouldn't risk my life … What happened with the guy [my brother's friend] who died scared me, and I said to myself, it's better not to go, or if I can pass normally [legally], it's better than to die there." --- Return migration A summary of the primary reasons for return migration are listed in Table 3. Reasons for returning were similar between the two communities and were often related to the initial decisions to migrate to the U.S., primarily economic and social motivations. About half of the youth who went because of financial need gave the impression that they had achieved their goals in the U.S. and were ready to come home. One 19 year old male from Tehuacán whose main motivation for migrating was to make money so that he could marry his girlfriend, returned to Mexico after two years because he was tired of working, had little time off, and missed his girlfriend. Six months after returning from California, where he resided with his parents and twenty members from his indigenous community, he was working in a factory and was engaged to get married. Another male migrant from Tehuacán also returned having achieved his migration objectives. While in the U.S., he had made $1300 weekly, of which he sent $1000 to his wife who used it to build their house. He described: "I went [to the U.S.] to build a house so I could live with my wife and child alone, apart from our family, and thanks to God, I built our house and bought land." In contrast to these success stories, a similar number of youth who migrated primarily for financial reasons returned home because they were not able to attain their goals. One young woman from Tehuacán travelled to the U.S. with her uncle and boyfriend with hopes of saving enough money to attend college in Mexico. Once she arrived in the U.S., she received little support from her relatives and became homeless. She moved in with a man who was a drug user so that she had a place to live and she soon became pregnant. She did not want her child to grow up in the poverty and drugs that surrounded her in the U.S., and chose instead to raise her child as a single parent in Mexico where she would have help from her mother. Two other female youth came back to Mexico to give birth so that they would have more family support to raise their children than they did in the U.S. Through comparison of the stories of youth who achieved their goals in the U.S. and those who did not, the importance of supportive social networks in the U.S. emerged as key to youths' success. For those reporting limited or no success in the U.S., support in the U.S. was frequently lacking and marred by drug use. This was especially true for youth who resided in the U.S. without their parents which was mostly youth from the low density migration community. Several youth became disillusioned with the U.S. saying that life there was more difficult than they imagined, particularly having to pay bills and rent. In fact, several mentioned that they did not ever want to go back to the U.S. A 22 year old male from Jiquilpan returned home after a couple of years and intended never to return. Even though he was unemployed and knew he would not make as much money in Mexico, he preferred to live in Mexico, because in his words and in the words of several other youth, life in Mexico was "más tranquilo" or calmer. Likewise, a female participant from Jiquilpan accompanied her mom to visit her siblings to see what their life was like in the U.S. After six months, having experienced racism and conflict with her extended family with whom she lived in the U.S., she and her mother decided to return home where "no one was telling them what to do." Just as several youth migrated upon short notice without planning, so did several youth return. Sometimes the return home resulted from a sudden and unexpected event that took place. For example, an earthquake caused one youth's family to become homeless, so after three years in the U.S. they returned to Mexico. Out of the seven youth who migrated due to their parents' decision, five also returned because their parents were returning. Death or illness of a family member in Mexico spurred several youth to migrate home though sometimes it was not the only reason for returning. For instance, a female youth from Tehuacán who had lived in the U.S. since the age of five said that her grandmother's illness was the impetus to return home, but she also mentioned that the house they were building in Tehuacán was almost completed. Additionally, two youth from Tehuacán returned earlier than expected because they were deported. --- DISCUSSION We conducted an in-depth exploration of the decision to migrate among youth who had migrated and returned to Mexico and youth from migrant families who had never migrated. Further, we examined reasons for returning among those with migration history. Many of youths' reasons for migration echoed previous findings among adults . Economic reasons of their own or of their parents were the primary motivation for migration. This was especially true for male youth and those from the low density migration community who faced more pressure to migrate since many of their own parents had not migrated. In the high density migration community, though employment opportunities and higher wages were a strong pull, social reasons like visiting relatives were also common. These findings echo those in a literature review on unaccompanied migrant children to the U.S., in which the top reasons stated for migration among repatriated children in Tijuana and Nogales included work and family reunification . At the same time, we observed aspects of the migration decision that may be especially pertinent to youth such as age, sibling order, educational aspirations, and risk preferences. Age and sibling order were strong determinants of migration and dictated how involved family members were in migration decisions. The strong relationship between educational aspirations and migration that emerged from youths' discourses in our study has been documented earlier . The observed higher risk tolerance among migrants relative to non-migrants has been alluded to in the migration literature but seldom tested and warrants further research. To our knowledge this is the first study to explore specifically why the majority of youth do not migrate. Previous research among non-migrants is sparse but suggests that those who stay behind may be motivated by the ties with family and friends in their community of origin . In our study, reasons for not having migrated varied widely. Several youth did not have a financial need to migrate or were not able to migrate because they were caring for children or parents. Other non-migrant youth wanted to migrate but were postponing until they were older, could go legally, or had completed school. Others expressed never wanting to migrate out of fear, because they had no interest, or to remain close to family in their communities where they believed they could be successful. The biggest difference observed between return migrants and non-migrants was the perceived value of education in Mexico and their willingness to wait for financial independence. According to the "culture of migration" theory, labor aspirations in the U.S. appear to compete with the desire to continue schooling in Mexico . In our study, migrants often chose the prospect of higher U.S. wages to achieve their dreams more quickly. Meanwhile, non-migrants communicated a desire to complete their education and succeed in Mexico making short-term sacrifices, such as putting off making money for a car and clothes, for long-term investments such as buying textbooks. Nonmigrants believed they could be just as or more successful in Mexico than their migrant counterparts in the U.S. Additionally, parents and older siblings' emphasis on the importance of education seemed to differ between the two groups. Non-migrants sometimes shared that they received counseling to finish school first. Migrant youth seemed to exhibit a more risk-friendly orientation which was consistent with non-migrants' perceptions of their migrant siblings as greater risk-takers. Migrant youth expressed curiosity and adventure in their motives to migrate. Meanwhile fear and a strong desire to succeed in their communities of origin were commonly voiced by non-migrants. Our study could not determine whether these different preferences observed between migrants and non-migrants are unique to youth or whether they operate similarly among adult migrants. Particularly for youth, because of developmental changes during adolescence, risk and temporal preferences may have important implications for the migration decision and for subsequent behaviors in the U.S. As in previous studies conducted among adult populations in Mexico, migration discourses were highly gendered . Male migration was more socially acceptable and male peers and relatives, especially uncles, played an influential role guiding younger males north . Females generally voiced fear and less desire to leave home and several did not want to work in the U.S. A few female migrants did not follow the traditional gender roles of migrating as daughters or wives Instead, they were the first in their family to migrate or went to join siblings already in the U.S. and were accompanied by a sister or comadre . For many, the decision to migrate reflected differences in perceived opportunities in Mexico and the U.S. With the increased uncertainty and risk of migration in Tehuacán, we might expect youth there to be more selected such that only those that had an extremely high risktolerance would migrate compared to youth in Jiquilpan where the migration context was more certain. However, for many Tehuacán youth, migration was the only choice they perceived to move ahead, whereas in Jiquilpan, migration was one of several options. It was also more common for non-migrant youth in Jiquilpan to perceive less opportunity in the U.S. and to have a more realistic perception of the struggles that immigrants may encounter in the U.S. Thus the threshold for risk-taking seemed to be influenced by what alternatives were available rather than merely individual preferences. This has important implications for immigrant youth who often settle in areas within the U.S. where opportunities are limited . Our study also documents the importance of youths' community of origin on the migration decision and journey. The differences in social capital between the two communities influenced with whom youth migrated and which routes they took. Research on acculturation points to the important role of pre-migration factors on the acculturation and health trajectories of immigrants in settlement societies . Studies examining adaptation among Mexican immigrants in the United States could be strengthened by including measures of pre-migration factors. A greater understanding of the context from which youth migrate may provide insight into the vulnerability of immigrant youth along the journey, to their adaptation in the U.S. , and their decision to return home. Reasons for return migration among youth in our study were similar to findings among adult migrant populations and were often linked to the initial reasons for migration . The same economic and social factors in their communities of origin that drove youth to migrate often influenced their decisions to return. For example, male youth returned once they had met their migration objectives to make money to buy land, build a house, or get married. Female youth who went out of curiosity to experience the life their cousins and siblings lived in the U.S., returned to Mexico once they became bored or did not get along with their relatives in the U.S. Several youth were forced to return due to family circumstances, such as when their parents decided to return or when a relative became ill or died in Mexico. Other youth were unable to 'salir adelante' as they had imagined and returned home. A supportive social network was a key factor in youths' ability to succeed and highlighted the particular vulnerability of unaccompanied youth and the importance of mechanisms for supporting young newly-arrived immigrants in the U.S. Reasons for return migration among youth have been neglected in the literature despite the fact that Mexican males between the ages of 25 and 29 have the highest rate of return . More research on the reasons for return migration among youth is warranted to provide insight into the selectivity of youth who stay in the U.S., as well as on the reintegration of youth who return home to Mexico. Recognizing the exploratory nature of this study, we are cautious in generalizing these findings because we interviewed a convenience sample of temporary migrants returning to Mexico who may differ from Mexican migrants residing in the U.S. However, a significant portion of undocumented Mexican migrants return home and do not settle permanently in the U.S. Return and cyclical migration is common and has important implications for both sides of the border . McKenzie, 2008 calculated that the median age of return migration from the U.S. to Mexico is 24 years, while the median time abroad for young return migrants from Mexico is three years. Due to the difficulty of recruiting representative samples of undocumented migrants in the U.S. , much of our knowledge of migration has come from studies that collect data from return migrants in their source community in Mexico . Thus, this study enabled access to a hardto-reach population often excluded from studies among youth in the U.S. . Furthermore, because interviews were conducted in Mexico, youth could share openly about their experiences with undocumented migration. --- CONCLUSION This study provides important information on the decision to migrate or not and the decision to return from the perspective of Mexican youth, a growing and yet understudied population. Not only does this study examine the personal factors associated with migration decisions among Mexican return migrant and non-migrant youth, but it also compares them between two sending communities in two states with differing migration densities. Both the number of youth migrating back and forth between Mexico and the U.S. and the rapidly growing Latino population in the U.S. warrant further bi-national efforts to improve our understanding of youth migration. Bautista is trained as a health economist. He received his BA in economics from the Universidad Autónoma Metropolitana in Mexico City and his Master of Health Economics at CIDE, also in Mexico City. His master's thesis evaluated the impact of Progresa/ Oportunidades on the utilization of primary health care services in rural Mexico. Sergio's current research includes a cost-effectiveness analysis to evaluate interventions for prevention and care of HIV/AIDS in Mexico as well as impact evaluations of health and anti-poverty programs. Hisp J Behav Sci. Author manuscript; available in PMC 2014 May 07. --- Biographies

We explored migration decisions using in-depth, semi-structured interviews with male and female youth ages 14 to 24 (n=47) from two Mexican communities, one with high and one with low U.S. migration density. Half were return migrants and half were non-migrants with relatives in the U.S. Migrant and non-migrant youth expressed different preferences, especially in terms of education and their ability to wait for financial gain. Reasons for migration were mostly similar across the two communities; however, the perceived risk of the migration journey was higher in the low density migration community while perceived opportunities in Mexico were higher in the high density migration community. Reasons for return were related to youths' initial social and economic motivations for migration. A greater understanding of factors influencing migration decisions may provide insight into the vulnerability of immigrant youth along the journey, their adaptation process in the U.S., and their reintegration in Mexico.

INTRODUCTION Among the social determinants of health, there is a great concern in regarding the effects of socioeconomic gradients in health 1 .More specifically, because socioeconomic disadvantages over the course of one's life may hasten poor health.In recent years, the effect ofchildhood socioeconomic position and how it may relate to health outcomes in adulthood have been investigated 2 . Systematic review has demonstrated associations between childhood SEP and an increased risk for coronary heart disease, stroke, and allcause mortality 3 . However, the mechanisms by which adverse social circumstances over the life course increase the risk of adverse health outcomes later on have not yet been fully explained. Social mobility can be characterized as the individual's trajectory within a social structure over their lifespan, and can be expressed through income, education, social class, or other forms of classification 4 . This type of mobility represents an individual's change from one category to another over a period of time. It is classified as "null" , "upward" , or "downward" 5 . Investigation into socioeconomic inequalities in health is a relatively young field in low and middle-income countries 6,7 . Brazil, even with advances in combating poverty, is still one RESUMO: Objetivo: Avaliar o efeito da classe econômica na infância e mobilidade social no crescimento linear ao longo da adolescência em uma coorte de base populacional. Métodos: Crianças nascidas entre 1994 -1999 em Cuiabá-MT, no centro-oeste do Brasil foram avaliadas pela primeira vez durante 1999 -2000 e novamente durante 2009 -2011 , sendo a estatura-para-idade avaliada nestes dois períodos. O Critério Brasil foi usado para classificar a classe econômica da família de cada criança como baixo, médio ou alto. A mobilidade social foi categorizada como mobilidade ascendente ou nenhuma mobilidade ascendente . Foram utilizados modelos lineares de efeitos mistos. Resultados: Foram avaliadas 1.716 crianças após 10 anos de seguimento, e 60,6% das famílias mostraram mobilidade ascendente, com uma percentagem mais elevada entre as classes econômicas mais baixas. Também foi observada uma maior altura-para-idade entre aqueles de famílias com elevada classe econômica, tanto na infância e adolescência , ao passo que a mobilidade ascendente não afetou o crescimento linear. Conclusão: Foi observada expressiva mobilidade social, mas a classe econômica na infância e a mobilidade social não apresentaram influência significativa na taxa de crescimento linear durante a infância nesta coorte do Centro-Oeste brasileiro. Palavras-chave: Nível socioeconômico. Mobilidade social. Crescimento. Estudos de coortes. Adolescente. Criança. of the world's most unequal countries, being the most unequal among the G20 countries 8 in regards to socioeconomic status and health. Between 1973 and 2008, a significant increase of social mobility was observed in Brazil 9 ; however, the greatest mobility has occurred in the last decade. Between 2001 and 2011, the 10% poorest section of the country experienced a cumulative revenue growth of 91.2%, while the richest part of the population hadan increase of 16.6% in the same period 10 . Changes in the economy and in public policies reduced social inequality in Brazil 11 . Growth is a widely used measure to gauge the quality of a child's environment because growth failure is linked to a broad range of adverse outcomes in life, and is related to later physical and cognitive development 12,13 . A positive association between SEP and height has been demonstrated in children in high-income countries as well as in low to middle-low income countries 14,15 , but few studies have examined the effect of SEP on height in adolescents 16,17 . Investigating the association between SEP and growth trajectories, however, requires longitudinal analyses of repeated height measurements. An investigation of this association in childhood would indicate whether the development of socioeconomic inequalities in adolescent or adult height could be partly attributed to inequalities in linear growth during this critical period.Therefore, this study aimed to evaluate the effect of one's socioeconomic position in childhood and of social mobility on linear growth through adolescence in a population-based cohort. --- MATERIAL AND METHODS --- SUBJECTS A cohort of children born between 1994 and 1999 in Cuiabá, the capital of the state of Mato Grosso, in central-western Brazil, was evaluated at primary health clinics from May 1999 to January 2000. Ten vaccination clinics were randomly selected, and the parents or guardians of approximately 240 children randomly chosen from each clinic were interviewed . All guardians who accompanied their children were invited to participate; the refusal rate was of 0.4%. The coverage in Brazil for the diphtheria, pertussis, tetanus vaccine at that point was 97%. A full description of the sampling plan had been described previously 18 . All the subjects enrolled at baseline were eligible for follow-up study at their schools between 2009 and 2011. In Brazil, approximately 95% of children aged 10-14 years and 78% of children aged 15 -17 years attend school 19 . The annual School Census in Brazil was used to identify the cohort;86.8% of the subjects and their schools were identified using the child's name, date of birth, and the mother's name. In addition, five deaths were recorded in the Mortality Information System. Further details are provided in the paper by Gonçalves-Silva et al. 18 . REV BRAS EPIDEMIOL JUL-SET 2017; 20: 514-525 Fieldwork was carried out between 2009 and 2011 in public and private schools in Cuiabá as well as in additional cities on one occasion in the follow-up period after 11 years; 14.4% of the baseline children had moved to other cities. One city, Várzea Grande, is geographically and economically integrated toCuiabá. The other subjects had relocated to 17 cities within the state of Mato Grosso and 5 capital cities from other Brazilian states . Information about the subjects' birth was obtained from hospital records, but all outcomes and the majority of thevariables during childhood and adolescence were measured by a trained team of nutritionists. A pretest of the research protocol and a pilot study were conducted in a similar group of parents and adolescents in order to test the understanding of the questionnaire, correct possible flaws, and standardize data collection. --- MEASURES AND DATA ANALYSIS At the first evaluation, when the children were under 5years old , information about demographic and socioeconomic characteristics was obtained by interviewing the parents or guardians. From 2009-2011, the subjects provided information about their socioeconomic and lifestyle factors using a pretested questionnaire. At both interviews, anthropometric measurements were collected by trained field workers according to the techniques recommended by Lohman et al. 20 . Using the anthropometric measures obtained at the first interview and during follow-up, the subjects' length/height for age and gender were calculated according to the growth curves published by the World Health Organization , and were expressed in z-scores 21,22 . The scores were calculated using the WHO Anthro program, version 3.1. The SEP of the families was based on the number of home appliances, cars, and paid maids, as well as the educational level of the head of the household according to the "Brazil Criterion" of economic classification 23,24 . This system has five classes: A , B, C, D, and E . For our analysis, individuals were reclassified into three SEP groups according to similar analyses by Muraro et al. 25 : low , medium , and high , according to the number of observations at categories. Maternal educational level was categorized as low , medium , and high . Social mobility was evaluated using the difference between the SEP of families from 1999 -2000 and from 2009 -2011, and it was separated into two categories: upward social mobility and no upward social mobility. We considered families with upward mobility as those with an increase of at least one economic class during the study period. Only 3.3% of the families suffered downward mobility, and then they were grouped with those who did not show social mobility, because both groups had no substantial changes in the results. REV BRAS EPIDEMIOL JUL-SET 2017; 20: 514-525 For the analysis, we considered the theoretical model of pathways by which maternal education might influence early childhood linear growth from Silva et al. 26 . The factors considered were:maternal characteristics like educational level, smoking during pregnancy, height ; and child characteristics like gender and birth weight . The mean z scores for length/height-for-age in childhood and adolescence were compared using the Student's t-test and analysis of variance . P values for trend were derived from the chi-square test for trend , or for the linear trend test using a one-way ANOVA. For the longitudinal analysis, the outcome was length/height-forage z-score. Linear mixed effects models were used to assess the effects of early childhood SEP and social mobility on height from birth to adolescence to account for the correlation between measurements 27 . The main effects of time and social mobility, and the interaction between social mobility and time were tested. A significant interaction provides evidence for a differential rate of growth in height over time. Regression residuals were examined graphically to assess the goodness of the fit. Analyses were performed using the Statistical Analysis Systems software package, version 9.3 . The study was approved by the Ethics Committee of the Júlio Müller University Hospital at the Universidade Federal do Mato Grosso . Parents or guardians of the participating adolescents provided written informed consent. --- RESULTS At the first evaluation, 2,405 pre-school children had their height and weight measured. Information regarding their length and weight at birth was obtained from hospital records for 2,394 subjects. After approximately 11 years, 1,716 adolescents were measured and interviewed . The average age at childhood was 1.5 years , while that at adolescence was 12.1 years . At baseline to 15.3% of children were classified as high economic position, 76.9% as medium and 7.8% as low. In the follow-up, 40.2% were classified as high and 59.8% as medium, therefore no children was classified as low economic class . Significant upward social mobility was observed in the study population , and mobility was more prevalent among families that started out in the lower economic class . Families with a low economic class showed a higher proportion of low maternal education, a higher prevalence of smoking during pregnancy, and a lower mean maternal height. A higher mean of length/height-for-age in both childhood and adolescence was observed among families of higher socioeconomic status . There was no difference in birth weight and height-for-age according to social mobility . Linear mixed effects regression, socioeconomic position at preschool age and social mobility had no impact on linear growth until adolescence . --- DISCUSSION The present study provides information regarding childhood SEP and also the effect of social mobility on linear growth until adolescence. In general, significant upward social mobility was observed in the study, mainly in families in the lower economic class. Children from families of a higher socioeconomic status at the first evaluation presented a higher mean of height-for-age in childhood and adolescence. The analysis revealed no significant impact of childhood SEP or of social mobility on linear growth until adolescence; those with a low SEP showed no overcompensation for their initial height deficit in comparison to those with a high SEP. Thus, it is possible that inequalities in adolescent health may already be determined in childhood and may be further accentuated by socioeconomic inequalities in behavior and educational attainment in childhood 16,28 . According to Barros et al. 16 , a worsening in SEP between birth and adolescence seems to also limit height, as individuals who were not poor at birth but became poor later on presented a mean height that was intermediate between those who were never poor and those who were poor at birth. However, it should be noted that most of the children in the cohort experienced upward change. While there seems to be a stronger relationship between a better childhood SEP and more positive health outcomes in childhood 29,30 , the relationship with social mobility remains uncertain. Some studies have found an association between social mobility and height in adolescence 16,17 , while others have reported a partial or weakrelationship 28,31 , ordidnot find any relationship at all. Barros et al. 16 analyzed the association of height with a change in SEP between birth and 19 years of age in a birth cohort from Pelotas, Brazil. Information on family income in 1982 and 2001 was used to classify the study population. Height showed a similar trend for men and women, with the never poor subjects presenting the highest mean, followed by those who were not poor at birth but later became poor. Those who were poor at birth, regardless of their later status, were the shortest subjects. Howe et al. 28 examined maternal education inequalities across multiple outcomes and how these inequalities changed across childhood and adolescence in the Avon Longitudinal Study of Parents and Children , a cohort born from 1991-1992 in southwest England . Those offspring with a higher maternal education level were taller, and there was evidence of a slight widening of socioeconomic inequalities along with height for ages between 11 and 15 years. Most studies have assessed just one or two measures of SEP at each stage of the life course 32 . Howe et al. 33 argued that there is no single ideal measure of SEP for all studies and contexts, and advised the use of that which is best suited to the specific research question. Our study evaluated the SEP of the families based upon the number of home appliances, cars, and paid maids, as well as the educational level of the head of the household, for the economic classification. This "Brazil Criterion" was not developed to consider health concerns and social welfare, but it has been used widely in the country and has an association with several morbidities and risk factors in the population 34 . Furthermore, the highest degree of upward mobility among those in the lower economic classes in the present study is similar to that observed in other studies using criteria other than the "Brazil Criterion" used here 16 . The social mobility in our study follows the same trend observed in national data 35 . This significant upward social mobility can be partly explained by the Bolsa Família Program . The BFP, which was created in 2003, is a cash transfer program for poor families that aims to promote immediate poverty alleviation through direct income transfers. In 2009, the BFP reached12.4 millionfamiliesand accounted for12% of thedecline in inequalitymeasured by the Gini Indexin recent years 36 . Attempts to measure the association between intergenerational mobility and health and risk behaviors among adolescents may present some methodological complications. According to Ritterman et al. 37 , household and parental indicators of SEP may be useful proxies for the social status of infants and younger children, but are less appropriate in assessing adolescent social positions and social and economic resources.Adolescents may already REV BRAS EPIDEMIOL JUL-SET 2017; 20: 514-525 have attained a different social position than that of their parents, such as a different educational attainment level or occupational grade. Instead, adolescent class identity may be more influenced by social processes associated with social position, such as national educational systems 38 , cultural norms and values, and future expectations 39 . Furthermore, adolescents are in a transition between being defined by their parents' social position and by their own, and it is possible that adolescent social status may be influenced by projections of a potential future socioeconomic trajectory 39 . The key strengths of this study are its use of longitudinal analyses of repeated height measurements and inclusion of the maternal height as a genetic proxy for adolescent height 40 . Maternal height represents influences from genetic and non-genetic factors, including intergenerational factors related to nutrition on growth, which may avoid the development of genetic height potential in low and middle-income countries 41,42 . One limitationof our study was the loss to follow-up ;the baseline characteristics of participants and those lost to follow-up were compared in a previous publication 43 , and the loss to follow-up was greater among adolescents who had low height-forage, mothers with less education and among those exposed to maternal smoking during pregnancy.However, few of them were lost due to a refusal to participate in the study, with most of the losses to follow-up occurring by family moving toother states 43 . Another limitation relates to the lack of assessment of the subject's sexual maturation. Furthermore, the information of economic class based onthe adolescent's report would be considered a limitation, but this criterion is generally used in epidemiological studies among adolescents. The results regarding economic disparities in health have been similar to that observed in adults and the elderly, such as the use of health services 44 . --- CONCLUSION In conclusion, up to adolescence, we found no evidence that differences in height due to SEP could improve with upward social mobility. Our results suggest that such inequalities in growthare established in early childhood. Other studies with methodological refinement to measure social mobility may contribute to future assessments of social mobility effect on height.

Objective: To assess the effect of socioeconomic position (SEP) in childhood and social mobility on linear growth through adolescence in a population-based cohort. Methods: Children born in Cuiabá-MT, centralwestern Brazil, were evaluated during 1994 -1999. They were first assessed during 1999 -2000 (0 -5 years) and again during 2009 -2011 (10 -17 years), and their height-for-age was evaluated during these two periods. Awealth index was used to classify the SEP of each child's family as low, medium, or high. Social mobility was categorized as upward mobility or no upward mobility. Linear mixed models were used. Results: We evaluated 1,716 children (71.4% of baseline) after 10 years, and 60.6% of the families showed upward mobility, with a higher percentage among the lowest economic classes. A higher height-for-age was also observed among those from families with a high SEP both in childhood (low SEP= -0.35 z-score; high SEP= 0.15 z-score, p < 0.01) and adolescence (low SEP= -0.01 z-score; high SEP= 0.45 z-score, p < 0.01), whereas upward mobility did not affect their linear growth. Conclusion: Expressive social mobility was observed, but SEP in childhood and social mobility did not greatly influence linear growth through childhood in this central-western Brazilian cohort.

Background Male circumcision reduces the risk of female-to-male transmission of human immunodeficiency virus [1][2][3]. Evidence from three large randomized controlled trials in Africa showed a reduction in HIV acquisition of up to 60% for heterosexual men [1]. Voluntary Medical Male Circumcision is promoted by the World Health Organization as a way of reducing HIV transmission [4]. In Papua New Guinea , a middle income country in the Pacific, HIV manifests in a concentrated epidemic which is largely heterosexually transmitted. Almost 0.5% of PNG's population live with HIV [5]. Some populations, such as transgender people in urban areas, have estimated HIV rates of up to 24% [6]. PNG faces many health and development challenges, one of which is how to respond to HIV. The acceptability and feasibility of male circumcision for HIV prevention in PNG was investigated during two large research studies to inform a national HIV policy response [7,8]. These studies revealed a diverse range of penile cutting and modification practices and documented a variety of opinions about the implications for women [9,10]. It is particularly important to understand the implications of male circumcision for women given the low social and educational status of many women in PNG [11]. The substantive area of enquiry for this transformational grounded theory study was to explore how women understand, experience and manage the outcomes of male circumcision and penile modification practices in PNG, including for HIV prevention. The aims of the study were to: 1. Explore women's understanding and experience of male circumcision and penile modification in PNG; 2. Describe and construct a theoretical model of the processes used by women to manage the outcomes of male circumcision and penile modification in PNG; 3. Identify the implications of results for local-level action, along with national HIV policy and planning in PNG. --- Methods The setting PNG is a diverse middle-income, Pacific Island nation of 6.8 million people, who gained independence from Australia in 1975. Over 800 languages are spoken in this hyper-diverse nation in the tropics, which has great cultural and biodiversity. The rural majority live a predominantly subsistence lifestyle. Alongside expanding extraction industries, there is a growing phenomenon of urbanisation with subsequent changes to social and cultural norms, and increasing violence [12]. For women in PNG, life can be challenging. Educational opportunities for many girls and women are limited, with approximately 50% of women in PNG considered literate [13]. While educational opportunities for girls are improving, currently only 12.4% of girls complete secondary school, which is half the number of boys who complete secondary school [14]. Maternal mortality ratio in PNG is the second highest in the Asia-Pacific Region and one of the highest in the world, with an estimated 733 maternal deaths per 100,000 live births . Of the 111 members of the National Parliament, only three are women [16]. In this study, we worked with women at two sites: a university campus near the national capital; and an oil palm plantation in rural PNG. These sites are culturally, socially and educationally diverse. --- Design of the study Transformational grounded theory methodology was used for this study. This methodology builds upon grounded theory by incorporating participatory action research strategies and decolonizing methodologies [17]. Consistent with transformational grounded theory methodology, people who participated in the study were considered co-researchers who helped to determine the focus, content and representation of the study findings. In Phase One of the study existing qualitative data from a large, mixed methods multi-site male circumcision study , conducted at four sites was theoretically sampled for data on women's understanding and experience of male circumcision. The data consisted of transcripts in English or Tok Pisin generated with 861 male and 519 female participants [7]. An analysis of this existing data was conducted by MRM to identify 'chunks of qualitative data' that could inform the development of codes and categories about women's understanding and experience of male circumcision [18]. Codes and developing categories were identified using constant comparison methods, supplemented by memoing [19][20][21][22]. In Phase Two of the study, the 'chunks of qualitative data from Phase One that described women's understanding and experience of male circumcision were presented to 64 female co-researchers who co-analyzed chunks of data during seven interpretive focus groups. Groups were facilitated by MRM and RT at an urban university and a rural oil palm plantation and were conducted in either Tok Pisin or English [13 interpretive focus groups]. The chunks of data were discussed by small groups of women in the interpretive focus groups in in a way they themselves determined. "This process evoked a sharing of personal experiences as women discussed their interpretation of the data, their personal positions in relation to the data, and shared much laughter as well" [18]. As a part of responding to, and analyzing the chunks of existing data, the 64 female co-researchers also contributed their unique experiences of male circumcision through guided oral discussions and drawing pictures on 'storyboards' [18,23]. Codes and developing categories were identified using constant comparison methods within each group, based on the oral discussions and pictures. Phase Two also included 11 semi-structured interviews which were conducted to further explore developing categories and ensure theoretical saturation. These 10 women and one man were identified as having specific knowledge about male circumcision, gender and society [20,24]. Seven of the 10 women had participated in interpretive focus groups, with three women participating in semi-structured interviews only. Oral discussion was recorded and photos were taken of pictures on storyboards. MRM analysed the storyboards, the transcripts, and handwritten notes in the language they were collected in and then used grounded theory data methods of analysis including constant comparison, memoing and initial, intermediate and advanced coding. These processes led to a developing transformational grounded theory [17]. In Phase Three of the study the developing transformational grounded theory, drafted in Phase Two, was presented to six discussion groups during a further round of fieldwork at the two sites . Co-researchers in these discussions were predominantly the same co-researchers who had participated in the original interpretive focus groups. The transformational grounded theory was presented and modified/clarified based on detailed discussions. In addition, the developing theory was presented to and/or discussed with over 100 other relevant stakeholders including health workers, company managers, and employees of non-government organisations. As a result, further modifications were made to the grounded theory that is reported below. --- Sampling and methods of analysis Phase One: Existing qualitative data from the mixed methods multi-site male circumcision study was theoretically sampled using grounded theory methods [19,20]. Data from transcripts of focus groups and semi-structured interviews were sampled . Semi-structured interviews and focus groups were theoretically sampled and analysed from the existing data set. Theoretical sampling was conducted by considering relevance of the existing transcript , richness of the transcript , and how the data "maximised opportunities to develop concepts in terms of their properties and dimensions, uncover variations, and identify relationships between concepts" . Phase Two: Primary data were co-generated with 67 women and one man all from PNG. Co-researchers were theoretically sampled depending upon their characteristics . Co-researchers in Phase Two were from two of the four original study sites . Phase Three: a series of six feedback discussions with co-researchers at two sites were facilitated to present the developing transformational grounded theory. During these discussions the grounded theory model was presented, feedback provided by co-researchers about what resonated in the theory and what minor changes needed to be made. In addition, possible local and national level health promoting actions were identified. Consistent with grounded theory dictum, 'all is data' [26], the only published article that had focused on women's experience of male circumcision in PNG [10], was also included in the dataset as extant data. The process of fracturing, analysing and integrating data was repeated, with codes from the semi-structured interviews, interpretive focus groups and the article by Kelly et al. [10] enfolded into codes already developing out of the initial analysis. Analysis of the data was conducted by the lead researcher , supported by JM and RT. Using the qualitative software NVivo, 162 codes in both English and Tok Pisin were identified from the combined dataset . These codes were then combined into 93 sub-categories and finally grouped into the core category and the four other categories of the transformational grounded theory. --- Results Many women know a lot about male circumcision and penile modification. Many women can describe the procedure of circumcision, including the variety of cuts that occur, the local names of the cuts and how they differ from each other. Names women use to describe male circumcision include raun kat which is the circumferential male circumcision and stret kat or split kat the longitudinal cut or dorsal slit of the foreskin. These types of cuts have a variety of local names such as kela , kaen blong Sepik , V-cut and banana split [27]. Women know the range of ages that boys and men are circumcised. Women report traditional circumcision conducted on boys from a few days old up until they are young men of 18-20 years of age. The preferred age for circumcision depends upon the cultural setting, the resources of the family who prepares gifts and food to host the feast after the circumcision and where the young man is living. Peer influence can result in circumcision of school boys, university students and older men. Women report male circumcision occurs in a myriad of nonclinical settings, including: tambu haus , haus man/haus boi , by the sea, by the river, under oil palms, in the forest, at school, in dormitories, in school grounds and in the settlements in urban areas. Some women report circumcision occurs in clinical settings such as clinics or a hospital, but this is much less common. A female university student said, "They boiled the razor blade, you know the razor blade, they boiled the razor blade and then they removed it and then they did a slit, but we were not allowed to know about it. We knew about it but were not allowed to go near the boys yeah, the boys stayed in the room and we stayed away from them " Semi-structured interview ). It is not uncommon for primary school and high school boys to cut each other's foreskins. This can result in serious injury and family disharmony. A group of women drew a picture on a storyboard that illustrated this type of informal cutting amongst students . Women know about male circumcision and penile modification from a variety of sources. They know because they are sexual partners, participants in community feasts, through observation, health practitioners and family members. Most women find out if their male partner has been circumcised once they are in a sexual relationship with them. One woman explained, "Only their wives will find out or whoever is having sexual relationship with that man" SSI). Another coresearcher said, "Basically the girl will notice during, like when they having sex and then the boy tells the girl" . Women who are nurses or health workers do not usually perform the circumcision, but they are often asked to provide bandages, gauze and medications and assist boys and men. In addition, co-researchers reported sharing sanitary napkins to assist boys in the management of bleeding post penile cutting in a school dormitory. Some women know about male circumcision because they are family members. When a man has been circumcised, women are often asked to prepare less food or different types of food such as dry kaukau so that the young men do not have too much fluid, thus reducing the need to urinate. "Ol bai stopim kaikai, stopim wara ol disp'la kain" [They will stop eating, stop drinking water or things like that] SSI). The preparation of special food often falls on the mother. Some women are happy for men to be circumcised, especially if they come from traditionally circumcising areas. They see circumcised men as manly, capable of caring for a family, strong and healthy. They are able to take their rightful place in the community. Others think circumcision is healthy or consistent with religious teachings. However, if circumcision is not a traditional cultural practice, women think men are motivated to be circumcised for different reasons. Increased attractiveness to women is frequently cited as the main reason men from non-traditionally circumcising areas get circumcised. "Taim ol katim…em isi lo kisim ol liklik yung gels tu, kain olsem" [when they get circumcised/cut… it is easy to attract young girls for intimate or sexual relationships, something like that] ). "So why do boys do the circumcision? Basically to attract the women, in a sexual way…the trend is that boys Fig. 1 The transformational grounded theory model have a lot of girlfriends. If you have a lot of girlfriends 'yu man' …so they circumcise themselves and the more women they sleep with, so it kind of, they also taken on the health side, but psychologically they want to be want seen as tru man " . Women have mixed views of male circumcision but unanimously oppose penile modifications, such as inserting objects or injecting fluids. Women describe men inserting ball bearings under the skin of their penis , and inserting parts of toothbrush handles and other objects in the wall of their penis. One coresearcher from the oil palm plantation explained her response to these practices, "So mi bin go againstim displa, lo olsem katim stret na lo sait blo health or displa kain safety em orite tasol, lo somapim bearing na wanem kain ol samting em ino gutpla" [I am against these practices, if it is like a straight cut for health reasons or this kind of safety, that is alright, sewing in bearing or something like that is bad] . Some women describe men injecting fluids into their penis. These women knew details about how long the substance stays in the wall of the penis, how hard the penis becomes and the type of damage this causes women when men have sex with them. Some coresearchers had assisted women in their village to seek medical help due to the excessive bleeding that had occurred after sex with a man who had injected fluid into his penis. Women know a lot about male circumcision and penile modification, despite it being a culturally taboo subject for them to discuss in public or with men. Women are often asked to take active, supporting roles when men are circumcised. Women know about the consequences of male circumcision or penile modifications for their relationships, including a potential change in the nature of their sexual relationship, or a perceived risk that men will increase the number of sexual partners they have. The level of detailed knowledge women have about male circumcision and penile modification raises questions about the nature of gendered, taboo knowledge. Thus a property of this category is the taboo nature of male circumcision. For women who come from traditionally circumcising areas, their fathers, mothers or other family members explain that circumcision is a taboo topic. "Cultural reason yeah… if I'm openly talking with my partner then I'm disregarding culture, because they don't talk about it" . Co-researchers reported negotiating with their husbands, brothers and uncles to speak to researchers about male circumcision, with one saying, "When I asked them , I said I'm going for an interview with this lady, so they felt sorry for me and they said "oh, it's okay" for me to tell you." . For some women, just keeping safe is their focus and they keep quiet about what they know. There are potentially negative consequences, perhaps even to their own safety, if they speak about this men's business to the 'wrong' people. It is from this data that the First Category of the grounded theory is: Women Know a Lot. This is the Base category and represents the plethora of knowledge women have about male circumcision and penile modification. Despite these activities being seen as Fig. 2 Storyboard picture of school boys cutting each other's foreskins under banana trees at the back of a school building 'men's business' , many women know when, where, how, by whom, costs, medications required and many other details about male circumcision and penile modification. Women know because they are wives, girlfriends, sisters, mothers, health workers and/or teachers. Although women know a lot, they are careful with this knowledge. Speaking about male circumcision and/or penile modification outside of accepted social or gendered groups can result in a threat to women's safety. From this Base category, we now explore how an increasing knowledge can influence women's experience of male circumcision and penile modifications. The second category of this transformational grounded theory is Increasing Knowledge. Women report that if they have more knowledge such as a formal education or informal training they have greater status in their community. This change is almost always positive for the woman and her family. It affords women more options and greater autonomy. She is more likely to be a person in the community that women go to if they have a problem. An educated woman has opportunity to move beyond usual social and cultural norms and mores. When discussing male circumcision in an interpretive focus group, one woman with a qualification in the agricultural field was referred to as the woman who had a pepa -a qualification. This woman was deferred to by other women during the group. Deferring to a leader known in PNG as a big man or big meri is the norm. PNG is comprised of hundreds of collective societies, each based upon values of group benefit. Important as a research finding is how formal education or training is highly valued by community members, and with this valuing comes possibilities not available to women who have not had educational opportunities. Educated women in PNG are invited to be experts in not only their field of expertise, but also in many other aspects of community life. Women have to be committed to undertaking education or training as individuals, but they also require the support of others in their family and broader community. A dimension of this category is the collective support required for a woman to participate in education or training. Participation in education or training requires both individual commitment and the approval and support by men and other women in the community. This includes approval by a woman's husband if she is married, or approval by a woman's father or brothers if the woman is not married. When a woman participates in training, other women will take on some of the responsibilities of the trainee, and thus there is a community-level impact. Women who train as volunteers, for example for Tingim Laip , must assess the benefits and risks of participating in informal education. One of the benefits of women participating in education and training is the increased number of options that result. Specifically, increasing knowledge results in a change in the way women understand and accept male circumcision. One co-researcher explained, "…mi no bin hamamas bipo taim mi bin lainim dispela samting…but bihain gen mi bin go insait long wok blong Tingim Laip, ol tok 'dispela em safe, dispela em safe na ol workim em safe so yupela no ken kross." [I was not happy before I learnt about this …but after I began the work with Tingim Laip, they told us 'this is safe, this is safe, they can do this safely so don't get angry'] . Women who have a formal education or some training opportunities have greater status in their community. This increased status brings a greater number of options for managing the implications of male circumcision and penile modifications, and also increases the range of choices in other areas of a woman's life. The third category of this transformational grounded theory is Increasing Options. If a woman has an education or some form of training her status in the family and community is increased and she will have more options. These options will be about whom she can have relationships with, the nature of those relationships, her power in those relationships and who she can influence. A woman with an education will have more choice about with whom she has intimate relationships, whether a man's circumcision status will influence that decision and whether to make a decision about male circumcision for her sons. As one educated co-researcher explained, "We can say if we want" . The increasing options available to trained and educated women lead to feelings of confidence and a plan to act. "I wasn't aware of that, now that I'm aware, believe me, I'm going to have all my boys in my family, I will do it and I'm gonna do it. I will have to get them circumcised, that's it." . Increasing options become available to women with increased knowledge and an enabling environment. A co-researcher was asked if women have influence in their families to give information and help them make healthy choices. The co-researcher responded, "Some would, some might not, depending on different factors. Maybe their education levels, back to the cultural, how men see women in their society and stuff" . However, it does not always follow that once a woman knows about her options that she can act. The fourth category of the transformational grounded theory is Acting on Choices. The women who have increased knowledge, increased status and an increased range of choices can act on their options. They can act to keep themselves and their families safe, choose healthy sexual relationships, choose the best for their male children and discuss and agree upon options with their male partner. Some women make decisions that are not culturally sanctioned or are against family wishes, while some women convince their husbands to act for themselves, for their young or teenage sons to be circumcised. Only a few women take this type of action. For women to act, they need to be convinced about the benefits of the action, have power to make a decision, have resources if the decision brings negative consequences and/or have the support of family. If a woman is educated, male circumcision for her husband or children is more likely to be a joint decision between her and her partner. "They can do things to satisfy their boyfriends and husbands, like when they are, when they become their husbands wife, they need to satisfy all of the needs that the guy has so instance of sexuality they can also satisfy the need, then the guy to respect the woman, the woman also has rights to when to have sex and when not to have sex" . Women describe a dimension of risk associated with moving beyond social and culturally sanctioned roles, particularly when it comes to sexual health issues, such as male circumcision. If someone takes offense at a woman's involvement in 'men's business' , whether that be her partner, her sons or other family or community members, there can be serious physical, social, cultural and economic costs for the woman and her family/community. Compensation can be requested of the woman's clan and she can experience isolation and a reduced status in the community. Women without the increased status afforded by an education are most likely to leave the decision about male circumcision to their male partners and son/s, even if they think male circumcision is a good thing. Women report not wanting to face negative consequences when their son/s grow up. Some women fear their son/s may 'blame' them for being circumcised. If a circumcised male were to grow up and object to the decision of his mother to have him circumcised, there could be serious social and cultural consequences as adult children in PNG are responsible for providing every aspect of care for aging parents. An older woman said she would not want to have her son circumcised in case he was unhappy and therefore not care for her later in life ). One woman explained, "An infant should not be …it should be done to those big enough to understand what's going on with them so they themselves should decide what's done with them" ). Women who take action to have their sons circumcised, counter to cultural and social norms, usually rely on their knowledge about health and HIV transmission to make the decision. There is both an individual and collective dimension to this category. Women in PNG experience extreme rates of family and sexual violence [28] and this influences many dayto-day decisions a women makes -where she walks, how she speaks to her partner or family members, to whom she says yes and no. Women often walk together to ensure safety. As one co-researcher described, "Walkabout wantaim sumpla narapla, tasol walkabout wanwan em i no sef" . If a woman does not act according to societal norms, her personal, cultural and/or economic safety can be at risk. Such behaviour can also affect her connection to her religious community, if she acts in a way that is not 'acceptable'. A woman's future ability to act on choices may be threatened by her partner, his family members of her partner or members of the village/community. The seeming trajectory of the transformational grounded theory model, as represented by the large red arrow , does not always go in one direction. If a woman's safety is compromised, she may return to the place she knows she has increased options but not be able to act, as represented by the red arrow between Acting on Choices and Increasing Options. A woman with knowledge may also know she has increased options but may feel she needs more education or training to increase her legitimacy to act, as represented by the red arrow moving between the categories Increasing Options and Increasing Knowledge. The 'core' category of this transformational grounded theory is Power of Choice. Power of Choice is the overarching category that represents the central phenomenon present in each of the other categories [19]. As the overarching category it brings together concepts which explains how women understand, experience and manage male circumcision and penile modification in PNG. Power of Choice includes dimensions of both individual and collective power. A person who has the ability or freedom to direct or influence outcomes for themselves and others in the context of family or clan has power. The core category of "Power of Choice" for this transformational grounded theory, is an in vivo code gifted during a semi-structured interview with a young female student at PAU . This core category has two key concepts, Power and Choice, that when combined, creates a phenomenon that encompasses the four categories and intervening condition of the theory. Choice is both an individual and collective experience, reflecting the cultural context in which the women in PNG live. The concepts of Power and Choice contextualise women's reported experience of male circumcision and penile modification. Women with more education have more power of choice. Power enables an ability to influence and make decisions. Women require both knowledge and an enabling environment to have the power to make and act upon a decision. Choice refers to a woman's power to choose between the possibilities available without fearing negative consequences. A key dimension of the core category is the collective nature of cultural and social organization in PNG. The experience of collectivism is encapsulated in the phrase, "I am not the child of my mother, I am a child of my tribe" . This predominantly collective society is organised around a context specific, reciprocal wantok system. The majority of the over 800 language groups in PNG are patriarchal, which defines the nature of relationships between women and men, women and their male children and between women. This context influences how individual women can or cannot make decisions and how women may experience and/or influence the decisions of men or other women. Western assumptions about the nature of a motherchild relationship, with related assumptions about women being able to decide about the circumcision of their male children rarely hold in PNG. --- Discussion This transformational grounded theory explains how coresearchers understand, experience and manage male circumcision and penile modification in PNG. The theoretical model centralises the core category, Power of Choice, while connecting categories of Women Know a Lot, Increasing Knowledge, Increasing Options and Acting on Choices . Properties and dimensions of each category are represented in the model and anchored in the text, with evidence from co-researchers. Strauss and Corbin describe intervening conditions as broad conditions that bear upon action [25]. Safety of women is an intervening condition that affects all categories in this transformational grounded theory. Safety contextualizes the overarching lived reality for women in PNG, invites the inclusion of men in the transformational grounded theory model, and explains relationships between men and women in this study. Encompassed by the core category of Power of Choice, the four categories of this transformational grounded theory explain how women understand, experience and manage the outcomes of male circumcision and penile modification in PNG. We locate this grounded theory in international health literature, by identifying 'social determinants of health' [29] as the theoretical code that increases the explanatory power of the grounded theory. Glaser explains theoretical codes conceptualise how codes relate to each other when integrated into a theory. Theoretical codes are also emergent "they weave the fractured story back together again" . In this grounded theory, the theoretical code of 'social determinants of health' is applied in order to contextualise the transformational grounded theory, which is focused on local understandings of male circumcision to inform health promoting action, including HIV prevention. As a theory of public health, social determinants of health provides a structural explanation as to why people experience different health status when living in the same communities [29]. The ten social determinants of health identified by Wilkinson and Marmot are: the social gradient; stress; early life; social exclusion; work; unemployment; social support; addiction, food and transport [29]. As shown in the reporting of this grounded theory, many women in PNG endure negative aspects of these social determinants of health. While many women know about male circumcision , including for HIV prevention, there are impediments such as a lack of social status in a predominantly patricidal society. There is also individual and community-level stress due to continued exposure to violence [31][32][33], and in a collective society, potential social exclusion for taking individual health promoting actions outside of social, cultural and religious norms [10]. These impediments also include a lack of opportunity to participate in formal education , a reduced ability to explore options when compared to their male counterparts and reduced opportunities to act upon health promoting decisions , with a sense of safety . This sense of safety is critical to healthwith peace referred to as a fundamental pre-requisite for health in the Ottawa Charter [31]. A social determinants frame assists us to weave greater explanatory power into this grounded theory. It also has the potential to inform and transform local level action, holds possibilities for national health policy action in PNG and suggests lessons for international action for HIV prevention. As Dean et al. explains, effective action on the social determinants of health: "requires having sufficient knowledge of the mechanisms influencing health inequities and adopting a conceptual framework that not only clarifies the relationship between social determinants and health inequities, but also helps to identify entry points for intervention" . Results from this study will inform policy makers of the potential risk of negative, unintended consequences for women that may result from proposed health prevention programs, such as mass male adult circumcision programs. --- Movement towards action Transformational grounded theory draws upon participatory action research approaches and includes an action component that is conducted in a partnered and power sharing manner [17]. Although the study concluded, health promoting action continued. As a result of recommendations from women during the study, brief intervention sexual health training was facilitated for supervisors and employees of the oil palm plantation by a senior sexual health practitioner [34]. Health workers working with the oil palm company, along with clinicians from provincial health services, were also provided with a two-day course to discuss implications of male circumcision, penile modification, sexual and relationship issues [34]. In addition, three other research projects were planned with funding sought to enable the health promoting work. This research continues to create opportunities for change at the local level. The strength of this grounded theory is that it has been developed with co-researchers to not only understand the phenomenon, but to plan action for change. --- Limitations of this study Limitations of this study include limited time of the lead researcher in the field and its potential impact upon the two-way approach to research. This was mitigated to some degree due to previously established relationships with research partners while working on the large multi-site study, previous time spent living and/or working at the field sites and continued research with RT, a senior researcher from PNG who has since commenced her own PhD in the field of male circumcision for HIV prevention. The limited time at the sites during the research also impacted on the way the grounded theory method of theoretical sampling was enacted. Time at each field site was often approximately one week. Sampling challenges were encountered as time was limited, however these were somewhat overcome by the support of research partners who negotiated with co-researchers with particular characteristics to participate in the study prior to the researchers re-entering the field. --- Conclusions Women in PNG know a lot about male circumcision and penile modification, despite it being 'men's business'. Women who have had opportunities for education have a greater range of choices and an increased opportunity to act upon these choices. However, women can only exercise their power of choice in the context of safety. This research has shown that women's ambivalence previously reported about male circumcision occurs in a context of lack of safety, one expression of gender inequality in PNG. Women who feel safe are able to make choices and experience less ambivalence about male circumcision. The use of the transformational grounded theory methodology has provided a means to understand how women experience and manage male circumcision and penile modification in PNG, including HIV prevention and has identified opportunities for health promoting action. --- --- Abbreviations HIV: Human immunodeficiency virus; PNG: Papua New Guinea Authors' contributions MRM: Conceived the study, co-generated and analysed data, drafted and edited the manuscript; JM: Supervised the research, provided important intellectual content and edited the manuscript; RT: Co-facilitated the fieldwork, provided important intellectual content and edited the manuscript; DM: Led the original male circumcision study, provided important intellectual content and edited the manuscript; RS: Co-supervised the research and edited the manuscript; WJHM: Provided important intellectual content for the original male circumcision study, co-supervised the research and edited the manuscript. All authors read and approved the final manuscript. --- --- --- Competing interests The authors declare there they have no competing interests. ---

Background: Male circumcision reduces the risk of female-to-male transmission of human immunodeficiency virus (HIV) and is being explored for HIV prevention in Papua New Guinea (PNG). PNG has a concentrated HIV epidemic which is largely heterosexually transmitted. There are a diverse range of male circumcision and penile modification practices across PNG. Exploring the implications of male circumcision for women in PNG is important to inform evidence-based health policy that will result in positive, intended consequences. Methods: The transformational grounded theory study incorporated participatory action research and decolonizing methodologies. In Phase One, an existing data set from a male circumcision study of 861 male and 519 female participants was theoretically sampled and analyzed for women's understanding and experience of male circumcision. In Phase Two of the study, primary data were co-generated with 64 women in seven interpretive focus group discussions and 11 semi-structured interviews to develop a theoretical model of the processes used by women to manage the outcomes of male circumcision. In Phase Three participants assisted to refine the developing transformational grounded theory and identify actions required to improve health. Results: Many women know a lot about male circumcision and penile modification and the consequences for themselves, their families and communities. Their ability to act on this knowledge is determined by numerous social, cultural and economic factors. A transformational grounded theory was developed with connecting categories of: Women Know a Lot, Increasing Knowledge; Increasing Options; and Acting on Choices. Properties and dimensions of each category are represented in the model, along with the intervening condition of Safety. The condition of Safety contextualises the overarching lived realty for women in PNG, enables the inclusion of men in the transformational grounded theory model, and helps to explain relationships between men and women. The theory presents the core category as Power of Choice.

Introduction In response to the U.S. election in 2016, many artists took it upon themselves to use their art as a form of resistance. Embodied practices of justice crystallized as activist performance, direct action in the streets, and community engagement in many forms. Although dance and movement have a long history of resistance, the emergence of previously hidden blatant, unapologetic racist rhetoric has brought a deeper-and perhaps more embodied-understanding of the threat of oppression. --- Dialogue Rose: Artists Coalition for Change Together was born amidst the complicated political terrain of the Trump era. Originally founded by the two of us and Heather Roffe, ACCT became an organization of progressive collaborators that sought to generate social change through advocacy, dissent, resistance, facilitation, and healing. Consistent with those dialogical practices, we have written this paper as a conversation that highlights "ACCTions" designed to face the multiple reckonings of the Trump presidency. Donna: Rose, you said ACCT was born; how did that happen? Rose: I had a physical response to the 2016 election: fear appeared as constriction in my body, sadness revealed itself as paralysis, and anger emerged as the fight activation in my psoas. After years of arts activism, the moment called for more. I cast a wide net, you and Heather responded, and thus began our collaboration. Despite working full-time, we reallocated our personal resources to move into action through art. Rose: Looking back at the four years of ACCT, what patterns emerged? Donna: As dancers and choreographers, we were passionate and a bit naïve when we started ACCT. We spent weeks contemplating the right title and its acronymwho and what would comprise the scope and purpose of the coalition. In hindsight we realize we focused on three goals: 1) anti-oppression, 2) radical joy, and 3) connecting communities across differences. Driving our organizing was commitment to change and the action it takes to make progress. These three outcomes, while not our initial mission, are revealed and prominent within the ACCT Chronicle , which lists 50 "ACCTions," including organizational meetings, community workshops, performances, protests, rallies, parades, art installations, and artist gatherings to create work. What began as an intellectual, political response became embodied as direct community action to drive cultural change. Highlights in Rochester, NY, included a curated dance performance, "Stop Motion/Start Action," at the Multi-use Community Cultural Center during the Fringe Festival; an ensemble dance and vocal homily at Spiritus Christi Church; a community workshop titled "Subtracting the Divide," which led to a performance of movement and spoken word; a "Families Belong Together" rally and art installation at Washington Square Park; a multi-generational outdoor pre-parade movement improvisation at Memorial Art Gallery to celebrate pride in the LGBTQI+ community; and two versions of "The Reckoning," which are described below. All these creative, communal activities were embodied by its members as activism. They were acts of protest, and yet we recall not only the rage and conviction but also the joy of community connection, the collaborative spirit that drives people to march together, to stand in solidarity in freezing cold and scorching heat, to join a very long history of marches against oppression in the USA. Donna: I have heard the term "Social Somatics" and I'm not sure what it means in relation to dance as protest. Rose: Like Somatics, Social Somatics integrates the body and mind, focuses on perception and awareness, and acknowledges the body's intelligence as it relates to the environment it moves within. It deepens the belief in the social impact on and of our bodies. If social systems influence the embodied lives of the individual, then people shape the body of culture. Drawing upon ancient eastern philosophies and healing practices, Somatics has long been part of the dance and movement field in the U.S. More recently, embodied practices have begun to emerge in social justice circles using Social Somatics practices as a path towards justice. These practices are spearheaded by people like somatic educator Martha Eddy; Resmaa Menakem working in somatic abolitionism; Dr. Sará King, integrating neuroscience and mindfulness; and organizations like Generative Somatics that focus on the body, social change, and climate justice. When we consider the holistic implications of Social Somatics as it relates to protest, it is quite powerful: every human being has a body that brings them into relationship with the world. When inner intentions are aligned with outer expression, our bodies become vehicles for acts of resistance. Rose: Speaking of embodied activism, didn't you teach a college course on embodied activism in the spring? Donna: Yes, I did. As a co-founder of a social justice studies college program, I was influenced and encouraged by Nkem Ndefo and Rae Johnson, who taught an Embodied Philosophy course in July and August 2020 titled "Embodied Activism: Navigating the Intersection of Embodiment and Social Justice." I joined hundreds of participants on-line to re-think social change, trauma, liberation, and resistance and to connect our learning to future embodied actions. When I designed this community arts course to focus on embodiment as an expression of inner conviction to drive change, I was thinking of the semester as a journey to deepen awareness, gratitude, and resilience, especially during a pandemic when students had to be at least six feet apart and personal safety existed within the virtual world. "Embodied activism," a familiar term in the early 21st century, signals the centrality of human integrity and passionate conviction. Rather than being triggered intellectually by political ideology, activism that is embodied starts from a somatic, internal site within the body. Absent the need for a witness, embodied direct action is contrasted with performative activism, which my students recognized as "being seen" protesting, a flawed definition of activism. By the end of the course, students shared performances and writing that expressed changed identities in relation to activism, drawing a connection to their own embodied experiences. Rose: Also as an educator, I was able to include my classroom by designing both of my Arts + Activism courses around ACCT's community art piece, The Reckoning. I had hoped to encourage future change makers as my students worked beside the seasoned activist-artists of ACCT. Donna: Because we created two "Reckonings" before major elections, modeled around notions of "dancing the vote," do you believe these live events became examples of arts production for social change? Rose: The Reckoning was an example of how movement and performance can be used to activate audiences. Conceived collaboratively, the production emerged out of fear of voter apathy and a desire to activate potential voters. In both iterations , collaborating artists crafted interdisciplinary performance pieces that addressed "hot button" issues on the ballot, which included immigration, healthcare, LGBTQIA+ rights, gun violence and more. The 2018 performance was set up like a house of horrors that ushered the audience through unpredicted spaces, each installation finding new ways of provoking engagement. In one movement installation on the topic of immigration, the audience became the performers. Cast members, dressed as immigration officials, asked questions from the U.S. citizenship test, allowing the audience to move forward only if they answered historical questions correctly. The success of 2018 led us to carry The Reckoning into 2020. Although the second iteration required a different format due to COVID, we had the opportunity to re-envision aspects of the event. With pandemic restrictions in mind, our central questions became: How do we safely engage the most diverse audience? Can we reach audiences that may not seek art as protest? How do we encourage a wider range of voter engagement? While the original intent of The Reckoning was preserved, the 2020 version could not have been more different. We extended our run of performances by producing an event for four weekends leading up to the election. We exploited Covid restrictions to generate creative problem solving, ultimately traveling to four different neighborhoods in Rochester. Each outdoor performance was unique: one dance piece with spoken word streamed fully online; one open air art installation on a college campus; one interdisciplinary movement and film event projected onto the side of a building; and one culminating event with live dance, music, and visual art outside a local art gallery. Regarding the possibility of arts production for social change, ACCT was founded upon the assumption that interactive elements encourage audiences to interrogate their own beliefs and histories. Acting in an embodied way is likely to have more impact and change in the body is more likely to sustain intentionality and translate to action. Donna: How has this journey with ACCT influenced you as a person? Rose: Working with ACCT gave me the opportunity to claim my activist identity, something I did not do within the dance world. The invisible code that discourages being "too political" within the dance field was imprinted upon my own body and art making. Working with ACCT has caused me to become more aware of who has access to my work. During our time together I have recognized problematic patterns in my own attempts at activist work. I often struggle with the balance of making art from a place of personal expression and ownership of the stories I can tell. I make art with my body first, often discovering a piece alone in a dance studio, yet I also want to tackle difficult themes of injustice. I struggle to situate critical concepts as movement in my body of privilege. What happens when my white, cis-gender, middle-class, American body should not be centered? Do I continue to take action in my art? My answer has been to differentiate the type of work I make, either personal or communal. In the latter, I remove my white body from the choreography and turn to the community, to the layered stories that engage the multiplicity of human experience. Donna: I feel deeply indebted to you, Rose, for your vision in November of 2016. As a result of your courage, we started a coalition, which existed for and with other activist artists and community partners. I must name some of these partners: Gandhi Institute for Nonviolence, Rochester Contemporary Art Center, 21st Century Arts and the Women of Color Art Collaborative, FrazeeFeet Dance Company, Spiritus Christi Church, and the Memorial Art Gallery, to mention only a few. Our work together became a small part of society's resistance to old injustices that were being praised and reinforced by new public rhetoric. We had two goals, first, to generate more equity and belonging for all marginalized people, and second, to get people out to vote against Trump. It seems that ACCT represents a microcosm of failures and successes in modern society-how grassroots activism can make a difference and how little difference it makes. On a personal level, as a woman of color with academic privilege, ACCT taught me to love community activism and to question academia. As a result of our journey as artists in ACCT, I can now admit to assimilating for years, trying to fit into a predominantly white institution, and therefore to being complicit with racism, even against myself. Internalized racial oppression is real, and I have now faced its pedagogical effects. I've asked myself what I have unconsciously taught my students. Even as our artists' coalition addressed multiple reckonings-Trump's presidency, Covid-19 pandemic, and the Black Lives Matter response to George Floyd's murder-we all had to face personal reckonings too and realize how each of us must coalesce new insights to resist oppression everywhere, even within ourselves. --- Conclusion While performance can leave a kinesthetic impression on those who witness it, we wonder if dance and performance can actually make change. Should it try? Dance is often assumed to be apolitical, yet can any human art be apolitical? Although dance may not change policy, we know that dance changes people; the essence of dance is change in the body as it comes into relationship with others. The power of embodiment resides within its liminal truth.

The struggle for justice is not new, yet the impact of intentional embodiment and dance in protest is on the rise. Bringing together embodied anti-racist work and dance as a practice of resistance, this artists' statement describes a grassroots coalition in Rochester, New York: Artists Coalition for Change Together. Co-founded in 2017 as a response to the recent election, ACCT began as an organization of progressive dance artists and collaborators who sought to generate social change through performance, direct action, community dialogues, and scholarship. Written as a dialogue between two founders of ACCT, this statement examines the history of the coalition and its acts of embodied protest. As moving bodies and art communicated resistance in direct "ACCTions," community activism was generated. Through the lens of existing research and college courses on arts activism, the authors reference multiple reckonings in the North American cultural body, which peaked during the pandemic in 2020.

Background Tailored psychosocial support for vulnerable groups, such as unaccompanied refugee minors , is of great importance. URMs are refugee children and adolescents under the age of 18, who have been separated from their primary caregivers [1]. In 2017, approximately 6000 URMs were under guardianship in the Netherlands, with the majority coming from Eritrea, Afghanistan, or Syria. In general, refugee minors are exposed to a variety of stressors before and during their flight, such as forced migration and other dangerous circumstances. After migration, these minors often continue to experience hardships, such as resettlement in a new and strange environment, racial discrimination, isolation, and insecurities concerning family reunification and their refugee status. The potentially traumatic events before and during the flight, as well as the continuous stressors related van Es et al. Child Adolesc Psychiatry Ment Health 15:53 to the post-migration context can affect the mental health of these refugee minors [2][3][4]. URMs are considered one of the most vulnerable groups of refugees [5]. Studies indicate that, compared to accompanied minors, many URMs were exposed to more adverse events, such as sexual assault and violence [5,6]. The vulnerability of URMs might be further enhanced by the separation from their caregivers and the lack of support that parental and family care can offer [7]. Altogether, URMs are at greater risk of developing mental health problems than accompanied minors. A substantial group of URMs develop symptoms of posttraumatic stress disorder and/or depression, which may become chronic [8]. As mental health problems during childhood and adolescence can be longlasting, it is crucial to address these problems promptly [9]. Furthermore, the continuous stressors URMs face are likely to maintain their mental health complaints [8,10]. For example, Eritrean URMs and their caregivers in the Netherlands stated that continuous stressors, such as worries about their family members, the complex family reunification procedure, and a lack of financial resources, highly affected their psychological functioning. URMs and their caregivers suggested that, as a result, URMs developed, or experienced an exacerbation of, sleeping problems, difficulties concentrating, and other health issues [11]. Although URMs have an elevated risk of developing mental health complaints, few studies have focused on interventions and programmes targeting these complaints. Most of these studies were qualitative or based on case descriptions [12,13]. To our knowledge, only one randomized controlled trial was conducted. This study showed that a trauma-focused group intervention for URMs, drawing on cognitive behavioural principles, improved self-reported symptoms of PTSD and depression, but not caregiver-reported symptoms [14]. Moreover, a case study and an uncontrolled pilot study on trauma-focused cognitive behavioural therapy yielded preliminary evidence that this therapy is feasible and may effectively reduce PTSD symptoms in traumatized URMs [2,13]. Finally, a recent pilot study suggested it is feasible to implement Narrative Exposure Therapy for URMs [14,15]. One possible reason for the lack of intervention studies is the number of barriers faced when offering traumafocused interventions to URMs. One study indicated that although approximately 60% of the participating URMs reported a need for professional support, only 11.7% had actually received this help [16]. Barriers to the provision of adequate mental health care include individual barriers-such as linguistic differences, taboos concerning mental health care, lack of knowledge of the mental health system and mental health disorders, as well as a fear of stigma-and structural barriers-such as a lack of financial coverage for treatment and interpreters, and poor access to services [7,12,[16][17][18][19]. We therefore face the challenge of diminishing these barriers and offering culturally sensitive and accessible interventions to this group of minors. To overcome barriers to mental health care, we developed a short-term multimodal trauma-focused treatment approach, specifically adapted for URMs. This approach, described in more detail below, includes CBT interventions to target continuous stressors and to alleviate symptoms of depression and stress. The treatment approach aims to overcome individual as well as structural barriers by offering the treatment at the living location of the minors or another location of the minor's choice, collaborating with intercultural mediators , who aided in understanding cultural and language differences between the minor and the therapist, and conducting a flexible, multimodal treatment approach that allows the treatment sessions to focus on the minors' current request for help. This paper describes a pilot study designed to evaluate this trauma-focused treatment approach for URMs. Specifically, we explored the main request for help among URMs in the Netherlands referred to the treatment approach; the treatment integrity and feasibility , and the course of symptoms of PTSD and/ or depression. The information collected during this pilot study was meant to help us in refining the treatment approach to meet the needs of URMs in the Netherlands adequately. The findings will also inform future efforts in studying treatment programmes for URMs. --- Methods --- --- Procedure Guardians were informed about the intervention by youth care professionals working at Nidos. Guardians identified minors who experienced symptoms of PTSD and/or depression in addition to barriers to regular mental health care, such as low motivation for treatment, long waiting lists, and the idea that Western health care would not correspond with the URM's needs and culture. The minors were consequently referred for the multimodal trauma-focused treatment approach. Next, a therapist and ICM visited the minor at their living location or another location of the minor's choice, for an intake interview. The trauma-focused treatment approach then started. Assessments of psychological complaints and potentially traumatic events were carried out during the session following the intake interview . Post-measurements of psychological complaints were conducted during the last session . After each session, therapists filled in a list assessing elements of programme integrity . After all treatments had been completed, all ICMs and therapists were asked to fill in a questionnaire to evaluate the trauma-focused treatment approach. The questionnaires were conducted online or over the telephone due to restrictions related to COVID-19. To assure data integrity and compliance with ethical and juridical aspects, the study was reviewed by the Medical Ethics Committee of the Leiden University Medical Centre. The Committee stated the study did not require ethical approval as it intended to improve care as usual. Minors received verbal information on the study and provided verbal or written informed consent. For minors below 16 years of age, legal guardians also gave verbal or written consent. The assessments with the minors were part of the treatment evaluation and also used for clinical purposes. Participating therapists and ICMs gave online consent for the use of their data for research purposes. --- Measurements All assessments of the minors were administered with help from an ICM. The Children's Revised Impact of Event Scale-13 is available in several languages . The Patient Health Questionnaire-9, modified for adolescents is available in English. When a minor spoke another language, the questionnaires were translated into this language by an ICM. --- Demographic information The following demographic information was collected: gender, age, country of origin, whether the minor came from a city or town, and whether the minor had any family members in the Netherlands. --- Request for help To map how URMs define their main problem and request for help, minors were asked to formulate their main problem. --- Life events During the second treatment session, therapists identified positive life events, adverse life events, and losses of loved ones the minors had been exposed to. They did so by laying down a lifeline, as is described in the Narrative Exposure Therapy for the treatment of traumatized children and adolescents' protocol . Minors were asked for their permission to share information concerning their treatment sessions, including the lifeline, with the researchers. --- Programme integrity To assess whether the treatment approach was performed as intended, two components of programme integrity, based on the conceptual framework of Carroll et al. [21] were assessed: exposure and adherence. Exposure was operationalized as the number of sessions per client and minutes per session. Adherence was operationalized as the extent to which therapists adhered to the specific components of the approach. The programme integrity list was completed by the therapist after each session. In addition, the programme integrity list included questions on the duration of the treatment approach, how therapists evaluated the collaboration with the ICM, the main theme of each session, whether a module protocol was used, whether the therapist deviated from the treatment van Es et al. Child Adolesc Psychiatry Ment Health 15:53 approach, and why they deviated from the treatment approach. --- Feasibility After finishing all sessions, the therapists and ICMs involved in the study were asked to evaluate the multimodal trauma-focused treatment approach by filling in an online questionnaire. The questionnaire was based on a brainstorming session in which the researchers discussed the evaluation aims. It consisted of mostly open-ended questions focusing on positive aspects of the intervention and elements that could be improved or changed. Therapists and ICMs were also asked to arrange aspects of the approach according to their importance, based on their experience. --- Symptoms of posttraumatic stress All minors completed the CRIES-13, a 13-item tool measuring symptoms of posttraumatic stress in children aged 8 and older. The scale includes three subscales: intrusion, avoidance, and arousal. Items are rated on a 4-point scale rating 0 = not at all, 1 = rarely, 3 = sometimes, and 5 = often. A total score of ≥ 30 suggests an increased risk of PTSD [22]. The CRIES-13 is a valid measure of posttraumatic stress that has been used extensively among children exposed to war and with different cultural backgrounds [22,23]. The Cronbach's alpha in this study was good . --- Symptoms of depression To measure symptoms of depression, the PHQ-A [24] was used. The PHQ-A is a modified version of the PHQ-9. For example, 'reading the newspaper' in the PHQ-9 questionnaire was changed to 'schoolwork and reading' in the PHQ-A. The PHQ-A includes nine items rated from 0 to 3 with a time period of the past 7 days. In line with the PHQ-9, a total score of ≥ 10 was considered as the cut-off score for detecting depression [25]. Despite a few differences, the measure is mostly consistent with DSM-5 criteria for a major depressive disorder. The PHQ-9 is a validated, frequently used instrument measuring depression in adolescents [26]. The last question of the PHQ-A is on self-harm and suicidality. All therapists were licensed and trained to explore suicidality, and suicidality was also included in the training for ICMs. If a minor was suicidal, it was explained to the URM that their guardian and/or mentor would be informed. If needed, a child and youth psychiatrist could be consulted. Although the reliability and validity of the PHQ-A for refugee minors from different backgrounds have, to our knowledge, not yet been established, the PHQ-A has been shown to have acceptable psychometric properties when completed by Arabic-speaking adolescent refugees [27]. The Cronbach's alpha in this study was acceptable . --- Treatment and therapists Before starting the present study, we performed a nonsystemic evaluation of the multimodal trauma-focused treatment approach [28]. After offering the treatment to 33 Eritrean URMs, the therapists and researchers reviewed the approach. They noted that it focused mainly on traumatic events in the past, however, they emphasized the importance of tailoring the treatment to meet the minor's individual needs, for example by focusing more on current, ongoing stressors. The approach was therefore adapted to allow for more flexibility. Moreover, the importance of involving the direct context of the URM was emphasized by the involved professionals. The procedure is presented in Fig. 1. URMs were offered the multimodal, culturally sensitive, traumafocused treatment approach. The approach started with a clinical intake interview. During the intake, confidentiality and mutual expectations were discussed. During the following session, psychoeducation about PTSD symptoms and trauma-focused treatment were offered. As URMs are often unfamiliar with such treatments, and differences in explanatory models between the therapist and the minors might exist, considerable time was spent on psychoeducation. For example, therapists normalized symptoms by explaining how complaints might be linked to traumatic experiences in the past and continuous stressors. Additionally, the treatment rationale and potential side effects of the treatment approach were explained. Finally, the minors' lifeline was laid out, either by drawing or by laying down a rope and placing flowers for their positive experiences and stones for their negative experiences. Next, it was decided which intervention module suited the minor best. This was done during multidisciplinary consultation based on the request for help and complaints as reported by the minor. Each session lasted approximately 90 min. The final session focused on leave-taking. If more help was needed, the therapist discussed with the minor and the guardian what was needed and, if necessary and suitable, the therapist provided this. If therapists were not able to offer this support, they assisted the URM and their guardian to find suitable help. For example, therapists helped finding suitable mental health care nearby. The treatment approach included different modules, including CBT Reprocessing , to address continuous stressors, and symptoms of depression and stress. Culturally sensitive adaptations of the modules included collaborating with ICMs before and during the sessions. Before the sessions, how to offer psychoeducation in a culturally sensitive way was discussed, for example by using suitable metaphors. The core components of the treatment modules were not adapted. NET The lifeline was construed based on KIDNET [29] and served as a way to identify key adverse life events, pos-itive life events, and losses of loved ones. Moreover, the lifeline provided insight into the link between complaints and traumatic events . NET is conditionally recommended by the American Psychological Association for the treatment of PTSD [30]. EMDR URMs who had developed traumatic stress reactions as a result of a clearly defined traumatic incident received EMDR to address these issues. EMDR is a frequently used psychotherapeutic treatment [31] focused on weakening negative and strengthen- ing positive cognitions associated with the traumatic event. A key element of EMDR is a dual attention task. Whilst the client focuses on an image of the traumatic memory, bilateral, rhythmic stimulation, for example evoking saccadic eye movements, is offered [32]. Oras et al. [33] suggested that EMDR is an effective treatment for traumatized refugee children and the guidelines of the National Institute for Health and Care Excellence cited EMDR as a promising treatment for children and adolescents [34,35]. After discussions with ICMs, therapists, professionals with experience working in Eritrea, and EMDR experts, it was decided to offer EMDR, because during EMDR minors do not have to verbalize all of their traumatic experiences, which is more often the case in other trauma-focused treatments, such as KIDNET. URMs might be unfamiliar with talking about their experiences and therefore might be hesitant to do so. CBT NICE recommends CBT for the treatment for depression [36]. During the intervention, CBT was offered, for example, by supporting URMs in increasing activities and improving social connections [37]. --- Therapists The trauma-focused treatment approach was performed by therapists working at various mental health care institutions throughout the Netherlands. Therapists were recruited based on the network of the project team involved in designing the treatment approach. Therapists were all licensed mental health care workers and trained EMDR-therapists with multiple years of experience with minors from different cultural backgrounds. Therapists took part in multidisciplinary consultation as part of their regular work, as well as multidisciplinary consultation and supervision as part of the multimodal treatment approach. The motivation for therapists to take part in offering the treatment approach included the collaboration with ICMs, being part of a network of experienced psychologists, and working together to increase the understanding of what works for URMs in the Netherlands. --- ICMs ICMs were persons close to the URMs in cultural background and experience, who aimed to facilitate communication between the therapist and the minor. They interpreted language and offered information on the cultural background of the URMs. All ICMs followed a training focused on trauma, stress, and selfcare before providing the approach and were given the opportunity to participate in intermediate supervision sessions. --- Analysis --- Main request for help After an initial assessment of the main requests for help of the minors their requests were categorized into psychological problems, psychosocial problems, and somatic problems. All requests for help were related to these categories. In addition, a previous study indicated that URMs in the Netherlands often report psychosocial stressors and psychological complaints [11]. Moreover, refugees often present somatic symptoms [38]. --- Treatment integrity and feasibility To evaluate programme integrity, exposure, and adherence were computed. Missing items were allowed on the programme integrity list. When percentages are reported, they apply to all questions that have been completed. Adverse events were categorized following the subsequent categories, in line with the World Health Organization World Mental Health Surveys [39]: [1] natural and man-made disasters and accidents; [20] combat, war, and refugee experiences; [6] sexual and interpersonal violence; [7] witnessing or perpetrating violence; and [8] death of a loved one. Being captured or held against one's will was added as a category as more than 25% of minors reported experiencing this adverse event. Data from the online questionnaires were analysed using the general inductive approach for analysing qualitative evaluation data [40] in the following steps. All steps were conducted in parallel . First, the text was read thoroughly. Second, specific text fragments related to the research questions were identified and labelled to create categories. Third, overlap and redundancy were reduced. Finally, the most important categories were identified. The categories were discussed among the two raters until a consensus was reached. During this process, an ongoing discussion was maintained with the rest of the research team. --- Course of symptoms Data of the pre-and post-treatment assessments of the CRIES-13 and PHQ-A were analysed using SPSS . For the CRIES-13, in line with Verlinden et al. [41], data were counted as missing if more than one item on a subscale was missing. For the PHQ-A, in line with prior studies on the PHQ-9 [42], data were counted as missing if more than two items were missing. Missing values on the PHQ-A were replaced by the mean of the completed items of the PHQ-A, as suggested by Kocalevent et al. [43] and missing items on the CRIES-13 were replaced by the mean of the completed items of the same subscale. A Chi-square test and independent t-test were conducted to evaluate whether any differences related to demographic information and baseline scores van Es et al. Child Adolesc Psychiatry Ment Health 15:53 on the CRIES-13 and PHQ-A existed between participants who completed the questionnaires and participants who did not. To assess whether any changes occurred in symptoms of depression and PTSD, paired t-tests and Wilcoxon-signed rank tests were conducted. To evaluate clinical significance of change, we calculated the Reliable Change Index for changes in symptoms of depression and PTSD, using the formula t3-t1 Sdiff [44], with t 3 referring to post-treatment assessments of the PHQ-A and CRIES-13, t 1 referring to pre-treatment assessments of the PHQ-A and CRIES-13, and S diff being calculated using the test-retest reliability coefficient of the questionnaires and the standard deviation of the pre-treatment scores. The test-retest coefficient of the CRIES-13 is 0.85 [45]. Kroenke et al. [46] found that the test-retest reliability of the PHQ-9 was 0.84. A calculated RCI larger than |1.96| indicated a clinically reliable change, with 95% certainty. The RCI resulted in numbers of minors improved, unchanged, and worsened from t 1 to t 3 . In addition to the RCI, clinically significant change was further evaluating by assessing whether clients had obtained a score of < 30 on the CRIES-13, and a score of < 10 on the PHQ-A after the treatment approach. --- Results In total, 61 minors were referred to the treatment approach. Fifteen did not start with the approach for a variety of reasons, including: not being able to contact the guardian and/or minor , not being motivated for treatment , no longer having a request for help because family reunification took place , and starting treatment at another facility whilst placed on the waitlist . In addition, four minors did not give consent to taking part in the study. Ultimately, 41 minors participated in the study. Three participants dropped out immediately after intake, for the following reasons: moving house, needing another form of support because of current threat of deportation, and placement in a secure psychiatric facility. A flowchart is provided in Fig. 2 but one of the ICMs completed the online questionnaires. The ICM who did not complete the questionnaire was contacted several times but did not reply to the requests of the researchers. --- Request for help Most participants reported psychological problems, such as difficulties sleeping and concentrating, flashbacks, and feelings of depression. Moreover, psychosocial problems, such as worries concerning family members and problems with housing, were reported by 27 participants . Finally, six participants stated that they experienced somatic complaints, such as headaches and incontinence. --- Programme integrity The data of minors who attended at least one session after intake were included in analyses based on the programme integrity list . Participants were offered 8 sessions on average , which lasted approximately 80 min. Three minors dropped out after 2 or 3 sessions for the following reasons: wanting to focus on their future and preparing for their asylum interview, not being able to find motivation for treatment, and experiencing too much stress after receiving negative news concerning their asylum status. Therapists collaborated with ICMs during 93% of the sessions. The main reasons for collaboration with ICMs reported by the therapists included ICMs bridging the cultural and/or language gap , that collaboration with ICMs is part of the protocol , and ICMs helping in creating a connection with the minors . Therapists sometimes worked with professional interpreters , or without an ICM or interpreter , mostly when the ICM was not able to join a session due to practical reasons. During the majority of the sessions, therapists rated their collaboration with ICMs positively , stating for example that they 'understand the culture' , and that they promote the building of a trusting relationship. In a few cases therapists said the collaboration could be improved-this was mostly when the ICM joined the session online or via telephone as a result of restrictions due to COVID-19. The sessions focused primarily on continuous stressors , trauma , laying the lifeline , getting to know each other , saying goodbye , and psychoeducation . During 56% of the sessions, therapists followed a treatment module protocol. In these sessions, therapists primarily offered EMDR and the lifeline derived from KIDNET . In other sessions where a protocol of a treatment module was followed, therapists used CBT , or specific protocols, for example focused on stabilization or dealing with anger. Therapists documented several deviations from the treatment approach. For example, they deviated to address the current needs of the minor, which often concerned dealing with continuous stressors , and news the minors had received . Moreover, therapists noted that they sometimes needed to take extra time for psychoeducation, explaining EMDR, building trust, laying down the lifeline and translation. The circumstances due to COVID-19 resulted in several deviations from the approach for approximately a third of the minors. For example, two of the minors received all sessions online/over the telephone. In addition, several minors were contacted online or via telephone by their therapist between sessions for continuity. These sessions resulted in several challenges, including arranging devices, a secure internet connection, and adequate space for the minors. Furthermore, the focus of these sessions sometimes shifted to dealing with social isolation and restrictions. Therapists reported they took time to involve the direct context of the minors, for example by having a guardian or mentor attend the intake interview, and by talking to a guardian or mentor concerning the minors' continuous stressors. In addition, therapists sometimes deviated from the EMDR protocol because minors reported they did not want to continue, for example because their arousal was too high or because they had an extreme headache. Therapists described many differences between the participating minors. For example, whereas some minors were very motivated and wanted to talk about their experiences, others had difficulties developing a trusting relationship. One minor stated he did not want to continue with EMDR as it reminded him of magic. In some cases, treatment was mostly focused on motivating the minor to receive mental health care. Treatment termination took place for different reasons. Most minors felt the treatment had helped them overcome their problems , some minors were referred to another mental health facility , and others wanted to focus on their future/current worries rather than therapy . --- Evaluation by therapists The therapists' experience with offering the treatment approach varied widely. For example, three therapists had treated one minor, and two others had seen approximately 20 minors. All therapists found the approach useful and wanted to continue offering it. According to the therapists the most important elements of the approach included outreach work, working on a mutual established goal, and working with an ICM with a similar cultural background to the minor. They also valued the flexibility of the approach and being able to deviate to suit the needs of the minor. Some therapists experienced difficulties whilst explaining EMDR to the minors, and some thought that EMDR did not always suit the minors' needs. Several stated that the protocol could be improved by some adaptations, for example by allowing for more flexibility, and by adding more psychoeducation, CBT exercises relaxation exercises, and homework exercises to the protocol. Finally, they underlined the importance of local professionals throughout the Netherlands offering the approach, and of sharing knowledge and expertise. Therapist: 'We reach minors that wouldn't normally be reached. We try to adapt to fit to the needs of the minor, instead of the minor adapting to the mental health care institution. ' --- Evaluation by ICMs Eight ICMs took part in the evaluation. Their experience with the approach varied from one treatment to more than ten treatments. The majority of ICMs were involved in more than four treatments. Seven ICMs found the approach useful and all ICMs wanted to continue to offer the approach. The ICM who did not find the approach useful stated that more time should be invested in getting to know each other, and that the therapists as well as the ICM should share more personal stories with the minors. ICMs especially valued that the treatment approach could be adapted to fit the minors' needs better. ICMs rated the KIDNET lifeline, the outreach work, and the engaged approach by the therapist and ICMs as the most important elements of the approach. Most ICMs felt their work was an important addition to the approach. For example, one ICM explained that ICMs can assist in formulating certain questions. Some ICMs added that the approach could be improved by offering more psychoeducation, and spending time on building trust. Several ICMs stated their work could be improved by receiving more training focused on their role in the treatment approach, and some ICMs stated that the communication between therapist and ICM could be improved. Finally, the ICMs highlighted the importance of involving the patient's context and added that this could be improved. ICM: 'Before the therapy begins, the ICM and therapist should spend more time with the minor. Otherwise, you can't build trust. My experience is that when more time is invested in the beginning, the outcomes are better' --- Course of symptoms Most participants did not complete the questionnaires. Twelve participants completed the Pre-and post-measurements of the PHQ-A, and 17 completed the preand post-measurements of the CRIES-13. Obstacles in obtaining complete measurements included drop-out; limited time to conduct the assessments; restrictions due to COVID-19 ; a lack of motivation to fill in the questionnaires; and difficulties with understanding the questions. Another reason for non-completion was that therapists were afraid to burden the minors by asking questions about their mental health. Based on an independent t-test and a Chi-square test, we found no differences in demographic information or baseline scores on the PHQ-A and CRIES-13 between minors who completed the questionnaires and minors who did not complete the questionnaires. Most participants who filled in the CRIES-13 pre-treatment had developed heightened symptoms of PTSD, and approximately half reported moderate to severe symptoms of depression. A paired t-test was conducted to compare pre-and post-intervention scores on the CRIES-13 and PHQ-A . A statistically significant reduction in PTSD scores was found. There was no significant change in depression scores. To ensure robustness, Wilcoxon signed-ranks tests were also conducted, yielding a significant difference for the CRIES-13, but not for the PHQ-A . Table 3 shows the CRIES-13 and PHQ-A scores at t 1 and t 3 , and the RCI scores for participants who completed pre-and post-treatment measurements. As indicated by the RCI, 10 of the 17 participants who completed the CRIES-13 improved , seven remained van Es et al. Child Adolesc Psychiatry Ment Health 15:53 unchanged , and none worsened. Nine of the participants who improved scored below the cut-off point after treatment. Concerning the PHQ-A, three of the 12 participants improved , eight remained unchanged , and one worsened . Two participants who improved scored below the cut-off score after treatment, whereas two participants who scored below the cut-off point before treatment, scored above the cut-off point after treatment. --- Discussion The aim of the current study was to provide an evaluation of a multimodal trauma-focused treatment approach offered to URMs in the Netherlands. Forty-one URMs, mostly from Eritrea and Syria, started with the intervention. These minors arrived in the Netherlands after a harrowing journey, often being exposed to atrocities, hunger, thirst, and losing loved ones including companions and friends. Minors reported they would like to receive help to alleviate psychological issues, such as difficulties sleeping. Moreover, most minors reported psychosocial problems, including worries about their family members and their asylum status. A few minors stated they wanted help with somatic issues. Somatization in refugees is connected with psychopathology and can be perceived as an idiom of distress [38]. Moreover, some minors experienced physical abuse and neglect and might have experienced somatic issues as a result. In addition, most minors had developed heightened symptoms of PTSD, and approximately half reported moderate to severe symptoms of depression. Minors attended approximately eight sessions of the trauma-focused treatment approach. Concerning programme integrity, therapists often deviated from the approach. ICMs and therapists elaborated that they often needed more time to offer psychoeducation. Moreover, therapists and ICMs emphasized the need to adapt the focus of the session to meet the current need of the minors. In addition to a focus on trauma, continuous stressors often played a prominent role in the treatment approach. The treatment approach often consisted of a combination of modules, including the lifeline , sessions focused on trauma, and sessions focusing on continuous stressors. The feasibility of the pre-and post-measurements of PTSD and depression was low. The majority of the minors did not complete the questionnaires, and as a result, these findings are not representative of the entire sample. Most minors who did complete the measurements reported a decrease in symptoms of PTSD. No significant change in symptoms of depression was found. An important reason for non-completion was that therapists were afraid to burden the minors by asking questions about their mental health. It is possible the minors who were experiencing mental health issues were spared by not conducting the questionnaires. This finding emphasizes the need of an assessor other than the therapist to conduct the questionnaires. The results suggest that the trauma-focused treatment approach is partly feasible and indicates that certain barriers to mental health care can be overcome by offering this short-term, outreach approach. For example, several minors who experienced barriers to regular mental health care, where motivated for referral to another mental health facility after getting offered the treatment approach. The professionals involved in the execution of the treatment approach evaluated it positively, and all of them said they wanted to continue to offer it. Moreover, the drop-out was similar to drop-out in other studies researching trauma-focused treatments [47]. Drop-out was often due to circumstances linked to the minors' asylum status, and needing more help than this short-term treatment approach can offer. There were several limitations to the feasibility of the treatment approach, including low treatment adherence and challenges faced during implementation of the protocol. Notably, factors limiting programme integrity and feasibility were frequently related to structural barriers. For example, deviations, reasons for drop-out, and missing data, were often due to factors such as news concerning asylum status, moving house, and restrictions due to COVID-19. When evaluating a treatment approach in the context of trauma and resettlement, structural barriers often arise. Although these barriers are out of reach for the minors and therapists, it is important to take these barriers into account while planning the evaluation and organizing funding. For example, it could be helpful to adopt a flexible approach to deal with the ever-changing context of these young refugees, and to support URMs in developing coping strategies to deal with these continuous, structural stressors. --- Strengths and limitations The current study is one of the first to evaluate a traumafocused treatment approach specifically for URMs [12,13]. Although there are several limitations due to the design of the study, this study offers an overview of a treatment approach that can be offered to this understudied population in a naturalistic setting. Conducting this culturally-sensitive outreach approach, we were able to reach a group subject to barriers to mental health care who often do not receive adequate care. Another strength of the current study is that different perspectives are presented, including the perspectives of the therapists, ICMs and URMs. A factor that might have contributed to the feasibility of the trauma-focused intervention was the flexibility of the treatment approach. During a previous evaluation [11], professionals noted the importance of offering a flexible approach. This study offered insight into the wide range of requests for help reported by the participating URMs. Treatment sessions were often tailored to meet the needs of the minors, which is reflected in the main themes of the sessions as well as in the deviations reported by the therapists. A multimodal approach was conducted, offering modules of KIDNET, EMDR, and CBT, all of which are promising or recommended treatments for PTSD and depression. Multimodal treatments have been offered successfully to children suffering from psychological problems, including depression, anxiety, and trauma [48,49]. Although the flexible approach was valued as a strength by both therapists and ICMs, it complicates drawing conclusions on what aspects of the approach contributed to its impact. Several limitations deserve attention. First, the data collection of the questionnaires in this study relied solely on therapists, as there was insufficient funding for researchers to conduct the measurements. Because many sessions did not contain the planned treatment, it is not possible to draw any inferences about the impact of the presented treatment approach. Second, therapists were recruited based on their training level and experience with minors with different cultural backgrounds. It is plausible that this resulted in the selection of highly motivated therapists, and therapists with less experience and affinity with this population might have evaluated the approach less positively. Third, the CRIES-13 was used in the current study to measure symptoms of PTSD, however, this questionnaire is based on the Diagnostic and Statistical Manual of Mental Disorders-IV . Future studies should include a more up-to-date questionnaire in line with the current DSM-5 or the International Classification of Diseases 11th Revision , such as the Child and Adolescent Trauma Screen [50]. Fourth, URMs form a heterogeneous group with a wide range of cultural backgrounds, experiences, and requests for help. The generalizability of the findings is limited, as the current study only focused on a small sample size with participants from three different countries. As a prior evaluation of this approach focused solely on Eritrean minors, guardians often referred Eritrean minors. The current design did not allow for comparison of the treatment's impact for different cultural groups. --- Research implications The current study aimed to explore the feasibility of the multimodal trauma-focused treatment approach. Future efforts should focus on examining the effectiveness of this treatment approach. Demazure et al. [12] state that there are many challenges to conducting double-blinded RCTs of the impact of interventions for URMs. However, the authors suggest that small-N designs could be a viable alternative to evaluate intervention effectiveness. In addition to a quantitative evaluation, future research can adopt a mixed-methods approach by including qualitative data, such as interviews with URMs who have received treatment, in order to hear the voices and experiences of the URMs themselves. --- Clinical implications The collaboration between psychologists and ICMs emerged as one of the strengths of the treatment approach. The involvement of ICMs went beyond the translation of language. ICMs can aid in providing a culturally sensitive explanation of the treatment rationale, in bridging the cultural gap, in building a trusting relationship, and in motivating minors. Nierkens et al. [51] emphasize that the quality and accessibility of health care can benefit from collaborating with ICMs. In line with the evaluation by ICMs, Qureshi et al. [52] note that the role of ICMs can be improved by offering clarity regarding their role. To contribute to the professionalization of ICMs, it is essential to continue to offer supervision, guidance and training. The current study offers several clinical implications for the multimodal trauma-focused treatment approach. Modifications include providing specific approaches to address continuous stressors, and offering more attention to psychoeducation, as an important deviation from the approach was that therapists often spent more time on continuous stressors and psychoeducation. Continuous stressors related to resettlement can play a crucial role in the development of mental health problems in refugees. A recent pilot study emphasized that current stressors, including the rejection of the asylum claim, can potentially increase symptoms of PTSD in URMs post-treatment. Bruhn et al. [53] state that identifying and dealing with postmigration stressors might limit the impact of these stressors on treatment. Although addressing these problems can be very complex and at times impossible , interventions can focus on dealing with these problems, for example by strengthening the social network, and by improving coping strategies. The protocol could therefore benefit from providing specific approaches to deal with continuous stressors. In addition, both therapists and ICMs noted they often needed more time for psychoeducation, which psychoeducation is suggested to have a positive impact on PTSD and psychosocial factors, such as social support [54]. The treatment approach protocol can be further improved by offering more attention to psychoeducation. One of the major future challenges is implementing the presented treatment approach in a sustainable way. The current design requires many resources, including highly trained and experienced ICMs and therapists. Therapists conducting the treatment approach have to be trained to offer KIDNET, EMDR and CBT. However, it is not always feasible to secure these resources. One solution might be to put effort into training and motivating therapists who are less experienced. Motivation for participation in offering this treatment approach might be the collaboration with ICMs and working together with a network of experienced therapists to find out what works for URMs. To assure quality of the treatment, continued supervision and multidisciplinary consultation needs to be offered. Another option is to invest in alternative models. For example, the World Health Organization developed Problem Management Plus , a brief intervention targeting common mental health problems, based on CBT and problem solving strategies [55]. A recent study found that PM + is acceptable, feasible, safe, and may be effective in improving mental health and psychosocial functioning in refugees. In addition, PM + can be delivered by non-specialist refugee helpers [56]. Furthermore, two RCTs are currently assessing the effect of a stepped care approach addressing PTSD in URMs [57,58]. Using a stepped care model lay counsellors or ICMs could be trained to offer psychoeducation and the lifeline as derived from KIDNET. The presented treatment approach seems to be indicated when URMs and their guardians experience barriers to regular mental health care. The outreach care and collaboration with ICMs aids in overcoming these barriers. However, it remains unclear whether the content of the treatment approach-including the lifeline as derived from KIDNET, EMDR, and CBT-is more suitable than other treatment approaches, such as TF-CBT. Future --- Conclusion This study offers a first evaluation of the feasibility of a multimodal trauma-focused treatment approach specifically for URMs in the Netherlands, taking into account their specific needs; the context; pre-and postmigration stressors; and language and cultural differences. The results are promising and provide a first indication that this approach mostly overcomes barriers to mental health care, and that the treatment is partly feasible. --- --- Code availability The SPSS syntax code is available from the corresponding author, CvE, upon reasonable request. --- --- Clinical trial registration The study was registered in the Netherlands Trial Register , 10 April 2020. --- --- Competing interests The authors have no conflicts of interest to declare that are relevant to the content of this article. ---

Background: This study evaluated the feasibility of a short-term, multimodal trauma-focused treatment approach adapted specifically for unaccompanied refugee minors (URMs) in the Netherlands. This approach aims to overcome barriers to mental health care and to reduce symptoms of posttraumatic stress disorder (PTSD) and depression. Methods: An uncontrolled study was conducted, evaluating the main request for help, treatment integrity and feasibility, and the course of symptoms of PTSD (Children's Revised Impact of Event Scale-13) and depression (Patient Health Questionnaire modified for Adolescents). Results: In total, 41 minors were included in the study. Most participants were male (n = 27), predominately from Eritrea (75.6%) with a mean age of 16.5 (SD = 1.5). Minors mostly reported psychological problems, such as problems sleeping, and psychosocial problems, including worries about family reunification. Deviations from the approach were made to meet the current needs of the minors. Factors limiting the feasibility of the approach were often related to continuous stressors, such as news concerning asylum status.The results provide a first indication that this approach partly overcomes barriers to mental health care and emphasize the added value of collaborating with intercultural mediators and offering outreach care.The study was registered in the Netherlands Trial Register (NL8585), 10 April 2020, Retrospectively registered, https:// www. trial regis ter. nl/ trial/ 8585.

Introduction A rural woman is a female living in the rural area, whose everyday life is characteristic of this context, with a family structure and lifestyle typical to the life spent in rural areas, spending her time in reproductive and housework and farming, i.e., agriculture . Although advancements were achieved in public policy, such as rural retirement and access to credit, the singularity of rural women is marked by socioeconomic hardships and behavioral and health problems arising from their context, which directly interfere with their quality of life . Historically, the sociocultural processes in Brazilian rural areas assign different roles to genders, impeding women from playing a productive role, i.e., they are considered helpers of their fathers and husbands in agriculture income-generating activities but are essential for reproductive activities and housework . The lives of women in rural areas are marked by the environment where they live, while living and work conditions directly influence their health, exposing them to the risk of diseases. Gender-related aspects also result in more negative QoL among rural women . According to the World Health Organization, QoL is a subjective and multidimensional construct that includes biopsychosocial aspects, beliefs, independence level, environmental characteristics, and social relationships . Even though women's work in rural areas is often overlooked, they do not lose their perseverance and courage and keep enduring, seeking improved living conditions supported by resilience processes. Resilience is a concept that represents an individual's flexibility and ability to bring resources to adapt to adverse situations, aiming to maintain a biopsychological balance . This study enables unveiling aspects related to the QoL of rural women, considering contextual inequalities that affect these women's history of life and their healthdisease-care process. It also contributes to developing interventions and public policies to promote improved QoL and, consequently, rural women's health, which should be considered a more comprehensive phenomenon than diseases. From this perspective, improved QoL represents one of the most difficult challenges imposed on public policies directed to populations in rural areas . QoL and resilience are considered aspects that cannot be understood only from the perspective of objective data, but rather one has to immerse in its subjective sphere, and the confluence of these two spheres confer reliability to the process this study is intended to analyze. Therefore, the following guiding question is proposed: How do rural women's perceptions corroborate the results showing that their resilience intersect with their QoL? In order to answer this question, this study's objective was to analyze the intersections between rural women's quality of life and resilience. --- Method Study design This convergent mixed methods design concomitantly implemented qualitative and quantitative elements. These studies were equally prioritized but kept independent during the analysis process, while the results were then combined for general interpretation. An implicit perspective of theorization was adopted, anchored on pragmatism, a paradigm proposed for the mixed methods design . A cross-sectional QUAN study and a QUAL study guided by oral history were triangulated. Note that data collection and analysis occurred simultaneously but independently, and data were integrated after each study was analyzed separately. --- Study setting This study was conducted in Nazarezinho, a town located in the backwoods of the state of Paraíba, Brazil. --- The Brazilian Institute of Geography and Statistics --- Selection criteria The sample comprised women aged 18+ years old who consented to participate and were performing or had performed agricultural labor and/or extractivism at some point in their lives. Women who lived in the rural area but did not present a lifestyle or pursuits related to life in rural areas or lived in the rural area for less than six months were excluded. --- Sampling A convenience sample without replacement was adopted in the QUAN study. It was calculated with a sample error of 5% and a 95% confidence level, considering a formula in which N represents the population , Silva BN, Santos JLG, Riquinho DL, Miranda FAN, Souza NL, Pinto ESG. n 0 represents an initial approximation of the sample size , and E 0 represents the tolerable sampling error , reaching a sample of 87 rural women. The participants were selected for the QUAL study based on the concept of oral history of the target community , colony , network, and point zero . The network was determined based on point zero . The Community Health Agent working in the micro-area nominated the first rural woman to be interviewed. The rural woman nominated by the CHA nominated a second rural woman to participate in the study, who in turn, nominated the following participant until the recruitment of participants ceased, i.e., when the reports enabled answering the study's questions and became repetitive, as recommended by the framework adopted . The network of collaborators who participated in the QUAL stage was composed of seven rural women. --- Instruments used to collect information The following instruments were adopted in the QUAN study: socio-demographic form to characterize the sample addressing objective questions related to socio-economic, occupational, and health-disease information. In addition, the Resilience Scale and Medical Outcomes Study 36-Item Short-Form Health Survey were used. The Resilience Scale measures the level of positive psychosocial adaptation in relevant life events. It was validated for the Brazilian Portuguese and is directed to adult populations. This instrument comprises 25 items rated on a seven-point Likert scale. The higher the score, the more resilient an individual is. For example, scores below 121 points indicate "poor resilience," while scores between 121 and 145 indicate "moderate resilience," and scores above 145 points indicate "high moderate" to "high resilience" . The SF-36 is designed to assess QoL and was validated in Brazil. The domains assessing QoL include physical functioning , role-physical , bodily pain, general health , vitality, social functioning , roleemotional , and mental health . Its final score ranges from 0 to 100 points and is calculated for each domain: 0 indicates the worst general health, and a 100 score indicates improved general health . The QUAL study included open-ended interviews recorded with an mp3 device. The first cue was: "tell me your history in the rural area." Other cues were presented as the participants shared their histories, such as "tell me more about it" to guide the process. --- Data collection The main author with experience in collecting data and a CHA working in the micro-area collected quantitative data between July and November 2020. The interviews were held at the rural women's homes using a printed form. The main author conducted the qualitative stage by personally interviewing the participants. The interviews had been previously scheduled via telephone between August and November 2020 and lasted 40 minutes on average. The relation of dependence between QoL domains and level of resilience was analyzed with a multiple regression linear model, in which the level of resilience was the dependent variable, while the eight domains of QoL were independent variables. Note that all the analyses were conducted with a level of significance established at 5% . --- Treatment and analysis of data Data produced in the qualitative analysis were transcribed verbatim, followed by textualization and transcreation, a stage in which the texts of the interviews were returned to the participants for them to review and legitimate. After the participants' feedback , the histories of life were treated using inductive thematic analysis, a method used to identify, analyze and report patterns within a set of data, enabling its organization and description with rich detailing. In this approach, coding does not assume a previously established theoretical framework; i.e., analysis was guided by data . The results were integrated by developing a convergences diagram and a joint-display table. Hence, meta-inferences of mixed methods were obtained, which resulted from combining quantitative and qualitative inferences. Finally, the results were interpreted from the perspective of gender, resilience, and QoL. Rev. Latino-Am. Enfermagem 2022;30:e3521. Regarding the regression model in which the level of resilience was a dependent variable, and the SF-36 eight domains were independent variables, it predicted 14.1% of the variance in the level of resilience . The statistically significant association between the level of resilience and the QoL SF domain, together with the result obtained in the regression analysis, indicates that the constructs QoL and resilience mutually influence each other. As for the qualitative results, the inductive thematic analysis of the histories of life enabled extracting the main themes that emerged in the analysis: rural life --- Results A sample of 87 rural women participated in the QUAN study. Most of these women reported being Caucasian , married , with incomplete primary education , and aged 50.15 years old on average. The following age groups were identified: 12.6% were between 18 and 30 years old; 13.8% were 31 to 40; 19.5% were between 41 and 50; 27.6% between 51 and 60; and 26.4% were 60+ years old. Regarding labor-related aspects, most participants were farmers for at least ten years, with planting being the most recurrent activity . However, note that even though most of the participants farmed for a long time, practically none received any pay . Regarding the social and clinical aspects and healthdisease continuum of the women participating in the QUAN study, almost half reported a chronic disease; systemic hypertension was the most common . None of the participants had private health insurance/ plan and health services were mostly provided by Primary Health Care . Regarding social aspects, most rural women reported they felt welcomed by their communities. The women participating in the QUAL study was 67 years old on average, most were married , had incomplete primary school , reported being Caucasian , and a monthly income of two times the minimum salary . Regarding the quantitative results, a moderate level of resilience ( =129. --- Divergence The women reported that physical strain demanded by rural labor is a positive intervenient factor in their health and QoL , and being enchanted with the rural life was a protective factor for the development of resilience, which conflicts with a lack of statistical significance between the level of resilience and the SF-36 PF domain found in the Mann-Whitney test. No statistically significant association was found between a moderate level of resilience ( =129. --- Divergence The rural women's concepts indicate that even though they faced limitations that impeded their activities, they were able to prevail, revealing resilience even when dealing with obstacles, which also conflicts with the lack of statistical significance found in the Mann-Whitney test. No statistically significant association was found between a moderate level of resilience ( =129. Divergence Bodily pain stands out as a factor that limits everyday activities of these rural women, impeding them from doing their work, though they see it as a source of resilience so that their pain is a fact that negatively interferes in their QoL. This finding also conflicts with the lack of statistical significance in the Mann-Whitney test. No statistically significant association was found between a moderate level of resilience and SF-36 GH domain ‡ --- Concepts on health and QoL My health is terrible, but I make it seem not so terrible because when I have problems, I put them in God's hands and ask Him to give me strength to overcome them [...] . Divergence Protective factors such as spirituality promote resilience among rural women enable them to develop coping strategies to deal with unfavorable health conditions, also differently from what was found in the Mann-Whitney test, which did not present any statistical significance between the level of resilience and the SF-36 GH domain. Rev. Latino-Am. Enfermagem 2022;30:e3521. In order to organize convergences that resulted from the influence of resilience on the QoL of rural women, we developed a diagram of convergences, represented by Figure --- Divergence Rural women find ways to live with quality based on their hope and resilient attitude, despite their psychosocial problems like anxiety, and keep their level of psychological well-being by facing obstacles from more than one perspective. However, the association between resilience and mental health was not statistically significant in the Mann-Whitney test. --- Relationship between the level of resilience and the SF-36 domains --- Themes and excerpts from the thematic analysis Meta-inferences The linear regression analysis with the SF-36 domains as independent variables predicted 14.1% of the variance in the level of resilience, which was a dependent variable . --- Rural life as a promoter of resilience and QoL --- I guess I'm resilient because I've suffered a lot where I came from [...] I experienced many hardships and obstacles, but I thank God for going through all these; I guess that I'm an overcomer. I've faced so much, but still, I moved on . --- Magnification The rural women's understanding shows that the adversities they face in their daily lives and work contributed to their health and QoL, working as protective factors that promote resilience processes. The QoL domains in the regression model corroborate these resilience processes, which positively feedback these women's health. *Physical functioning; † Role-physical; ‡ General health; § Social functioning; || Role-emotional; ¶ Mental health --- Discussion The mean score obtained for resilience could not be compared with the literature because there are no Brazilian or international studies assessing the relationship between resilience and QoL among rural women. However, a study conducted in Portuga l , using another resilience scale, reports a moderate level of resilience among rural elderly individuals, though the analysis did not consider gender differences. Other researchers , using the same instrument, identified a moderate level of resilience among individuals living in rural areas affected by drought. The moderate level of resilience identified in this study was similar to that found in a sample of the Brazilian population composed of 2,038 individuals, in which researchers found a moderate level of resilience of approximately 124.60 points. In addition, the analysis did not consider differences between genders and indicated that higher levels of resilience were associated with higher educational levels. However, this study's moderate level of resilience might be related to a lower level of education and older age. Note that even though the mean score indicates a moderate level of resilience, there is a need to invest in strategies and interventions to enable rural women to follow paths other than farm work, strengthening and promoting their resilience. These interventions are not supposed to romanticize inequalities and inequities but emphasize that resilience helps them promote courage, skills, and confidence to main their health even when their context favors marginalization and vulnerability . Researchers indicate that the individual level of resilience varies depending on the individuals' characteristics and their social context, directly contributing to QoL. Consequently, resilient individuals experience improved QoL arising from their ability to deal with and overcome hardships and general problems more easily . Regarding QoL, the worst mean scores were obtained in the GH and RP domains, similar to the scores reported by a cross-sectional study conducted with 355 rural workers in the state of Bahia, Brazil, which indicated demographic and socioeconomic variables as related motives . In addition, studies note that rural communities present the lowest levels of QoL , revealing that these communities are often and historically overlooked by public policies and health care models, directly affecting this population's QoL . Characteristics such as age, poor education, and lower socioeconomic levels are associated with lower levels of QoL, showing that the poorest and less educated individuals are the most vulnerable groups. Additionally, the older an individual, the more negative his/her QoL is because of the relationship between age and the emergence of various morbidities and limitations . As for the conceptions of health and QoL expressed by the rural women in the QUAL study, health in the rural context can be understood as an ability to perform everyday tasks, while health needs are secondary to work needs. This study also shows that health is represented as the ability to overcome adversities and is related to the resilience concept itself . From this perspective, having knowledge of resilience in the occupational health field, in this case, rural workers, can support a better understanding of these workers' environment, helping them maintain or recover their health, which results in QoL . The strategies that can be implemented in interventions to improve the QoL of rural women include improving family income and facilitating access to high-quality health care services . Likewise, the implementation of public health programs intended to improve the QoL of the rural population is essential, considering that it is an indicator that produces information capable of identifying and screening this population's health needs . Thus, public Rev. Latino-Am. Enfermagem 2022;30:e3521. agencies should implement specific programs directed to rural workers to promote their integral health . Therefore, even though quantitative findings indicate that resilience only influences one of the QoL domains, the qualitative results unveil divergences, enlarging an understanding that this influence may not be quantifiable. Additionally, the regression analysis showed the influence of the eight domains of QoL on the level of resilience, indicating the mutual influence of these constructs, as reported by a study developed with adolescents . Even though there are no other studies specifically reporting the influence of resilience on rural women's QoL, the association of these two constructs was found in other studies among different populations. For example, a study with 108 elderly individuals with chronic pain was conducted in São Paulo, Brazil, and verified that those with the highest levels of resilience also experienced improved QoL. Additionally, higher levels of resilience positively anticipate QoL . Another study found a synergy between resilience and QoL among adolescents in Cuenca, Ecuador. The meta-inferences proposed by this study and reinforced by studies developed in different contexts addressing different populations suggest that the QoL of rural women is influenced by their level of resilience, which is favored by protective factors. Hence, these factors culminate in resilience processes that promote QoL and are seen as processes that support changes and positive adaptation when facing risks. These processes may be individual , at the family level , or social support . Social support networks stood out among these factors and were confirmed by the reports of the women participating in the QUAL study. Most of the women in the QUANT study also reported a sense of being welcome, and an association was found between a moderate level of resilience and the SF domain of QoL. Social support is essential for developing resilience and improving QoL . Despite the role these networks play, the low score obtained in the SF-36 RE domain may be explained by the fact that this domain assesses the repercussions of the psychological characteristics on the individuals' wellbeing, which are individualized and subject to the influence of other domains that also obtained low scores, such as GH, RP, and bodily pain, which, unlike the SF domain, assess the integration of the individuals in social activities . Religious beliefs and spirituality were also identified as protective factors linked to the fact that these individuals' lives are permeated by a transcendental dimension , producing in people a feeling of belonging to something higher, timeless, and unlimited that produces an existential meaning . Spirituality is a factor that can be used in times of crisis to facilitate the process when transitioning to a healthy adjustment, considering that believing in something higher has a calming effect, attenuating a negative emotional load arising from adverse events, enabling individuals to more effectively manage events that could break one's Self . Other protective factors the women reported included having hope in the work they perform, a finding that other researchers highlight, and being enchanted with farming, summarized as a project of work and life, valuing the farmers' customs and knowledge in the development process, encouraging their dream of living with freedom and the desire to cultivate their own land, master their work, and live with their families in the countryside with dignity . However, despite this enchantment, even though the conception of resilience adopted here is not intended to justify the inequalities rural women experience, who engender the resources needed to face their work with resilience, this enchantment is admittedly idealized. Hence, there is a need to discuss these aspects from the perspective of gender, which emerges in social sciences to define identities, values, roles, representations, and/or symbolic feminine and masculine attributes not as an effect of "nature", but as the product of different ways of socialization produced and reproduced by the individuals, highlighting the asymmetry of social relations between men and women . From this perspective, it is apparent that starting in adolescence, men are more likely to achieve higher levels of education and are expected to inherit their families' rural possessions, while women remain as caregivers . Even though women's participation in the world of work is gradually increasing, this division still exists and reaffirms the sexual division of labor that overlook the work of women, delegating to them specific cultural roles: that of caring serving, which express socio-cultural representations of the society's imaginary and often leads rural women to omit their individual histories and everyday lives, disregarding their aspirations, so that they adapt or reproduce roles constructed by social representations of the roles of homemaker and maternity . This invisibility is incorporated in the female rural imaginary and leads women to devalue their own work, even though they perform the same tasks men perform, such as their fathers and brothers, in addition to the housework. Even if wives and daughters farm, they are considered mere helpers or are seen as the "producer's wife," and when they are paid, they are not allowed to decide about the money they receive , which is Silva BN, Santos JLG, Riquinho DL, Miranda FAN, Souza NL, Pinto ESG. confirmed by this study; approximately 97.7% of the rural women who participated in the QUAN study did not receive pay for their rural labor. However, even though women are linked to the reproductive domain , in a rural context, women are also active in the productive domain, i.e., cultivation and farming, which is often attributed to men. Thus, women have responsibilities in the two spheres, which leads to work overload . Even though the challenges and adversities of the farming life are acknowledged, this is a fact that rural women considered to be a source of strength and joy for being able to overcome adversities, which appears as a resilient behavior to overcome challenges, which when overcome, promote even greater resilience, as the reports show. However, even though the fact mentioned above works as a protective factor, gender roles are implicated in their enchantment with farming because rural women generally do not have a choice about what work they will perform, i.e., they are introduced to the context of agricultural activity since their childhood . Researchers suggest that enchantment may be related to the symbolic value assigned to labor, originated from the social imaginary of being inserted in the productive chain, and feeling beneficial to society, which confers status. It may be reinforced by the association the rural women made between farming and freedom and autonomy, being satisfied for not being submitted to urban standards and pursuits, valuing and being guided by nature, and perceiving that they are transforming their context . Even though farming renders women vulnerable, they find meaning in their work, which influences their QoL . Nevertheless, rural women need to be empowered to face their work beyond enchantment and broaden their knowledge regarding their contexts to propose alternatives and consider it an important instrument to improve their life conditions . The conception of belonging to one's social context was also evidenced as a protective factor, which, in addition to being linked to support networks, portrays a feeling that is surrounded by close bonds established with their community and which is strongly related to the identity of rural women, also reported by other studies . From this perspective, individuals living in cities, for instance, do not understand the feeling of belonging that rural people experience because they do not have their roots in the rural area and do not put their hands and feet on the land, which impedes the establishment of a relationship with the territory they inhabit . The protective factors identified in the reports of rural women can be used in psychosocial interventions --- Conclusion This study analyzes the intersections between QoL and resilience among rural women, showing that when these women develop protective factors, they also become more resilient, a factor that influences and is influenced by their QoL. These factors include spirituality, establishing social support networks, and a feeling of belonging. The results reinforce the need for a more effective Note that the influence of the constructs found in this study would not be understood if only one approach was employed -quantitative or qualitative -highlighting the contribution of this mixed methods design to understanding complex and imbricated phenomena, such as QoL and resilience. The only quantitative association found and limited prediction level could have led to erroneous interpretations regarding the intersections between the constructs. Note that cultural aspects and the participants' level of education limited the participants' ability to answer the instruments in the quantitative stage, a limitation --- Rev. Latino-Am. Enfermagem 2022;30:e3521. overcome by individualized interviews based on Oral History. Hence, the mixed methods design validated the results presented in this investigation. --- Authors' contribution --- All authors approved the final version of the text. Conflict of interest: the authors have declared that there is no conflict of interest.

Paper extracted from master's thesis "Intersecções entre qualidade de vida e resiliência em mulheres rurais: estudo de métodos mistos", presented to Universidade Federal do Rio Grande do Norte, Natal, RN, Brazil.

Introduction COVID-19 spread rapidly throughout the world, with dire consequences for low -resource countries like Nepal [1]. Nepal reported its index case on January 23, 2020, and implemented strict measures like nationwide lockdown, social distancing, and travel restrictions after the isolation of the second case on 23 rd March [2]. The preparedness, readiness, and response status of any country are crucial in identifying, managing, and preventing health crisis such as the COVID-19 pandemic [3]. However, this was not the case for a low-income country like Nepal [4]. The scarcity of resources such as personal protective equipment, health personnel, infrastructure, and other necessary resources tremendously burdened the existing health care services, even during the last wave of COVID-19 [5,6]. Despite the multiple challenges and a fragile health care system, Nepal struggled to manage the COVID-19 crisis by increasing free testing at public facilities and offering free in-patient COVID-19 health services at designated COVID-19 hospitals, which was noteworthy. However, the efforts to contain COVID-19 were not adequate [7,8]. The COVID-19 outbreak raised many public mental health concerns. With unknown dimensions, this disease not only disrupted the lives of COVID-19 patients but also their caregivers [9]. Caregiving involves different types of support, such as emotional, physical, and financial support [10]. Due to the lack of formal caregiving services, family members, also referred to as informal caregivers shoulder the major responsibility of caregiving [11]. Informal caregivers of COVID-19 faced their own exposure risks and increased concerns about self-care and health. When the patient with signs of severe COVID-19 needed hospitalization, caregivers experiences a mix of positive and negative experiences including stigma, change in relationships, fear of contracting the disease, psychological stress, visitation restrictions, economic worries, and grieving, making the caregiving task even more challenging [12][13][14]. Additionally, several studies indicated that caregivers experienced social isolation, chronic stress, psychological distress, and physical and mental health problems exacerbated by the contagious nature of COVID-19 [15][16][17]. This study was conducted during the last wave of COVID19 in a tertiary level hospital, B.P. Koirala Institute of Health Sciences, which had a 100 bed COVID-19 hospital and served as the referral centre in Province1. The institute provided a separate caregiver residential facility for the caregivers of COVID-19 patients admitted to the hospital. Despite being the only residential facility in Nepal, this caregiver facility was an open shaded structure with no separate provision for males and females and had the bare minimum of facilities. Phenomenology is a philosophical method of inquiry that enables researchers to understand the fundamental structures of experiences and is useful to explore previously unknown and overlooked experiences [18]. Since, this was the only caregiver residential facility in Nepal, it was crucial to understand the experiences of caregivers. This understanding will help identify specific areas of intervention to offer greater support to family caregivers during and beyond the pandemic. Therefore, we used this descriptive phenomenology approach to gain an in-depth understanding of caregiver experiences and the challenges of informal caregiving for COVID-19 patients during the pandemic. --- Materials and methods --- Ethics Ethical clearance was obtained from the Institutional Review Committee of the B. P. Koirala Institute of Health Sciences, Nepal . Informed written consent was obtained from each of the participants before enrolling in the study. We followed the consolidated criteria for reporting qualitative research guidelines in this study [19]. --- Research team and reflexivity The research team comprised multidisciplinary members from the hospital, the Department of Psychiatric Nursing, and the Department of Psychiatry, BPKIHS, as well as the Department of Psychiatry, Patan Academy of Health Sciences, Nepal. The lead author, Prekshya Thapa , is a psychiatric nurse with previous experience conducting qualitative studies; she conducted the data collection. The data collection process was assisted by Roshni Thapa , a community nurse, and Pratik Uprety , a medical doctor, who recorded the interviews. The data collection was supervised by Gayatri Rai , the hospital's matron, and its quality was ensured by the other researchers not affiliated with the hospital, namely Sami Lama and Nirmala Pradhan . Madhur Basnet , a language expert and psychiatrist, was involved in translating the interview guide and overseeing the research design, data analysis, and manuscript preparation. Finally, Nidesh Sapkota , who was external to the team, reviewed the data analysis process. None of the research team members were previously known to the participants. --- Study design This qualitative study used a descriptive phenomenological approach [20] to understand the family caregivers' experiences, as less is known regarding care experiences related to a newly emerging disease . --- Study setting The study was conducted at the COVID-19 hospital of the B.P. Koirala Institute of Health Sciences. This hospital has 100 beds and was consistently occupied, especially during the surge of omicron variant of COVID-19. As a tertiary referral centre in Province 1, it caters to a diverse population, leading to a heterogeneous sample in terms of caste, ethnicity, age, socio-economic status, ecological regions and other demographic variables. --- Sample population and sampling technique Thirteen family caregivers of COVID-19 patients admitted to the ward who assumed primary responsibility and stayed for at least one week at the COVID-19 caregiver residential facility were purposively selected from April 15 th to June 30 th , 2022. Sample size estimation was determined based on a review of the literatures, which typically recommends using a smaller sample size sufficient to answer the main research questions [21,22]. Sampling continued until no new information emerged and caregivers provided maximum information, ensuring data saturation on the phenomenon. Maximum variation sampling, considering factors such as age, gender, educational levels, occupational status, and residential location, was used to obtain comprehensive and diverse data [23]. --- Data collection The caregiver was contacted by PT, who explained the purpose of the study, read out the participant information sheet, and obtained the informed consent of the participant. After taking written consent for participation in the study, he or she was asked about socio-demographic information, and pre-structured open-ended questions were asked using an interview schedule guide in a separate room nearby the COVID-19 hospital, maintaining confidentiality and comfort by the lead investigator herself . A face-to-face interview was conducted to encourage the participants to discuss their experiences of caring for a patient with COVID-19, using the interview guide developed and validated by the research team for this purpose . The interview guide was piloted on two of the caregivers of COVID-19 patients who were not admitted to COVID hospital, and questionnaire modifications were made based on the feedback from pilot testing. Physical distancing and precautionary measures were taken into consideration while conducting the interview. The participants were allowed to express their feelings openly, and when necessary, probing questions were asked to encourage discussions and more clarifications. Each interview lasted 35 to 45 minutes and was audio-taped by RT and PU and later transcribed. Field notes, including non-verbal communication, were taken by RT and PU. Permission was also obtained to audio record the one-on-one interviews. All interviews were conducted in Nepali, the mother tongue of the participants and the data collector. --- Data analysis Audio recordings of the interviews were transcribed verbatim in Nepali and later translated into English by the lead researcher, PT. Inductive thematic analysis [24] was conducted by identifying and reporting the themes generated from the data. This process involved familiarizing oneself with the data, coding the information, searching for appropriate themes, reviewing the identified themes and defining and naming the themes. MB and PT coded the data, and samples from the transcripts were cross-checked by SL. No software was used for the data analysis. The analyzed themes and categories were shared with research team members for their final review. The final themes were discussed among the investigators and were then confirmed by all team members. Trustworthiness in the study was maintained according to the criteria proposed by Lincoln and Guba [25]. The co-investigator, RT, contacted the participants by phone and reported the findings to ensure that the extracted codes were consistent with their viewpoints. For dependability, the study results were reviewed by a research expert, NS, who was external to the research team. To ensure conformability, peer checks were done by three co-investigators to ensure the accuracy of the descriptions and categories. The transferability of data was achieved by the maximum variation of the samples, different levels of age, education, occupation, and family relationship with the patient. --- Results --- Socio-demographic characteristics of the caregivers Among the 13 caregivers, the six were male and seven were female. The age of the caregiver ranges from 20 years to 65 years. Most of them were married, and had completed secondary education and had agriculture as their occupation . --- Qualitative result findings We summarized our findings and categorized the themes under five domains: challenges encountered, changes in physical and mental health needs, changes in role, relationship and responsibilities, positive experience and strategies that could ease caregiving tasks . 1. Domain of inquiry: Challenges encountered. Theme 1: Financial challenges. Almost all the caregivers reported that increased expenses have created difficulties in managing the situation. One of them said that they were not prepared for the expenses to be this high and had to manage them by borrowing money. "Now the expenses for staying for 7-10days are likely to be around 70-80. We came with a little sum of 10-20 thousand as a loan for this seriously ill patient. We are feeling burdened due to financial difficulties."-IDNO.12. They also expressed that although the treatment inside the COVID hospital is free of charge, the cost of medicines and other necessary supplies has to be covered by themselves. "Although they say the treatment is free, we have to pay for the expenses of the medicines ourselves."-IDNO.1. Another caregiver expressed that some investigations are required to be done outside the COVID hospital, which is very expensive and unaffordable. "They ask for blood tests at the hospital, but they instruct them to do the test outside. The charges are significantly higher when done outside, and now our money is finished." -IDNO.2. The increased financial burden had additionally hampered caregivers with limited sources of income who came from remote places."4-5 lakhs have already been spent, including medicines and travels. You see, we have a limited source of income. In the hills, you cannot easily sell your land; people prefer to leave. We have to take loans; need to take loans for treatment." -IDNO. 7. Theme 2: Communication challenges. Another major challenge encountered by most of the caregivers was communication problems such as uncooperative caregivers and unreliable information, as well as improper communication from health personnel. One caregiver reported greater difficulty when the other caregivers seemed indifferent and less cooperative. "Some friends come, but do not inquire about their whereabouts, do not speak even when called; and do not care about anything. It's difficult when this happens."-IDNO.2. Some of the caregivers also expressed the feeling that communication from the health personnel was doubtful and unreliable as they were misinformed about the patient's condition. "Now, it's like this here-one doctor says one thing and another says other thing. There is one doctor, I do not know his name, who scares us a lot when we ask him about the condition of the patient."-IDNO 4 Additionally, caregivers expressed concerns about the final death rituals. They mentioned that if the patient dies in the hospital, nobody will care, and they will not be able to perform the necessary rituals. One caregiver explained, "If the person dies at home, we could perform our rituals. If they die in the hospital, there's nothing we can do. We face stigma, discrimination, and neglect. People stay away and avoid us. It feels like there's no recognition of the dead person, no care, whether it's the death of a person, or a dog, or a cat." -IDNO.7". Theme 4: Gender disparities and safety concerns. Female caregivers perceived more stress and vulnerability due to mistrust and fear of unknown male caregivers' intentions, as they had to share the same roof in the open shade of the residential facility. One caregiver expressed, as "What to say, it's very difficult to stay for a female alone. Not all males consider females to be their sisters. Sometimes I feel terrified. When I'm alone, I have to stay awake the whole night. The day before, I was the only female here and could not sleep the whole night."-IDNO.2. In contrast, another female caregiver expressed that she felt more secure due to the presence of male caregivers and pointed out that it might be more challenging to be the only caregiver since the facility is located near a jungle area and is not well facilitated. "We gather courage to stay because there are other males here. Otherwise, there is no facility, and it's really hard for females to stay. Sometimes, when alone, it feels like a jungle and becomes scary. It feels unsafe. I cannot stay alone: it's very quiet, and there isn't much movements." -IDNO.12. Fear of theft, looting, and other criminal acts due to the open space and the remote location were some of the caregivers' serious concerns of the caregivers. "This place is so open; it feels uncomfortable when I'm alone." -IDNO.8. Theme 5: Caregivers' struggles under visitor restriction policies. Most of the caregiver complained that the visitor restriction policy made them unassured and uncertain about the patient's condition inside the hospital. This created doubt, mistrust, and confusion among the caregivers. "I am not able to see how my mother is doing inside. My anxious heart fills with doubt, wondering if she might be dead, especially when I cannot see her. Sometimes I feel like taking her home instead, said IDNO1". Theme 6: Substandard caregiver residential facility and services. Caregiver reported inadequate basic facilities such as restrooms, lighting, water and sanitation. They expressed, "There is only a toilet, no bathroom; you have to bathe in the toilet only if you need it." -IDNO 3. Another caregiver stated, "The toilet has a lock, but it's broken." -IDNO 6. IDNO.1 mentioned, "The toilet and bathroom are very dirty. We could clean, but there are no cleaning supplies."-IDNO 1. Improper waste disposal and the presence of mosquitoes at night posed health risk for caregiver. One caregiver reported, "Wastes are disposed of here; no vehicle collects them. Mosquitoes are everywhere because of the waste."-IDNO 4. Additionally, caregivers faced stress due to the canteen, lab services, and drug store being located outside the COVID hospital premises. They complained that we don't get good food here in Dharmashala. We have to go to the market. The available food is not clean; sometimes the vegetables are dirty."-IDNO13. Another caregiver mentioned, "Medical services are far for everyone. There's also a problem with drinking water here." IDNO 6. Furthermore, caregiver faced difficulties accessing emergency labs and other services due to lack of lighting. " One caregiver shared, "There should be lighting, but there's none. Yesterday I had to go to the emergency lab at 12:30am, and it's far. The distance is manageable, but without light, it's scary. The path isn't safe, especially without light."-ID No. 8. --- Domain of inquiry: Changes in physical and mental health. Theme 1: Emotional turmoil and mental strain. Almost every caregiver expressed feeling stressed, burdened, and overstretched as a result of caregiving. Caregivers mentioned being physically exhausted and drained, stating, "There's a lot of exertion; need to go so far to buy medicines and bring meals for patient; there's a lot of tension. There's so much tension, my god."-IDNO5" "We feel anxious when the patient is sicker."-IDNO 2 They expressed a sense of hopelessness as "I am hopeless now. I shall bear the treatment as long as my condition and finances allow, and then I will take her home if I am not able to bear the costs. Now also, I am sustaining on donated money."-IDNO 1. "The fear of contracting the virus and uncertainty regarding the patient's condition made caregivers worry. One of them shared, "I feel tensed about my mother-in-law's condition. My husband is not here; he's abroad, and he's also worried. After all, she is his mother, who gave him birth. I feel worried if she might not get well. There's also the fear if I might get COVID myself."-IDNO 10". "It's scary; it feels like if something happens to a patient. It's scary either way."-IDNO.2. Theme 2: Physical strain and neglected health: The toll of exhaustive caregiving responsibilities. Almost every caregiver felt that their basic physical needs, such as food, sleep, and rest, were unmet due to exhaustive responsibilities. They expressed that they had to be alert anytime as they could be called from the hospital at any time, which hampered their sleep. "Now, it's difficult to sleep; they ask for medicine anytime, irrespective of night or daytime, and cannot sleep just like that." -IDNO 1. Headaches, backaches, and fatigue were also reported by caregivers as a result of overexertion and lack of rest. "I am feeling exhausted with a headache and body ache."-IDNo.11. Because of inadequate facilities near the COVID hospital, caregivers had to run here and there for necessary supplies and related investigations, making them physically drained and restless. "It's hustle and bustle for two people; they are calling at night and calling during daytime as well."-ID No. 1. "You have to go here and there now and then."-IDNO 7. --- Domain of inquiry: Changes in roles and relationships. Theme 1: Burden of altered roles: caregivers' struggles with evolving family responsibilities and guilt feelings. Caregivers have reported that they are unable to fulfil family responsibilities due to their caregiving duties. One caregiver expressed her struggles, saying, "It's difficult to stay here, leaving my children behind at home. One has college in the morning and another has school in the afternoon, so there's a problem regarding who will prepare the meals. They end up eating the meals prepared in the evening the next morning as well, and that's a source of stress."-IDNO 10. Another caregiver shared her feelings of guilt, as she could not properly care for her COVID-positive mother due to the contagious nature of the disease. She explained, "Yesterday, I was allowed to meet mom. She went out without a mask to bask in the sun, and I had to move away when she tried to come near me, which made me very uncomfortable. I could not make her hair help her with her dresses while sitting beside her. I felt like she approached me thinking I would, but I had to keep my distance due to the fear of COVID-19, and that made me feel somewhat bad." -IDNO 1". Theme 2: The impact of caregiving on employment and livelihoods. Caregivers also expressed that their caregiving role has hampered their work roles and responsibilities and resulted in financial losses. One participant shared, "My business is also facing financial losses. There are losses here and there as well. It would be okay, at least if my relative gets well."-IDNO3. My husband was working at the site when suddenly my mother tested COVID positive for COVID-19, so he had to leave his work and come here" -IDNo.1. --- Domain of inquiry: Post-COVID resilience and positive adaptation: Navigating challenges, finding strength, and acknowledging support. Despite the challenges, COVID caregivers also expressed positive experiences, such as normalizing COVID-19, adapting and positively coping, appreciating the hospital facilities and services, and acknowledging the support from family. Theme 1: Post Covid concerns and positive adaptation. Several caregivers voiced post-COVID concerns, particularly regarding the risk of infection after receiving the COVID-19 vaccination. One caregiver shared a personal experience, stating, "Two months after vaccination, I fell ill and remained bedridden, affecting my daily activities" . Similarly, another caregiver expressed apprehension, mentioning, "I heard that vaccination can lead to infection, and my mother indeed contracted the virus after being vaccinated."-IDNO.10. Furthermore, caregivers acknowledged a decline in the intensity of COVID-19-related stigma in society. One participant remarked, "Previously, COVID-19 was highly stigmatized, although it still carries some stigma, it is not as severe as before."-IDNO5. Caregivers have normalized their experience with COVID19, as stated by IDNO6. "After understanding it better this time, I felt COVID-19 is not as dangerous as we thought. It's not that dreadful a disease, provided we exercise and take care of our diet. It's our own weakness to perceive it that way."-IDNO 6. Some of the caregivers also expressed their positivity even in difficult situations, acknowledging that there is no other alternative but to stay positive. "Whatever it may be, we should handle it in a good way, keep laughing and talking as mentioned by ID NO. 13. This observation highlights the evolving social perceptions surrounding COVID-19, indicating a shift in public attitudes while underscoring persistent concerns and misconceptions among caregivers. Theme 2: Appreciation of services and support. A caregiver expressed satisfaction with hospital services and the care provided by healthcare providers, despite the discomfort of staying in caregiving home. They stated,-"Although there is difficulty in staying here, the way they are taking care of the patients and treating them is satisfactory; it's good."-IDNO10." Another caregiver mentioned, "They have made this facility, and so we need not wait outside, need not sit in the sun. We are staying here like a family, and it's fine. We do get a bit cold, but it's OK. When we have no other option, we need to be satisfied with whatever we have."- IDNO 11". Yet another caregiver appreciated the opportunity to visit the patient inside the hospital, saying, "My patient is inside, and they allowed one visitor to go in today. We were concerned about the condition inside, but we saw that our patient was OK. We appreciate this."-IDNO4". One caregiver acknowledged the support from their family and expressed that "I do have my family responsibilities, but my husband is helping me in this".-IDNO 2. 5. Domain of inquiry: Strategies that could make caregiving task easier. Nearly every caregiver emphasized the need for improved hospital facilities and services to facilitate caregiving tasks. They highlighted the importance of maintaining hygiene and sanitation in caregiving, which is crucial for managing health and preventing illness. Similarly, caregivers expected proper communication and a positive attitude from healthcare providers to support them in their caregiving roles. Theme 1: The need for essential services within hospital premises for caregivers. Many caregivers emphasized the importance of having essential facilities like an emergency lab, medical store, bill counter, and canteen facilities within the hospital premises. One caregiver stated, "It's quite far if you need to ask for medicines. It should be near as much as possible."-IDNO1 "The issue with getting medicines is that it's far; it would be better if nearer."-ID No2. IDNO 5 added, "The drug store should be nearby." IDNO 6 suggested, "It would be easier if you owned a pharmacy! It would be better if we did not have to go far to get medicines; they should be available here." IDNO 7 emphasized the need for a pharmacy within the hospital premises, especially for outsiders, stating, "There should be a pharmacy here so that we don't have to go elsewhere at night. People from outside might not know the place and could end up going in the wrong direction. The facilities should be right here". Additionally, another caregiver emphasized the importance of having the bill counter and canteen facilities nearby. IDNO 5 said, "The bill counter should be close to the hospital, and the canteen should provide healthy and affordable services. We shouldn't have to go far to pay bills." IDNO 8 echoed this sentiment, stating, "We have to travel quite a distance to get meals; it would be much better if everything we need was available around here". Additionally, another caregiver emphasized the importance of having the bill counter and canteen facilities nearby. IDNO 5 said, "The bill counter should be close to the hospital, and the canteen should provide healthy and affordable services. We shouldn't have to go far to pay bills." IDNO 8 echoed this sentiment, stating, "We have to travel quite a distance to get meals; it would be much better if everything we need was available around here". Theme 2: Promoting safety and comfort: Reconstruction and maintenance of facilities for secure and hygienic caregiving environments. Caregivers expressed that they faced multiple challenges due to the open caregiver residential facility. The open facility should be made more secure in order to ensure caregivers feel comfortable. Caregiver also mentioned that the existing hospital facilities need maintenance, and that open spaces should be partitioned for the safety and comfort of the caregivers. One caregiver stated, "It would be better if the sleeping place is well organized. This area is open and needs to be organized. It would be better to have separate rooms with locking doors." -IDNO 5. Additionally, concerns were raised about the adequacy of the existing caregiving home, especially during surges in cases. "This facility is okay for the time being, but if the COVID cases suddenly surge, this space won't be enough. It's better to have one more building. Also, lots of ants come here, and it's better to get it plastered." -IDNO 1Maintaining sanitation and hygiene was also highlighted by caregivers. They suggested, "It would be better if the dustbins are emptied daily. Also, separate toilets for ladies and gents would be better." -IDNO 3". Theme 3: Effective communication and positive attitude. Caregivers expect clear communication from health personnel, which would assure them regarding patient condition. One caregiver expressed, "When we hand over our meals to our patients , we should be clearly informed whether our dear ones have eaten or not. Sometimes they just say that they did not eat, but we should also be told why they did not eat. We cannot have trust if we are not communicated well, as we have not seen inside." -ID NO 4". Caregivers also expect a positive attitude from health personnel, which would make them feel better. They expressed, "It would be better if we were seen positively when we went there. I hope we are not viewed negatively as caregivers of such sick people. I wish doctors treated our patients without negligence, as we have entrusted our patients to their care, and we would be grateful if they were treated well."-IDNO 6. Theme 4: Visitation allowance to provide comfort and support. Some caregivers have expressed their opinion that they should be compulsorily allowed to visit their patients inside the hospital, which would significantly improve the patient's well-being. One caregiver stated, "There should be one to two visitors allowed for each patient here. They should be nearby, inside with the patient. It feels meaningless when the patient is inside and we are out here. The patient would feel comforted by seeing her family members and would have high morale. She would also stay there satisfied. We could also ask her what she needs and wants to eat."-IDNO 7. Another caregiver emphasized, "My point is that at least one person should be allowed to stay with the patient." -IDNO5. Some caregivers also expressed that the presence of a caregiver comforts the patient and improves their overall well-being. One of them stated, "Whatever the case, we want to go in and see the patient. It's ok, even if allowed once a day. Patients also feel bad if relatives don't visit. They would feel comforted if we visit."-IDNO 6. --- Discussion The present study aimed to explore the experiences of caregivers residing in COVID-19 caregiver residential facilities while supporting their hospitalized relatives. The findings were summarized and categorized into five domains of inquiry. Informal caregivers play a vital role in in low-and middle-income countries like Nepal [11]. However, they face numerous challenges in providing care [26]. Our findings revealed various difficulties, including financial challenges, communication issues, stigmatization, insecurity, visitation restrictions, and problems related to substandard caregiver accommodations. Almost every caregiver reported financial burdens consistent with the other s*tudy findings [12,27] likely due to intermittent lockdowns causing unemployment, especially for daily wage-based labourers and vulnerable populations like disadvantaged and marginalized groups [28,29]. Additionally, caregiving responsibilities often prevented them from fulfilling work duties, leading to job loss or a lack of payment for daily wage workers. Nepal's already overstretched healthcare system was ill-equipped to handle the crisis, resulting in increased treatment costs and substantial out-of-pocket expenses [27]. Female caregivers felt unsafe due to shared accommodations, with one caregiver expressing fear as a sole woman among male caregivers. Concerns about theft and personal safety added to their emotional burden. To address this, we recommend separate male and female rooms with secure locking systems in caregiver facilities for safety and peace of mind. Our study, consistent with others [30,31], highlighted caregivers' experiences of stigma, discrimination, neglect, and lack of support from relatives, the community, and health care personnel. Visitation restrictions and communication gaps exacerbated caregivers' uncertainty about patients' conditions, leading to negative emotions such as doubt and mistrust, consistent with other studies [32]. This emphasized the need for clear communication and information dissemination about hospital policies and patient status [32]. The open and shared caregiver accommodations posed challenges in maintaining basic hygiene, contributing to caregiver distress. Improper waste disposal, inadequate facilities, and distant services added to physical exhaustion and burden. The study identified changes in caregivers' physical and mental health needs. Constant vigilance due to shared accommodations, fear of contracting the virus, financial burden, unfulfilled roles, stigma, and inefficient management led to negative emotions such as fear, anxiety, helplessness, and burnout, consistent with previous studies [33][34][35]. Early psychological intervention is crucial to promoting emotional release and mental well-being among caregivers during epidemics. The study also highlighted the burden of assuming additional responsibilities and adjusting to a new caregiving routine, which is consistent with other studies findings [36,37]. Female caregivers, particularly in Nepal, faced challenges balancing caregiving with household responsibilities, while male caregivers found their occupational roles affected [38]. Addressing discriminatory gender norms is essential to encouraging equal sharing of caregiving responsibilities [39]. In conclusion, our findings emphasize the need for targeted support, improved communication, and gender-sensitive policies to alleviate the challenges faced by caregivers in similar contexts. --- On the positive note On a positive note, this study was conducted during the omicron wave of COVID-19 where the cases spiked in a very short duration [40]. However, the public was more aware during this phase of the pandemic and had somewhat adapted to the difficult situation created by the pandemic. Some of the positive experiences were as follows: they had adapted positively to the COVID-19 situation and were appreciative of the governments and health personnel's efforts, as well as the existing facilities of the hospital. Caregivers also acknowledged the support from relatives, the community, and the nation, which can be seen as silver linings and is consistent with other study findings [30,41]. --- Recommendations from the caregivers This study has unveiled several key factors that can improve the challenges faced by caregivers during the health crisis, based on the experiences of those attending COVID-19 hospitals. Providing amenities such as basic health facilities, medical stores, bill counters, and canteen services within the hospital premises can substantially reduce the need for caregivers to travel, thereby minimizing physical exhaustion and burden. Moreover, caregivers stressed the urgency of restructuring the existing open accommodation facilities into separate cabins equipped with locking systems to ensure their safety. Caregivers reported feelings of fear, vulnerability, and physical threats due to the open and distant accommodation setup. Hence, they suggested the establishment of a secure environment within the hospital premises, along with maintaining proper sanitation and hygiene in the caregiving area, to promote both physical and emotional well-being among caregivers. Additionally, caregivers emphasized the importance of health personnel displaying a positive attitude and practicing clear communication, echoing findings from previous studies [42,43]. Another vital suggestion was to permit at least one visitor to see their ill relatives inside the COVID-19 hospital every day. Caregivers highlighted that such visits could enhance patient comfort, allowing them to express their emotions more freely. Moreover, caregivers themselves would find solace in being able to see their relatives, contributing to their overall satisfaction and well-being. --- Strengths and limitations This study was conducted in the only caregiver residential home affiliated with a COVID hospital operated by the government in the eastern region of Nepal, providing a distinctive perspective on caregiving experiences. While existing literature has delved into the challenges faced by family caregivers of COVID-19 patients, our research is pioneering in its focus on caregivers residing within a dedicated caregiving facility within a COVID hospital. However, there are notable limitations to our study. Firstly, despite our best efforts to ensure diverse participant selection, the findings may not encompass the full range of experiences prevalent among the wider population. Future research could explore contextual factors such as income, age, religion, education, social support, and residential status, shedding light on the varied experiences of caregivers of COVID-positive patients. Secondly, a key limitation is the timing of our research, conducted during the omicron variant of COVID-19. At this point, society had somewhat normalized and adapted to the crisis situation, potentially influencing the caregivers' experiences. It is crucial for subsequent studies to investigate the earlier waves of the pandemic, capturing the unique challenges faced during the initial stages when the situation was more acute. Furthermore, our study highlighted that caregiver dealing with elderly patients, postpartum mothers, or those facing comorbid illnesses experienced heightened stress and burden. To gain a comprehensive understanding, future research should specifically focus on exploring the additional caregiver burden associated with patients having comorbid conditions alongside COVID-19. Addressing these limitations can contribute significantly to the depth and breadth of knowledge in this critical area of study. --- Implications to policy and practice Implementing a range of initiatives can significantly enhance the caregiving environment for COVID-19 patients in hospitals. Financial support mechanisms, including insurance coverage and subsidies, can be tailored to ease the economic burdens faced by informal caregivers. Healthcare professionals can undergo comprehensive communication training, ensuring empathetic and clear interactions with caregivers. Public awareness campaigns targeting the reduction of stigmatization against both COVID-19 caregivers and patients can be launched. Hospitals can allocate funds to improve accommodation facilities, focusing on security and hygiene. Visitation policies can be reviewed to strike a balance between patient safety and the emotional well-being of caregivers, possibly incorporating flexible hours or virtual options. Initiatives like formal recognition programmes and awards can acknowledge exemplary caregiving efforts, fostering pride and motivation among caregivers. Research collaboration with institutions can inform policies, ensuring ongoing improvement, while national guidelines for crisis preparedness, including caregiver training, can enhance their confidence and efficiency during emergencies. These concerted efforts can create a compassionate and supportive caregiving environment within COVID-19 hospitals in Nepal. --- Conclusion This study demonstrates that family caregivers attending to COVID-19 patients in hospitals and residing in caregiver residential facilities face numerous challenges. These caregiving challenges not only affect their physical health needs and activities of daily living but also lead to --- PLOS ONE emotional and psychological issues such as fear, anxiety, and stress. Stigma and communication problems further exacerbate the caregivers' conditions. However, amid these challenges, caregiver also reported positive experiences of coping and adapting to the difficult situation, as well as appreciating the existing facilities of the hospital. Our study revealed several key factors that can facilitate care, including improving hospital services and facilities, maintaining a positive attitude and effective communication, providing allowances for visitation to support the patients, and ensuring sanitation in caregiving homes. These facilitators could play a crucial role in managing future health crises. --- Supporting information S1 --- --- Data curation: Prekshya Thapa, Madhur Basnet. Formal analysis: Prekshya Thapa, Sami Lama, Nidesh Sapkota, Madhur Basnet. Investigation: Prekshya Thapa, Roshni Thapa, Madhur Basnet. Methodology: Prekshya Thapa, Sami Lama, Gayatri Rai, Nidesh Sapkota, Nirmala Pradhan, Roshni Thapa, Pratik Uprety, Madhur Basnet. Project administration: Prekshya Thapa, Roshni Thapa, Pratik Uprety, Madhur Basnet. Resources: Prekshya Thapa, Roshni Thapa, Pratik Uprety, Madhur Basnet. Software: Prekshya Thapa, Madhur Basnet. Supervision: Gayatri Rai, Nidesh Sapkota, Madhur Basnet. Validation: Prekshya Thapa, Sami Lama, Nidesh Sapkota, Nirmala Pradhan, Madhur Basnet. Visualization: Prekshya Thapa, Madhur Basnet. Writing -original draft: Prekshya Thapa, Sami Lama. Writing -review & editing: Prekshya Thapa, Sami Lama, Gayatri Rai, Nidesh Sapkota, Nirmala Pradhan, Roshni Thapa, Pratik Uprety, Madhur Basnet.

Informal caregivers played a significant role in caring for COVID-19 patients during hospitalization in Nepal. This study aimed to understand the experiences of family caregivers attending to their relatives in a COVID-19 hospital in Nepal.A descriptive phenomenological approach was adopted to understand the caregiver's experience in supporting their relatives admitted to the COVID-19 hospital of the BP Koirala Institute of Health Sciences (BPKIHS). Thirteen caregivers of COVID-19 positive patients were purposively selected from April to June, 2022. Face-to-face interviews were conducted at a caregiver residential facility using the interview schedule developed by the department for the purpose of data collection after obtaining ethical clearance from the Institutional review committee of BPKIHS. Data were audio recorded and manually analyzed.Among the 13 caregivers, six were male and seven were female. The findings were categorized into five domains: challenges encountered, changes in physical and mental health, changes in roles and responsibilities, positive experiences, and strategies to ease caregiving tasks. Major challenges included financial burdens, communication problems, stigmatization, Insecurity, substandard accommodation, and visitor restrictions. Caregivers reported negative emotions, unmet physical health needs, and shifts in family and occupational roles. Despite these challenges, caregivers attempted to cope positively and acknowledged the efforts of healthcare personnel and other family members. Suggestions for improving caregiving included providing essential medical and basic services within the

Introduction Trade in wildlife poses one of the greatest threats to biodiversity in the tropics [1] and is receiving attention from researchers, conservationists and policymakers alike [2,3]. Southeast Asia, in particular, is a "hub" for wildlife trade, both legal and illegal [4][5][6]. Vietnam, for instance, has an illegal trade that generates estimated annual revenues of US$67 million [7]. Vietnam contributes to this trade in several ways. Like other countries in the region, it is a producer and thoroughfare of wildlife traded internationally. It is also, however, a significant consumer of wildlife, particularly wild meat, in urban restaurant settings [8,9]. To meet this demand, poaching of wildlife is widespread threatening regional and international biodiversity [1,9,11]. Generally, interventions to tackle illegal and potentially unsustainable trade in wildlife have focused on limiting supply, typically through the enforcement of regulatory measures [2,12]. In Vietnam, an array of legislation regulating much of the wild meat supply chain has been enacted. Decree 32/2006/ND-CP and Decree 82/2006/ND-CP, for example, prohibit the harvest, trade, use and consumption of all species protected under Vietnamese Law. However, ineffective enforcement, the presence of increasingly organised criminal networks and corruption undermine these efforts [7,13,14]. As such, the Government of Vietnam promotes captive breeding of wildlife as an alternative solution to the exploitation of wild populations and has established a policy framework regulating the increasing number of commercial wildlife farms that produce a variety of taxa, many of which are consumed as meat [15]. According to economic theory, flooding the market with farmed substitutes for harvested wild species will lower prices and therefore incentives to poach from wild populations [16]. However, doubts have been raised over the appropriateness and efficacy of farming as a conservation tool following findings that wild meat consumers prefer wild-sourced meat over farmed substitutes [17] and findings that wildlife farms continue to launder wild-caught animals [15,18]. In light of both the limitations of these approaches and the rapidly rising demand for wildlife products [6], attention is turning to demand management and tools such as awareness raising and social marketing [5,9,20,21,22], which attempt to engineer desirable conservation outcomes e.g., reduced demand. However, for such tools to be effective in changing consumer behaviour an in depth understanding of consumer choices among target audiences is first needed. In this paper we explore consumer preferences for wild meat and the drivers of consumption among residents of Ho Chi Minh City, Vietnam. We employ a mixed-methods research approach encompassing a hypothetical choice modelling survey, to provide an indication of the relative impact of different attributes of wild meat on consumer choice, and a series of qualitative interviews with wild meat restaurant staff and consumers, to explore the drivers of consumption and consumer choice in depth. --- Material and Methods --- Study site This research was conducted in Ho Chi Minh City , one of several urban centres in Vietnam where wild meat consumption has been quantified. A recent survey found that 50% of sampled HCMC residents had used wild animal products during their lifetime, 75% of which were wild meat [23]. --- Ethics statement Research ethics clearance was obtained from the Research Ethics Advisory Group at the University of Kent. Informed verbal consent was obtained from all respondents prior to participating in the study. Verbal consent was deemed appropriate given the clandestine and sensitive nature of the survey content, and was specifically approved by the institutional ethics board. The date and time of the interview or questionnaire, along with the name of the researcher obtaining consent, were documented prior to commencing the interview or survey. All data collected were anonymous. A business visa was acquired for Vietnam and the research was carried out in association with the Institute of Tropical Biology, part of the Vietnam Academy of Science and Technology, based in Ho Chi Minh City. --- Choice modelling Choice modelling is a survey based stated preference valuation technique that makes use of hypothetical markets to infer respondents' preferences and demand for goods [24]. A respondent is required to make choices between different hypothetical product alternatives, each described by varying levels of the same attributes, and each providing different levels of utility to the respondent. Consistent with random utility theory and Lancasterian consumer theory, choice modelling is grounded in the assumption that an individual will choose the product alternative whose attributes provide him the greatest utility [25]. It can consequently be used to infer information on which attributes influence consumer choice, the implied ranking of these attributes, and marginal willingness to pay for any increase or decrease in a significant attribute [24]. In our study we used this method to attempt to understand consumer preferences for various attributes of wild meat by presenting respondents with hypothetical menus containing several wild meat products, each defined by different levels of the same product attributes. The choice survey was a D-efficient Bayesian design produced using Bayesian prior distributions generated from the results of a small-scale pilot [26], also carried out in HCMC. The final design consisted of 10 choice scenarios, which we blocked into two groups of five . Respondents were randomly assigned to either A or B. Each choice scenario, or 'menu', contained three 'wild meat' options and a 'status quo' option and for each the respondent was asked to choose their preferred option . We selected attributes and their levels following a review of relevant literature and considering data collected through informal interviews with wild meat consumers and restaurant staff. Price: Price is universally recognised as a primary determinant of demand, with demand falling as prices rises. Nonetheless, there is evidence suggesting that some consumers in Vietnam value and demand wild meat because of its expense [17]. To explore this, we included price as a five level attribute with the levels reflecting the variety in market-prices for different types of wild meat. The context for the choice task was given careful consideration. As wild meat consumption in Vietnam is a social activity [9], we presented price to respondents as the price for enough meat for them and three 'friends'. This also allowed us to control for different choices that might be made in different consumption contexts. Source of the meat: A number of taxa consumed as wild meat are commercially farmed in Vietnam, including deer , soft-Table 1. Description of all attributes and their levels chosen for the choice modelling survey . --- Attribute Description Price Price for enough meat for four persons divided into five levels: 300,000 , 600,000 , 1,100,000 , 2,200,000 , 4,500,000 . Wild meat-type Type of wild meat divided into five levels: wild pig , deer , civet , king cobra , pangolin . --- Source Whether the meat was sourced from the wild or from a farm . doi:10.1371/journal.pone.0134787.t001 shelled turtle , king cobra and Southeast Asian porcupine [18]. However, the use of farmed wildlife as a substitute has been questioned by some conservationists given the apparent preference for wild-sourced meat by consumers [17]. To investigate this further, we included meat source as a binary attribute: wild and farmed. Wild meat-type: Wild pig , deer and civet were chosen as they were among the most commonly eaten wild meat-types, as reported in recent surveys [9,23]. Pangolin and king cobra snake were selected as the fourth and fifth levels because both are consumed for their meat in Vietnam despite being protected under Vietnamese Law and listed in Vietnam's Red Data Book of threatened species. Respondents also perceived them to be 'rare', a characteristic that Drury [17] suggested was important to consumers. This is significant because where a species' value is tied to its rarity it can result in an extinction vortex as its value rises all the while stocks deplete [27]. Understanding how consumer behaviour responds to perceived rarity is, therefore, important. 331 responses were generated . Respondents were opportunistically sampled while seated in restaurants and cafes; this was thought to be the only realistic sampling strategy given the clandestine nature of the survey content. In order to target wild meat consumers, each sampled individual was asked if they had ever eaten wild meat. Those who responded affirmatively, and non-consumers who indicated that they would consider consuming wild meat in the future were invited to complete the survey. While our sample was unbalanced in terms of gender, it does reflect the typical urban wild meat consumer as identified in previous studies [9]. We began our model estimation by employing a Multinomial Logit Model to determine aggregate preferences. We then used a Latent Class Model to take account of respondent heterogeneity. Model specification. In terms of model specification the random utility model has two parts: an observable deterministic component; and an unobservable random component. We assume a respondent n makes one choice from a finite set Z. The utility respondent n obtains from selecting an alternative i is: Uni ¼ bXni þ εnið1Þ where U is the utility obtained by respondent n, β is a vector of parameters to be estimated, X is a vector of the attributes from the survey and ε is the unobservable random component assumed to be a type 1 extreme value distribution. To identify the extent of preference heterogeneity we assume that within the population there are a finite number of segments S such that respondent n belongs to segment s . Given this we can re-express the utility that respondent n obtains from selecting an alternative i as: Unijs ¼ bsXni þ εnijsð2Þ such that the utility parameters are segment specific. The deterministic part of Eq can be divided in two: the specific attributes of the choice made; and individual specific characteristics . It follows that the choice probability for respondent n, given that they belong to segment s, will select an alternative i is: Pr nijs ¼ e b 0 s X ni X A k¼1 e b 0 s X nk 0 B B B B @ 1 C C C C Að3Þ Next we use a MNL to place a respondent into a specific segment as follows: Pr ns ¼ e a 0 s Z n X S s¼1 e a 0 s Z n 0 B B B B @ 1 C C C C Að4Þ where Z n is a vector of respondent-specific variables and α s a vector of segment specific Eq is estimated using maximum likelihood estimation requiring that the number of segments S be set in advance. Model Selection. We used model log likelihood ; the minimum Akaike Information Criterion and McFaddens Pseudo R 2 to select the optimal number of segments in the LCM. Results indicated significant model improvement as the number of segments increased. Specifically, the AIC decreased until we had four segments and the rate of change for all criteria was such that that a four segment LCM fitted the data best . --- Interviews Where research can reveal illegal activities, informal, unstructured methods may be more appropriate in terms of accessing data of high internal validity [28,29]. Accordingly, we conducted interviews , largely in an informal or unstructured format [30], in addition to and alongside the hypothetical choice modelling survey. Interviewees were sampled on completion of the survey, or, during opportunistic visits to restaurants selling wild meat. For a breakdown of sampling strategies, interview types and respective sample sizes see Table 3. Interviews were conducted with individuals or groups and some respondents were interviewed more than once. All interviews were conducted with the aid of a pre-prepared guide, however minimal structuring was provided by the researcher. Questions were open-ended and neutral and intended to elicit original and undirected responses from interviewees. Interviews were completed in Vietnamese or English, according to the interviewee's preference and lasted between 10-30 minutes. Due to the sensitive nature of the subject-matter interviews were not recorded, rather, notes were made during the interview. Analysis followed an iterative approach; themes were not predefined but were identified as they emerged from the data [29,31]. Moreover, it took into account the fact that translation seldom conveys exact equivalence in meaning [28]. As such, quotes do not claim to be a precise --- Results --- Choice modelling results We report the MNL , which indicates aggregate preferences, and a four segment LCM . The four segments of the LCM each captured different aspects of respondent behaviour. In all segments price was negative and statistically significant indicating a preference for the lower priced menu options. Source was statistically significant in segments two and three with the negative coefficient in both segments indicating a preference for wild as opposed to farmsourced wild meat. Summing the probability of class membership, segments two and three accounted for approximately 55% of respondents, thus the source of the meat mattered to the majority of respondents. In terms of the meat-types, pig, deer and king cobra were generally preferred , except segment four where pangolin was the most-preferred. Where statistically significant, the coefficients show increases or decreases in utility on the average respondent for changes in each attribute level away from the baselines described in Table 2. For the dummy coded wild meat types, pangolin is taken as the excluded level. In terms of segment membership the best model specification included dummy variables for education and occupation. We found that segment 1 members are associated with higher education levels but lower status occupation-types. . Membership of segment 3 is associated with higher status occupation-types. Members of this latter group were likely to prefer wild-sourced wild meat. --- Attribute Finally, we were able to estimate marginal effects for the preferred LCM specification. The marginal effect tells us by how much the probability of a choice changes given a unit change in price. Thus it is akin to 'price elasticity of demand' . Considering the price attribute in detail we found the marginal effect to be -0.15, which indicates that choice was very price inelastic. --- Qualitative results Price. Interview data confirmed that price was a key aspect of decision making. Interviewees considered lower-priced wild meat a suitable or novel selection in more casual, recreational consumption or drinking contexts: "I see it on the menu, strange meats, at an acceptable price, and I want to try them" "For what reasons do you eat wild meat? To try because my friends tell me it's good. Me and my friends eat it on the weekend when we go out drinking. When I go out drinking I choose between seafood and wild meat, it's a habit now" Contrastingly, more expensive wild meat-types were associated with wealthy and highranking individuals who consume it to communicate their own status and wealth: "Rich people to show their high class because is expensive" . Equally, interviewees described how expensive wild meat-types are purchased to convey the status of, and therefore, respect of others. For this reason they are widely used in business to promote good business-relations and aid the contract-signing process. This, reported interviewees, "is the business culture" . "Pangolin has the highest price that's why you eat it with business contacts it's a question of respect, well not really respect but about showing the status of the invited persons and the invitees" Expensive wild meats were also considered an effective "diplomatic" tool, used to secure favours from those in influential positions. Interviewees explained that in these consumption contexts, not only is high cost a symbolically important product characteristic, but, it is also inconsequential as they have access to corporate or institutional funds. One consumer explained how: "It doesn't matter about the price I can order as high as I like the company pays" Others indicated that even if they had to pay out of their own funds, still the cost would be insignificant in light of what they stand to gain: "They do not care about the money they spend on the meal because what they pay they will gain back" Wild meat-type. Interviewees reported that the rarer and more expensive wild meats are eaten by those situated towards the top-end of society. "The price of pangolin is 10 times higher than civet so there must be a difference in the income of the people who eat these different meats. The people who eat pangolin can throw money at you and make you die. The people who eat civet have an income a little higher than medium the people who eat king cobra have an income between those who eat civet and pangolin" What about bear or pangolin? "Only my boss and company managers would eat that" Interviewees also associated different wild meat-types with different consumption contexts, and linked the consumption of rarer and more expensive wild meat-types with lucrative business or institutional contexts: "Which wild meats would you eat with friends? Ones that are not very expensive and rare. For close friends I would choose the same as I would choose with business contacts, strange ones which are quite expensive and rare" "For friends soft-shelled turtle and cobra snake we usually hang out and the most important thing is fun. For business pangolin and civet they are the highest quality" Interviewees indicated that underlying this heterogeneity is a value attached by certain consumers to those wild meat-types which are expensive and rare. These product characteristics render wild meat suitable for displays of personal wealth and status, and also, for showing the wealth and status of others. "I have to choose the most expensive to show my respect to them" "It is a Vietnamese belief that the more rare the food is the more respect you show to your business contacts" Source of the meat. In agreement with the results of the choice modelling interviewees expressed a widespread preference for wild as opposed to farm-sourced wild meat and considered the farmed substitute inferior. Interviewees distinguished between active wild animals which eat a natural diet and inactive farmed animals which eat an artificial and chemically enhanced diet. This distinction means that wild-sourced meat is perceived as superior in terms of its: quality, nutritional value, health benefits and taste. Additionally, the perceived aphrodisiac abilities of wild-sourced wild meat were alluded to by many interviewees. Interviewees also mentioned a perceived link between the source of wild meat and its exclusivity, rarity and cost. Wild animals are perceived as more rare and expensive than farmed substitutes. They are therefore symbolically more appropriate for demonstrating wealth and status. "Everyone knows that wild civet has a high value. . .so when they want to show their value they eat it. But farmed civet, everyone can have it, so its value is not high. . . so it does not show their high value" For these reasons, for certain high-status individuals and in some business consumption contexts the farmed substitute was considered unacceptable. "If you take them [business contacts] to a wild meat restaurant, if it is not wild wild meat, it is like you don't respect them, like somehow you have tricked them" On the other hand, and in spite of a general preference for the wild product, some interviewees explained that they do eat farmed equivalents: they are more affordable, suitable for casual consumption or drinking contexts, and buying the farmed product means that the consumer avoids being cheated by restaurants which, according to informants, do sell farmed substitutes under the guise of being wild. "The best solution is to order farmed because the price is so different so then you will be sure that you are not being tricked" --- Discussion Over the last few decades the Vietnamese government has committed to tackling the illegal and potentially unsustainable consumption of species for their meat by enacting legislation that regulates the harvest, trade, and consumption of wildlife, and by promoting the captive breeding of consumed species. Recent years have also seen an increased focus on various 'demand reduction' strategies [19][20][21][22]. While all have likely had some impact, the illegal consumption of wild meat remains a problem. Here we explore the implications of our results for conservation actions noting from the outset that this was an exploratory case study, with relatively small sample sizes, and which employed a targeted sampling strategy. Results are not, therefore, generalizable to the population as a whole. --- Regulation Enacting and enforcing regulation is a vital component of strategies to address wildlife crime and recent seizure data attest to its impact. Nonetheless, our results suggest that the role of regulation in this context is likely to be impeded by the following factors. First, and consistent with the findings of related research [9,17], our results indicate that high-status individuals and business people purposefully select rarer, wild sourced and more expensive wild meats to secure business and social advantage. In order to fully understand the extent of the social benefits conferred by wild meat, it is enlightening to consider the broader socio-cultural context of consumption. In collectivist Asian cultures, a society is typically very hierarchical and its individuals preoccupied with their own position within it. Accordingly, displays of social status and 'face' are common, and are reinforced by a cultural inclination to conform to expected behaviours [34,35]. Like other luxury goods, wild meat appears to be appropriate for conveying status and face in a hierarchically conscious society, in this case Vietnam. It may also be the case that the perceived differences in the value of distinctive wild meat-types provide the consumer with the opportunity to assert inter-group differences in status and face [35]. The social and financial benefits conferred by consuming wild meat and the apparent relationship between wild meat and the politico-business nexus will make enforcement difficult. This will be particularly so where these individuals or institutions are willing and able to pay rising prices and where they have the funds or political leverage to avoid prosecution. Second, wild meat is considered more attractive for communicating status and hence more valuable in a business setting if the species consumed are rare and expensive. This is significant because where a species is valued for its rarity and high price it can become trapped in an extinction vortex as its value will continue to rise as its stock in the wild deplete [27]. It is also important because it suggests that a reduction in supply due to enforcement activities may, ceteris paribus, increase the perceived rarity of certain wild meat-types among some consumers, potentially rendering them more appealing [14]. --- Farming According to economic theory, farmed substitutes for harvested wild species will lower prices and thereby incentives to poach wild populations [16]. However, our results suggest that farmed animals may only act as effective substitutes in certain circumstances. While the results of our choice modelling reflect a widespread preference among the sample for wild-sourced wild meat, our qualitative data suggest that farmed substitutes are satisfactory in more casual eating or drinking contexts, and among consumers attracted by their lower prices. However, two important points must be noted here. First, even where farmed animals are satisfactory substitutes, there remains a risk to wild populations as costs associated with farming create incentives to launder wild animals. This has been widely reported [36,15,18,37]. Second, it is entirely possible that farmed wild meat is creating a new market, attracting new consumers, rather than displacing existing demand for wild animals. This issue was not explored here, but warrants further investigation. For a 'super-elite' segment of Vietnamese society, whose members consume wild sourced animals to convey status and wealth, farmed sourced wild meat is not an appropriate substitute as it lacks the product characteristics needed to symbolically convey status and wealth-expense and rarity. Indeed, it is possible that the availability of substitutes in the market is causing these consumers to place increased emphasis on finding the rarer, wild specimens to assert inter-group differences in status and face. Where combined with a willingness and ability to pay rising prices, this could incentivize the exploitation of the last rare, wild individuals of farmed species [27], or alternatively, shift demand to those species whose biology precludes their being farmed. Of future significance here is Vietnam's booming economy-between 1990 and 2010 it grew at an average annual rate of 7.3%, and the per capita income almost quintupled [38]. It may therefore be anticipated that demand for highly-valued, rare, wild-sourced wild meat may also increase as the emerging 'super-elite' seek consumption as a means of demonstrating their economic achievement and high societal position. It is additionally worth noting that where demand is driven by the perceived health and nutritional benefits of eating wild-sourced wildmeat, which reflect aspects of Chinese medicinal philosophy, farmed substitutes are also unlikely to be considered satisfactory substitutes. --- Consumer-centric campaigns Recent literature advocates the use of consumer-targeted campaigns to shift behaviour and reduce demand for illegal wildlife products [9,14,20]. However, for this to be a realistic conservation option a greater understanding of the consumer and the drivers of consumption is needed in order to be able to create effective messaging and target the right audience. Results from our exploratory study indicate that effecting behaviour change in this context will not be easy. A successful campaign will have to approach the issue in a culturally-grounded way and overcome the following barriers to behaviour change: the role of wild meat consumption in communicating status, its perceived health benefits, and the apparent relationship between wildmeat and the politico-business nexus. It will also need to be highly targeted and account for heterogeneous demand . None of these barriers to change are, however, an excuse for inaction and many of them do in fact present opportunities for leveraging behaviour change. For instance, given consumers' preoccupations with meat-related health concerns, any campaign may do well to emphasise the potentially detrimental health implications of consuming unregulated, wild and illegally sourced meat. Campaigns that seek to undermine wild meat's role as a status symbol might also be a good starting point. Looking ahead, we suggest that further research is needed to understand not only consumer decisions about wild meat but also how to integrate behaviour change into the broader sociocultural and institutional landscape. Equally important will be understanding the media and information sources that are typically used and that are trusted and esteemed by target audience members , as well as the type and form of message that is likely to produce changes in behaviour [14]. --- All relevant data is available via Dryad . --- Supporting Information

Vietnam is a significant consumer of wildlife, particularly wild meat, in urban restaurant settings. To meet this demand, poaching of wildlife is widespread, threatening regional and international biodiversity. Previous interventions to tackle illegal and potentially unsustainable consumption of wild meat in Vietnam have generally focused on limiting supply. While critical, they have been impeded by a lack of resources, the presence of increasingly organised criminal networks and corruption. Attention is, therefore, turning to the consumer, but a paucity of research investigating consumer demand for wild meat will impede the creation of effective consumer-centred interventions. Here we used a mixed-methods research approach comprising a hypothetical choice modelling survey and qualitative interviews to explore the drivers of wild meat consumption and consumer preferences among residents of Ho Chi Minh City, Vietnam. Our findings indicate that demand for wild meat is heterogeneous and highly context specific. Wild-sourced, rare, and expensive wild meat-types are eaten by those situated towards the top of the societal hierarchy to convey wealth and status and are commonly consumed in lucrative business contexts. Cheaper, legal and farmed substitutes for wild-sourced meats are also consumed, but typically in more casual consumption or social drinking settings. We explore the implications of our results for current conservation interventions in Vietnam that attempt to tackle illegal and potentially unsustainable trade in and consumption of wild meat and detail how our research informs future consumer-centric conservation actions.

Introduction And Background The COVID-19 pandemic exposed the global healthcare system to an unprecedented challenge. Indeed, few people thought of a "global" healthcare system until the pandemic. In Latin America, the first case of COVID-19 was reported in Brazil in February 2020, much later than the first cases in China, the United States, and Europe [1]. Necessary equipment ranging from personal protective equipment to ventilators was in short supply. Quarantines, lockdowns, masking, and social distancing meant that healthcare-as-usual was interrupted. On the one hand, new telehealth applications were developed to work around these problems, but, on the other hand, many routines and non-emergency care were halted. This has had a particularly adverse impact on patients dealing with chronic pain. As more is being learned about COVID-19 and its management, new insights and techniques into pain care have been advanced. --- Review COVID-19 epidemiology in Latin America The original timeline of the pandemic described patient 0 as a senior man in Wuhan, China, who encountered the novel SARS-CoV-2 virus from a game sold at a wet market [2], but new evidence reports that the first patient may have been infected in China in October 2019 [3]. The origin of the virus has become controversial [4]. The first cluster of patients reported was in China in January 2020, and by January 12, 2020, the genomic sequence of the virus was published. While polymorphisms accumulate in all RNA viruses, specific proofreading functions in SARS-CoV-2 have slowed its replication rate, making mutations take longer than other viruses, such as influenza or norovirus [5]. COVID-19 arrived in Latin America on February 26, 2020, when a 61-year-old Brazilian man returned home from a visit to Italy, which at the time was experiencing a COVID-19 outbreak [6]. As of October 5, 2021, 4.8 million people have died of COVID-19 worldwide, and there have been 236 million confirmed cases in 221 countries and territories [7]. Worldwide, 34.5% of the population is vaccinated, with an estimated 6.36 billion vaccine doses administered to date [8]. The pandemic exposed broad global disparities in healthcare. For instance, only 2.3% of the population of underdeveloped countries have received at least one dose of the vaccine [8]. Seventy-one percent of Brazilians have received one or more doses of the vaccine, much higher than the global average of 46% [8]. Peru and Mexico are about at the mean with 49% vaccination, while Uruguay has 79% of its population vaccinated. South America has an overall vaccination rate of 63% population vaccinated with at least one dose [8]. Latin America has been particularly hard hit by the pandemic, with an estimated 90 million people infected. Accurate statistics can be challenging to obtain and verify because the pandemic is a real-time event, moves quickly, and testing capacity as well as the ability of various countries to gather statistical data varies. The so-called "third wave" of COVID-19 has had a devastating impact on parts of Latin America and Mexico [9]. Latin American nations responded in different ways. For example, lockdowns in some Latin American nations were lax, late, or incomplete. However, Peru, a particularly hard-hit nation, had one of the first lockdowns in Latin America, and it was strictly enforced [10]. Some Latin American nations did not prioritize senior citizens for the vaccine, exposing a vulnerable subpopulation to an elevated risk for infection. Efforts to make people aware of the benefits of masking were sometimes haphazard. Some nations in Latin America, such as Mexico, shared a similar problem as the United States, in that public health is decentralized to the point that centralized efforts to curb infections could not always be implemented [9]. The highest mortality rate for COVID-19 in Latin America occurred in Peru with 606.41 per 100,000 inhabitants, followed by Brazil , Colombia , and Argentina [11]. However, Latin America is a large and heterogeneous region, and the pandemic has resulted in asymmetric crises within Latin America in terms of the healthcare system, the economic system, and social structures. Structural weaknesses were exposed. In general, Latin America does not have a robust hospital infrastructure to manage emergency care on the scale of this pandemic. There were not always sufficient healthcare professionals to manage local caseloads, and many hospitals lacked sufficient critical-care beds to meet the needs of the pandemic. In addition, Latin America faced specific challenges: there was a general distrust of the healthcare system, a disbelief in the seriousness of the pandemic, and a strong interest in traditional medicine and alternative treatments. The pandemic highlighted an urgent need for the education of healthcare workers and reliable information for patients and the general public. Vaccination efforts have been strongly promoted in Latin America with good results, but the world is now confronting variants, new outbreaks, and postviral complications, all of which have added new burdens [12,13]. Anxiety, illness, financial insecurity, and the pressures of prolonged lockdowns have created stressful situations for the general population. The COVID-19 pandemic also showed that many hospitals around the world, including those in Latin America, need more training to manage advanced diseases, better equipment, more stores of personal protective equipment , specific protocols for end of life, and better training of healthcare professionals for pain management. Urgent care was also insufficient in many places because people suffering dyspnea, high fevers, respiratory distress, and tachycardia were not always rapidly and appropriately treated. There is much that can be learned from the pandemic. Greater healthcare literacy is needed. Care in the pandemic should be "de-medicalized" so that different treatment options are available depending on the patient and the trajectory of the illness. We must encourage the population to be autonomous and empowered in managing their risk of infection and its consequences. This means, in part, providing the general population with realistic strategies for treatment, supportive care, accurate but understandable public health information, and encouragement. Virtual care has emerged as an unintended benefit of the COVID-19 pandemic, and patients seem more accepting of telehealth than they were previously. Support groups and networks of patients who can encourage each other should be facilitated to proactively address mental health issues. There is much that COVID-19 has taught healthcare professionals and the public health system. The pandemic caused global healthcare to face up to systemic vulnerabilities that must be recognized and defined to be corrected. COVID-19 has created a climate of uncertainty that challenged clinicians, but in a crisis, the healthcare system must be able to manage uncertainty. Clinicians faced great adversity during the pandemic as they struggled with difficult caseloads, insufficient supplies and equipment, lack of PPE, uncertain pandemic progression, lack of clear-cut public health guidance, and their own risk of infection. Ideally, these healthcare workers need to build robust personal networks of family and friends and other support structures to see them through periods of crisis. More education is needed in infectious disease, triage, and other aspects of emergency medicine. Finally, healthcare workers around the world must have and maintain high levels of professional empathy, even when they face what can seem like insurmountable challenges. COVID-19 will become endemic and lose its pandemic status in the future, but epidemics and pandemics will recur, and we need to learn these important lessons now to weather the next outbreaks more safely and effectively. --- Painful syndromes in COVID-19 The COVID-19 pandemic has had broad ramifications, adversely impacting people who contracted the infection, healthcare professionals who cared for them, healthy people seeking to avoid infection, and the large population of individuals who struggle daily with chronic pain [2]. Each has faced unique and important challenges in this pandemic. COVID-19 has presented extreme situations for the general population: abrupt sickness and death of loved ones, financial loss, lockdowns, emotional devastation, insecurity, fear of infection, anxiety about the future, separation from family and friends, drastic changes in work and travel, a sense of "disconnectedness," uncertainty about public health data and statistics, and other psychosocial stresses which will likely extend into the future [14]. The pandemic exposed the general public and the healthcare profession to an avalanche of sometimes contradictory information, which in some cases made people mistrustful [15]. Chronic pain patients have been disproportionately impacted by the pandemic [16]. Chronic pain has been described as a multidimensional biopsychosocial phenomenon involving the complex interplay of biological, psychological, and social factors [17]. Thus, the psychological and social stresses of a prolonged pandemic could increase the prevalence of chronic pain and exacerbate painful symptoms. Patients already dealing with chronic pain now suddenly were confronted with personal and financial loss, isolation, new medical paradigms for obtaining care, in some cases denial or delay of care, and the day-to-day fear of contracting the infection. Long-haul COVID-19 or long COVID-19 is a postviral syndrome that was first reported online by people discussing their symptoms after COVID-19 resolution and has since been described in the literature [18]. Symptoms of long COVID-19 are many, but prominent among them are chronic fatigue, diffuse myalgia, and sleep disorders [19]. Postviral syndromes have occurred with other viruses, such as the severe acute respiratory syndrome epidemics in 2002 [20]. Gastrointestinal infections are strongly linked to postviral syndromes in general, and about 10% of patients who develop certain enteric infections will develop irritable bowel syndrome [21]. The exact trajectory of long COVID-19 remains unknown, but it appears that risk factors for prolonged symptoms include the severity of the acute COVID-19 infection, its duration, and the premorbid psychological status of the patient [22]. Although older age is a risk factor for acute COVID-19 infection, it does not seem to be a risk factor for long COVID-19 [22]. Moreover, the symptoms that occurred in acute COVID-19 are not necessarily the same as the symptoms that emerge in long COVID-19 [19]. In the aftermath of the apex of the pandemic, there emerged three distinct new subpopulations of chronic pain patients. The first are those whose chronic pain first emerged after COVID-19 infection and might be considered a form of postviral chronic pain; the pain relates either directly to long COVID-19 or to organ damage that occurred during the acute infection [13]. The second subpopulation includes those who had preexisting chronic pain that worsened after the COVID-19 infection. The third emerging chronic pain subpopulation is less well studied and consists of individuals who did not have chronic pain before acute infection but who developed chronic pain during the pandemic. These individuals had psychological and psychosocial risk factors that predisposed them to chronic pain, such as anxiety, catastrophizing, fearfulness, loneliness, depression, and the stresses of the pandemic triggered painful symptoms. In addition to these three new subpopulations, there is a broad population of people with pre-existing chronic pain whose pain continued but did not worsen or change throughout the pandemic. In this connection, it is important to recognize that much intensive care unit patients develop "postintensive-care syndrome" that can be associated with chronic pain [23]. ICU hospitalization for any cause may result in cognitive dysfunction, weakness, and intrusive thought patterns similar to post-traumatic stress disorder that may persist long after ICU discharge [24]. Of ICU patients on mechanical ventilation, 56% report some form of impairment after critical illness enduring for at least a year [25]. Among survivors of the earlier severe acute respiratory syndrome epidemic in Hong Kong, 25% to 44% suffered posttraumatic stress disorder, and 15% had depression after the infection [26]. The symptoms of long COVID-19 include limb and joint pain, headache, neuralgia, myalgia, abdominal pain, chest pain, angina, and others [27]. Among COVID-19 survivors, many will experience chronic pain with a neuropathic component after the infection [28]. Different body systems can be affected, such as the respiratory, cardiovascular, hepatic, and other systems [27]. The distinction between pain in long COVID-19 and new-onset chronic pain is not clear because it may be possible for certain painful conditions in long COVID-19 to transition to chronic pain. Patients who survived COVID-19 have a significantly higher incidence of de novo chronic pain than patients who were not infected and a higher prevalence of de novo headaches [29]. While acute and long COVID-19 symptoms are not always correlated, headache at the onset of acute COVID-19 infection has been associated with long-term headache and fatigue that persisted up to seven months after hospital discharge [30]. When headache occurs in acute COVID-19 infection, the patient should be monitored for persistent headache following the infection. Among COVID-19 patients discharged from the hospital, 27.3% reported persistent joint pain [31]. In a study of 41 hospitalized COVID-19 patients in China, 44% had myalgia, of whom 33% had elevated levels of creatine-kinase [32]. In a case series of 99 patients with diagnosed COVID-19, 11% of patients reported myalgia, of whom 13% had elevated creatine-kinase levels [33]. The SARS-CoV-2 virus enters the body by way of the interaction of its spike domain with the host's angiotensin-converting enzyme -2 receptor. ACE-2 receptors are dense in the musculoskeletal system, and it has been speculated that the virus may directly impact skeletal muscle [34]. The body's systemic inflammatory response during infection may also play a role in myalgia [34]. Perhaps most important in considering chronic pain after COVID-19 is the role of neuropathic pain, which may manifest as Guillain-Barre syndrome, myopathy, polyneuropathy, peripheral nerve injury, and other forms of peripheral or central neuropathy [28]. Neuropathic pain has been reported after Zika, chikungunya, herpes zoster, HIV, and other viruses [28]. However, neuropathic pain has been associated with ICU hospitalization for any cause. Neuropathic pain following ICU hospitalization has been attributed to prolonged time spent in a prone position, lack of mobility, neuromuscular blocks, muscle atrophy, complications due to procedures such as chest tubes or tracheotomy, critical illness myopathy, and polyneuropathy [35]. ( --- TABLE 1: Risk Factors for Chronic Pain Among Hospitalized COVID-19 Patients Chronic neuropathic pain may occur in up to 2.3% of hospitalized COVID-19 patients, although the mechanisms remain to be elucidated [36]. Stroke, myelitis, or Guillain-Barré syndrome, are prevalent among COVID-19 patients and associated with an elevated risk for neuropathic pain [36]. Patients who entered the era of COVID-19 with chronic painful conditions may have suffered worsening symptoms. The lockdowns and overtaxed healthcare systems meant that chronic pain patients could not always get the medical treatments on which they had come to depend. Unemployment, financial distress, overburdened hospitals and clinics, and lapses in insurance coverage also limited care for chronic pain. The stress, anxiety, and loneliness imposed by the pandemic may also have worsened painful symptoms. Finally, during the lockdown, many people, not just chronic pain patients, were unable to get their usual exercise or pursue healthful outdoor activities as normal. These things set the stage for an exacerbation of pre-existing chronic pain. These same factors may have triggered de novo chronic painful conditions in others, likely attributable to fear, anxiety, depression, poor sleep, and forced inactivity. The extreme stress associated with the pandemic affected some people to the extreme, possibly even exacerbating substance use disorder [37,38] or leading to suicidality [39,40]. COVID-19 will have far-reaching implications for pain medicine, and there are urgent and unmet medical needs to help treat the pain-related symptoms of acute and long COVID-19. Dedicated multidisciplinary COVID-19 clinics and registries of infected patients may be helpful to better understand how COVID-19 has affected chronic pain patients. The role of telehealth is a positive consequence of the pandemic and may serve pain patients in the future as a new and convenient way to deliver certain types of healthcare. Above all, it must be recognized that de novo chronic pain or worsening pre-existing chronic pain is prevalent in the pandemic, and prompt, targeted treatment is needed [41]. --- Pain management in COVID-19 People with pre-existing chronic pain, regardless of their COVID-19 status, may have suffered worsening symptoms as a result of the psychological stresses and social limitations associated with prolonged isolation [41]. Many chronic pain patients dependent on the healthcare system and their caregivers may have felt abandoned during lockdowns if their routine care had to be interrupted [42]. Chronic pain patients who contracted COVID-19 ideally require a nuanced approach to care that both controls pain and manages viral symptoms; such care may not have been available or accessible. Tele-triage describes a telehealth practice that allows a remote clinician to assess patient symptoms, measure the body temperature, assess pain, and categorize the patient. The triage levels in this system should be low for chronic pain patients who have controlled pain and do not have COVID-19; elevated when it is suspected such patients may have COVID-19; and highest when it is suspected that emergency care may be appropriate [43]. Murphy and Latif recommend that triage levels be described as elective, urgent-elective, and urgent [44]. Interventional pain patients may not have received their routine treatments during the pandemic, and a triage system for interventional procedures has been proposed. For example, urgent triage levels are appropriate for chronic pain patients experiencing a worsening of complex regional pain syndrome or degenerative or neurological pain resulting in impaired mobility, while emergency triage levels included pain or conditions that impaired the patient's ability to conduct the activities of daily living. The lowest triage level, elective, is reserved for conditions that are stable and can be safely and effectively managed remotely, using alternative treatments rather than an interventional procedure [45]. Many interventional pain patients require ultrasound-guided treatments, and during the pandemic, this posed a problem even for those who could get to a hospital or clinic for pain treatment. The sonographer or sonologist must be near the patient during imaging, which requires strict protective measures. The SARS-CoV-2 virus may survive up to nine days on ultrasound equipment and can exist in inert form for up to 96 hours. Critical devices such as ultrasound equipment must be cleaned carefully between patients using a high-grade disinfectant [46]. Ultrasound personnel should use adequate PPE during scanning. Some clinics might not have offered ultrasound during various phases of the pandemic or that patients were too fearful to undergo treatment. Since perioperative COVID-19 infection increases postoperative morbidity and mortality, patients diagnosed with COVID-19 are asked to delay surgery, if possible, for at least seven weeks until the infection has completely resolved. Those who have symptoms after seven weeks may require further delay [47]. This meant that important procedures to relieve pain had to be postponed for many patients. The use of analgesics during the pandemic has generated considerable controversy and confusion. Nonsteroidal anti-inflammatory drugs may be appropriate for pain control in certain patients, but there remain some unknowns in their use with COVID-19 patients [48]. At first, there was controversy about the use of ibuprofen, often chosen for its antipyretic properties, but these concerns were shown to be unfounded [49]. It was further speculated that NSAIDs may be safe but could potentially mask symptoms of the infection that would be better to recognize early. Pain patients and others may require information about the use of over-the-counter pain relievers such as paracetamol or NSAIDs, as there are risks to their use, particularly over a prolonged period. Opioids are effective analgesics and may be appropriate for treating chronic pain in certain patients, but since opioids can suppress the immune system, their use must be scrutinized and carefully considered in a pandemic [50]. Opioid therapy may also mask symptoms of acute infection, which could delay COVID-19 treatment. Transdermal opioid delivery systems should be used with caution, as heat, even from a fever, can increase drug release or otherwise negatively affect treatment [51]. Patients with prolonged exposure to opioid analgesics may develop withdrawal symptoms if opioids are tapered too quickly or discontinued abruptly. In some cases, telemedicine may be used to provide advice to patients who may need to taper their dose [52,53]. Chronic pain patients may take antidepressants, benzodiazepines, and other adjuvant agents that could potentially result in drug-drug interactions with certain COVID-19 treatments. Antidepressants may interact with hydroxychloroquine, chloroquine, and azithromycin, which may elevate the risk for QTsegment prolongation. Lithium is contraindicated for concurrent use with hydroxychloroquine, chloroquine, and azithromycin. Anticonvulsants and benzodiazepines present little or no risk for adverse interaction with COVID-19 treatments [54]. The use of corticosteroid injections during COVID-19 is controversial. There is some evidence that one intra-articular corticosteroid injection increases the risk of contracting influenza; no study has been done specifically on COVID-19, but this finding is suggestive [55]. No studies have been conducted as to whether a corticosteroid injection might affect the clinical course of COVID-19. If a patient might benefit from a corticosteroid injection during the pandemic or if the patient has acute COVID-19, it is important to exercise clinical prudence in appropriate patient selection, explaining the potential risks to the patient, and obtaining informed consent. If a corticosteroid is to be used, it should be employed at the lowest effective dose for the briefest effective course of treatment [55]. It is not clear if corticosteroid injections might affect the efficacy of the messenger ribonucleic acid vaccine for COVID-19. Based on the understanding that both epidural and intra-articular corticosteroid injections suppress the hypothalamic-pituitary-adrenal axis, out of an abundance of caution, it has been suggested that a corticosteroid injection be administered at least two weeks before an mRNA vaccine or no fewer than one week after [56]. The disruptive impact of COVID-19 on the world has not been entirely negative. Telemedicine, which met with some resistance before the pandemic, has expanded and improved to the extent that it is much more helpful and accepted today than it was just a few years ago. For example, the literature describes a virtual spinal examination, which allows the patient to actively participate in a remote spinal examination that can guide the treatment of back problems [57]. COVID-19 has likely accelerated such innovative telehealth approaches. The pandemic has created several serious challenges in the treatment of pain. Healthcare professionals must exercise caution and mitigate risk when patients and clinicians are in close contact. Pain patients have faced unprecedented challenges in getting care due to clinic lockdowns, fear of infection, and social distancing. The medical profession must recognize that the pandemic has affected the general population not just physically but also in terms of mental health, and pain patients may not be resilient but instead vulnerable to depression, anxiety, and stress. Healthcare systems can be stretched to the breaking point, forcing clinicians to sometimes make difficult decisions in terms of where and how to direct their efforts [58]. Pain specialists must consider all of these factors and seek to strike a balance in offering responsible care for their patients, protecting themselves, and promoting trustworthy public health policies. --- Palliative care Epidemics and even pandemics have always been with us. COVID-19 is a well-documented and global pandemic, but it does not surpass the plague of Justinian in 541-549 AD in virulence and devastation. This plague, which affected the Roman Empire all of the ways to its eastern capital in Constantinople, was responsible for an estimated 50 million deaths. The plague of Justinian was just the first wave of what would later be known as the Black Plague, which waxed and waned over the centuries. The worst outbreak occurred between 1346-1353 AD in Europe when it is estimated that up to 200 million souls perished. At that time, neither the cause nor the treatment for the plague was known, heightening the fear and terror of the time. Even today, the humanitarian crisis of a pandemic strains our social systems, challenges medicine, collapses the economy, and overwhelms our public health efforts to care for sick and dying individuals. These plagues divert considerable economic and human resources away from the common good and toward local and national solutions aimed only at mitigating suffering, providing whatever incomplete relief is possible. Even when costly, these solutions offer only partial remedies. In the modern world, pandemics should be viewed as humanitarian crises that demand a coherent, coordinated, and global response. During these crises, palliation emerges as an almost alien concept that contradicts a fundamental notion of modern medicine, namely that patients must be treated as aggressively as possible in the struggle against death. To some extent, the healthcare systems view death as defeat and illness as inherently manageable, even if it cannot always be cured. COVID-19 has shown our healthcare systems that we must think seriously about the role of palliative care. existing laws to modify the nation's current telemedicine system into the Teleatiendo, which allows for patients to be monitored, treated, and attended by phone or instant messaging service throughout their illness [67]. Further legal modifications were quickly implemented to allow for opioids to be prescribed by phone. In these ways, Peru was a leader in adapting healthcare in innovative ways to meet the challenge of the pandemic [59]. However, this raised an important and ethically delicate issue: certain patients were not going to survive COVID-19. These patients were offered palliative care, allowing the healthcare system to prioritize scarce resources toward those patients with a better likelihood of survival but without neglecting the needs of patients nearing the end of life. Palliative care alleviates pain and suffering and does not facilitate death. Instead, palliative care recognizes the inevitability of death for certain patients and seeks to provide them comfort, dignity, spiritual support, and symptomatic care in their final days [60]. The concepts of palliative care are relatively new, and even healthcare professionals may find them offputting. Training is urgently needed to raise awareness about the value of palliative care and its main principles. In palliation, an individual patient assessment must be made regarding the disease, its severity, and its prognosis, taking into account individual patient factors, such as comorbidities, frailty, and patient age. If the patient's situation is severe, not likely to change, and would not benefit from treatment , the goal shifts from treatment to alleviation of suffering. The patient is not abandoned, but he or she is not subjected to extreme and possibly painful efforts to delay death through aggressive treatments. A core goal of palliation is providing the patient comfort and peace as life comes to an end, which has particular relevance in the COVID-19 pandemic [61]. The isolation imposed by lockdowns, the strict protocols enforced by hospitals, and the fearfulness in society may serve to heighten the alienation, suffering, and pain in the dying patient. As much as possible, palliative care patients must be allowed to say goodbye to family and friends and to take what small comforts they can find in familiar settings. Palliative care patients should be given medications to ease their pain and relieve their emotional turmoil. The goal of palliation is to allow as much peace and dignity for the patient and their loved ones as possible as death approaches [62]. Palliative care training is essential for healthcare professionals who must work in epidemics, pandemics, or disasters. The fundamental principles include triage, effective but compassionate communications, pain assessment, and alleviation of suffering [63]. This must overcome modern clinical training that emphasizes the "hero mentality" where aggressive and even risky treatments may be used to prolong life at all costs. It is not a medical failure to recognize that some patients will not survive, and these patients require a different kind of care. Instead, palliative care makes the distinction between those who can fight and survive versus those who should be comforted as they pass away. Pandemics, epidemics, outbreaks of virulent infectious diseases, and other disasters will undoubtedly recur. They deserve more study to increase global and local preparedness and find optimal strategies to take decisive, effective actions as soon as possible. At the same time, the concept of palliation must not be jettisoned. Governmental, non-governmental, and other official agencies must be transparent and forthright when sharing information. The climate of fear around COVID-19 was in part due to confusion among people who were unsure about the risks of the disease or how to protect themselves. Better and more direct communication is essential. A centralized crisis management system may be helpful to create a "defense system" of epidemiologists, public health experts, healthcare professionals, and government leaders who can respond quickly to disease outbreaks [64]. In new outbreaks, priority should be given to developing reliable diagnostic tools, effective treatments, and vaccines, if appropriate. PPE and other specialized equipment must be stockpiled for rapid deployment where needed. Emergency stores should be set up so that even under-developed and poor nations have ready access to PPE and other equipment. This process can be formalized so that local governments may request from a central authority the basic necessities, medicines, PPE, and other necessary tools to fight disease when an outbreak is declared. This sort of system could distribute supplies where needed fairly and transparently. Peru has implemented and expanded several telemedicine centers that treat not only COVID-19 but also oncology patients and others at high risk and/or with complex conditions during the pandemic [65]. In Peru, 2,267 new cancer patients were treated with telemedicine in 2016; this number increased to 2,472 in 2018 [65]. In January and February, as well as July and August of 2020, Peru saw the highest number of telemedicine consultations, which mirrored pandemic activity [65]. While there is considerable interpatient variation, the approximate average time for a telemedicine consultation is about 45 or 60 minutes for an existing or new patient, respectively. Rapid and effective response to a healthcare emergency is possible if a trained team and an established protocol are already in place. In Peru, we could provide care for virtually all of the population infected with COVID-19 during the pandemic. Remote regions of Peru, like remote regions in all parts of the world, were hard hit [66]. Systems must proactively find solutions to send resources quickly to remote regions. These response efforts must also recognize that palliative care is an essential component of emergency healthcare. In Peru, it was possible to provide palliation to many appropriate patients using telemedicine models and healthcare professionals trained in palliative care. What is needed in Peru and around the world is an influx of healthcare professionals better trained for emergency and palliative care. The COVID-19 pandemic is not over. It is imperative for us as medical authorities and healthcare professionals to acknowledge that the pandemic may worsen, may resolve but recur, or those new variants may arise. --- Conclusions In Latin America, as in the rest of the world, COVID-19 exposed numerous vulnerabilities in our healthcare system as hospitals were strained to accommodate burgeoning caseloads. Latin America faced specific challenges in that there was a cultural distrust of the healthcare system, a skepticism about the seriousness of the pandemic, and a strong interest in alternative medicine. However, vaccination uptake has been generally strong in Latin America. COVID-19 is associated with certain painful symptoms, such as myalgia and headache, and post-viral syndrome or "long COVID-19" has been linked with a range of symptoms, including neuropathic pain. Patients with pre-existing chronic pain before the pandemic may have had their pain care routines interrupted, treatments delayed or cancelled, and symptoms exacerbated by lockdowns, social isolation, anxiety about the disease, and financial setbacks. Triage systems have been developed to prioritize pain care in COVID-19, which is complicated by the risk of disease transmission. Ultrasound guidance, for example, used in certain interventional pain treatments, was sometimes off-limits during COVID-19 for fear of disease transmission. While COVID-19 has been devastating on many fronts, it also brought with it advances and greater acceptance of telemedicine and greater awareness of the need for palliative care for patients at the end of life. --- Additional Information --- Disclosures --- 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.

Vaccinations and therapeutics have been developed for COVID-19, but vaccine uptake varies markedly among countries. Public health responses have also varied, in particular, with lockdown efforts and school closing. All over the world, the pandemic exposed healthcare and economic weaknesses. COVID-19 exacerbated mental health issues by exposing the population to prolonged periods of fear, anxiety, financial stress, psychological uncertainties, and sometimes isolation from even family and friends. Chronic pain patients have been disproportionately affected. The pandemic-associated stresses may have exacerbated their already painful symptoms while at the same time interrupting their access to care. The ramifications of the COVID-19 post-viral syndrome ("long are not yet known. COVID-19 viral infection has been associated with neuropathic pain symptoms. Tele-triage and telehealth applications can help manage chronic pain patients in the COVID-19 era, but many interventional procedures, injections, or other treatments have been delayed. The role of palliative care for patients with terminal cases of infection must be re-examined. Palliative care is a relatively new medical specialty and allows terminally ill patients to die in as much comfort and peace as can be afforded to them. More training in palliative care for all clinicians is urgently needed. COVID-19 exposed much that is wrong or weak or inadequate in our healthcare systems, but it also allowed us to embrace new technologies and develop better systems to manage the challenge of a pandemic.

Introduction The COVID-19, which began in December 2019, has swept the world, causing not only huge losses to the lives and economies of all mankind, but also seriously affecting the mental health of ordinary people . The COVID-19 outbreak was first detected in December 2019 in Wuhan, and China was the first country affected by the coronavirus pandemic, thus the development and reply of China's COVID-19 epidemic has been concerned internationally . After March 2020, in China, the epidemic of COVID-19 was well under control and people's lives gradually returned to normal. A follow-up study of the mental health of Chinese residents found that the levels of stress, anxiety and depression had been relatively stable since March 2020 . However, in January 2021, China's COVID-19 partially reoccurred in Shijiazhuang, Hebei Province. China once again launched wartime control measures: the whole city was under closure management; airports, railways and public transport were all suspended; about 10.25 million residents were quarantined at home, and on January 7, 12, 20 all residents carried out three nucleic acid tests. More than 20,000 people were in a centralized isolation during this outbreak. Since China was deemed to have "defeated the COVID-19 epidemic, " the second round of the local concentrated outbreak dragged the Chinese from bystanders to face the pandemic. The pressure and anxiety associated with COVID-19 have become apparent. After the global outbreak of the COVID-19 in 2019, the media is the main source of information about COVID-19 , especially social media . China's major social media sites have a dedicated "COVID-19 Fight" section. WeChat added more than 100 "mini-programs" related to COVID-19, updating information related to the outbreak simultaneously. In addition, many self-media and netizens also published and disseminated relevant information on WeChat, Weibo, and other social media . Thus, the impact of social media on people's mental health has been highly concerned by scholars, governments and the International Health Organization . The research conclusions of the existing literature are consistent, that is, more reliance on social media has led to severe depression, anxiety, and other psychological problems after the outbreak . Since the pandemic began in 2019, social media has been bombarded with false information and reports about COVID-19, sparking groundless fears among many netizens . A large number of gossips and misinformation pose a serious threat to public health . The COVID-19 epidemic has developed into an unprecedented "information epidemic" . In fact, prior to the outbreak in 2019, the support of social media for people's mental health was confirmed by many scholars. Researches by some scholars have suggested that social media can provide users with affluent information, help obtain solutions to problems such as doubts, health crises, etc., and help improve emotional threats and protect mental health . What's more, social networking provides valuable emotional support, through which users can feel being accepted, loved and respected , a sense of belonging and self-affirmation , which in turn improves mental health issues . Therefore, we found that the impact of social media on people's mental health varies among different social contexts and media environments, or even the complete opposite. As a result, in such a special social environment as CODIV-19 outbreak, it is necessary to explore the relationship between social media use and the mental health of residents in different social media environments. In any case, as some scholars have pointed out, the significant impact of social media on people's mental health has been widely recognized. Governments of countries and health departments are struggling to find effective solutions to the "information epidemic." While the Chinese government has taken strong measures to control the COVID-19, it has also stepped up its efforts to govern the social media environment. On April 16, 2020, the National Health Council of China issued guidelines for local governments to promote psychological crisis intervention for residents during public health emergencies. However, after the social media environment has changed, it remains unclear whether the mental health problems of Chinese residents during another local outbreak in China, and whether the impact of social media on mental health will change at this time. Therefore, there is a need for a rapid assessment of the factors affecting the mental health during the second outbreak . In addition, numerous studies have also focused on the heterogeneity of depression that occurred during the pandemic . For example, during the period, the prevalence of depression in women was significantly higher than that in men , and young people under 40 are more susceptible to depression . However, so far the impact of social media on public psychology during the COVID-19 pandemic has received due attention, while the heterogeneity of how social media use can affect mental health remains almost unnoted and how it differently works on the disadvantaged and the advantaged remains unclear. Based on the above-mentioned current Chinese situation, as well as the literature review and analysis, our research raises the following questions: --- Methods --- Design and participants The data of the study were cross-sectional data from the online survey, and the questionnaires were distributed by nondirectional random, with participants aged 18 and older. We surveyed urban residents of Shijiazhuang who were living in isolation at home and those who had restricted mobility outside Shijiazhuang. The survey was conducted between January 10 and January 15, 2021, 1 week after the closure and home isolation in Shijiazhuang during the second round of China's COVID-19. The survey was conducted through the questionnaire star platform , which invited Chinese urban residents to participate online. All participants in the survey volunteered to fill out the questionnaire, which was conducted under the supervision of the Academic Committee of University of Shanghai for Science and Technology. All participants in the survey were asked to answer specific questions about demographics, social media use, depression, anxiety, and other changes in mental health. In the questionnaire design, we set multiple restrictions to ensure the validity of the data. Specifically include: Filtering for duplicates of IP. 1. Duplicate questionnaires from the same device, IP address or Wechat account were deemed invalid. 2. Time limit for answering questions. Questionnaires that take <150 s to fill out were considered invalid. 3. The design of questionnaire questions. The questionnaire was occasionally interspersed with basic cognitive questions, including "Where is the capital of China?", "Which picture is a square?", "3 + 6 = ?". If one of the above questions is answered incorrectly, the questionnaire will be considered invalid. Once the questionnaire is valid, the participant will receive a bonus . We finally screened out 904 valid samples from the 1,204 samples, including 412 samples from Shijiazhuang residents who were isolated at home and 492 samples from residents with restricted mobility in other areas. As of this writing Shijiazhuang is still under lockdown, during which the psychological health of residents is uncertain, so it is particularly important to obtain rapid and timely data. Although the network survey is not a rigorous sample survey, the sample representation is not perfect, but the advantage of the network survey is the rapid access to data in the event of emergency. The distribution of the sample population is shown in Table 1. From the statistical results of valid samples, although it is not completely consistent with the distribution of China's population, it also covers the majority of groups. Therefore, our sample has some representative significance. --- Measurement Mental health According to previous studies, depression and anxiety assessment have become the most important measurement of mental health. Depression is assessed through a Chinese version of depression by The Chinese version of WHO-Five Well-Being Index , including five positive emotional items: feeling happy and comfortable, feeling calm and relaxed, feeling energetic, feeling sober after waking up, getting enough rest, every day's life is abound with interesting things. Participants were asked how often they had these positive emotions since the COVID-19 outbreak. A score of 6 points was used, from all times to no time . Less than 13 scores indicate depression. Anxiety is assessed with the widely used Anxiety Scale GAD-7 , including 7 negative items: feeling nervous, worried and anxious, being unable to stop or control worry, worrying too much about all kinds of things, being difficult to relax, being unable to sit still due to restlessness, becoming prone to trouble or impatience, feeling that something terrible would happen and scared. Participants were asked how often they had these negative emotions since the COVID-19 outbreak. Response options are "not at all, " "less than half the time, " "more than half the time, " and "almost daily, " with assignments of 0, 1, 2, and 3, respectively. A total score of more than 10 represents that the respondents suffers from anxiety. --- Social media usage Social media use is done by asking respondents how often they have accessed information about COVID-19 through social media since the outbreak . In order to more accurately reveal the impact of social media on mental health, we included both social media and traditional media as independent variables in the model to compare their effects on mental health. Traditional media mainly include television, radio, newspapers, official websites, magazines and so on. Social media include WeChat, Weblog, Zhihu, Douyin, Toutiao, news networks and so on. The response is measured by using the five-point Likert scale of the "never" , "occasionally" , "sometimes" , "often" , and "always" . --- Covariates The regression model is adjusted according to a number of personal factors, including income, education , gender, age, marital status , number of people living together, years of residence, self-rated health . First, our model controlled variables that may be related to social classes, such as income and education. The rationale behind is that social classes are found to be associated with mental health . Second, as a large number of literatures have identified the effects of living conditions on mental health , we controlled two variables: the number of people living together and the number of years of residence to increase robustness of the model. Finally, we controlled SRH, because SRH has a direct impact on emotional health, covariate to SRH, but also help to increase the robustness and effectiveness of the model. --- Statistical analysis This study selects descriptive statistical analysis and structural equation model for statistical analysis. In order to test whether the data is suitable for the method of SEM, we have significant t-checking of all observed variables, which are split by 27 and 73 scales, and the results show that all variables have good identification. The factor verification analysis of depression and anxiety measurement model is carried out. The results showed that the component reliability of the measurement model was >0.6, the average variance extraction was >0.5, the factor load of the observation variables was >0.6, and the reliability coefficient was >0.36 , which showed that the measurement model all has good reliability and validity, and were suitable for SEM analysis. The results of depression model and anxiety model fitting showed that the fit index GFI value, RMSER, card-square degree of freedom ratio , AGFI value, IFI value and CFI value all meet the ideal standard, and the model has good fitness. In addition, our study is based on cross-sectional data, so in order to explain the relationship between social media and residents' mental health as much as possible under limited conditions, and to avoid the effects of confounding effects as much as possible, we used Propensity Score Matching to analyze and verify the robustness of the structural equation model results. PSM achieves a random assignment effect by matching the samples of the treatment group and the control group one by one to control the self-selecting mechanism interference caused by observable variables, and by controlling the propensity value, it is possible to "approximately" satisfy the non-obfuscation assumption under the statistical counterfactual framework and thus make causal inferences . The basic process of PSM is to estimate the probability of each sample being grouped into a processing group by using the Logit model according to the observable confusion variables, obtain its propensity score, and then match the samples with the closest propensity values but belonging to the two groups one by one. We used proximity matching to obtain a design effect similar to that of randomized trials. Finally, the average treatment effect on treatment and significance of social media on depression and anxiety in residents were obtained. --- Results --- Descriptive statistics for the sample Mental health Depression Depression symptoms were present in 38.9% of the overall sample. Age 39-, female , master's degrees and above , low income , unmarried , students , poor SRH , number of people living together 4+ , and years of residence 4- had a higher prevalence of depression, all by more than 40%. --- Anxiety 12.6% of the overall sample had anxiety symptoms. Age 39-, female , college and below , low income , unmarried , students , poor SRH , number of people living together 4+ , and years of residence 5+ were more likely to have higher anxiety. It is worth noting that such groups as Age 39-, female, low income, unmarried, students, poor SRH, the number of co-residents 4+ reflect a higher prevalence of depression and anxiety. See Table 1 for details. Through the comparison of model results, we found no significant difference in the association between social media For a clearer picture of the comparison, we have merged "never, " "occasionally, " "sometimes, " "often, " "always, " "always" level 5 items into level 3, "never, " "occasionally" into "less, " "often, " and "always" into "many". use and mental health among Shijiazhuang residents living in isolation and those living in other areas. --- Social media usage --- Traditional media In the overall sample usage frequency, the proportions of choosing "less, " "sometimes, " and "often" were 24.3, 18.1, and 57.5%, respectively. The groups, including middle-age and above, college and below, local personnel, very good SRH, use traditional media significantly more than others. The frequency of using traditional media in these groups is 51.9, 63.5, 60.3, and 65.1%, respectively. --- Social media In the overall sample usage frequency, the percentages of "less, " "sometimes, " and "often" were 10.8, 9.1, and 80.1%, respectively. Groups, including Age 39-, master's degrees and above, high-income, students, and local people groups, use social media significantly more frequently than other groups, which frequency of using social media is in order: 81.6, 89.2, 88.6, and 81.8%. See Table 2 for details. --- The overall model Table 3 shows the model fitting results based on the overall sample. When gender, age, education, income, marital status, number of people living together, years of residence, and SRH were controlled, social media had a significant positive correlation between depression and anxiety at 1%level. The effects of standardization were 0.161 and 0.120 , respectively However, the use of traditional media is not significantly related to anxiety and depression. In addition, we can also see that age and SRH have a significant positive effect on depression at 1% level, with standardized effect coefficients of 0.103 , 0.260 , and no significant correlation between other covariates and depression . Education and SRH had a significant positive effect on anxiety at 1% level, with standardized influence coefficients of 0.096 and 0.235 , respectively. Marriage had a significant negative effect on anxiety at 1% level, with a standardized effect factor of -0.140 , and no significant correlation between the other covariates and anxiety . --- FIGURE The standardized path of depression influenced by social media. We used PSM to analyze the results of structural equation models for robustness analysis and sensitivity Analyses. First, we divided the samples of residents into experimental and control groups according to the difference in social media use frequency, with the top 50% of the samples in social media use frequency as the experimental group and the rest as the control group. --- FIGURE The standardized path of anxiety influenced by social media. The gender, age, education, income, marital status, number of people living together, years of residence, and SRH are then used as disruptors through binary Logit. The model estimates the probability of each sample being grouped into the experimental group and obtains its propensity score. Match matching is then performed, matching the samples with the closest propensity values but belonging to two groups one by one. We used the nearest neighbor matching method for matching, and the results showed that the matching success rate was 100%. The propensity score matching effect is detected by parallel hypothesis testing. The results showed that the absolute values of the normalized deviation after matching were <20%, and the normalized deviation value was significantly reduced, and the matching effect was better; the t-test after matching was not significant , age, Marital status, income, and self-assessment health were several variables that were significant before the t-test , and the t-test after matching had no significance , indicating that the matching effect is good. The matched data were analyzed by the Average treatment effect , and it can be seen that the ATT effect value after matching is still significant , i.e., the PSM analysis showed significant differences between social media use and depression and anxiety, and social media use has a positive effect on both depression and anxiety. --- Comparison of paths among di erent group models Table 6 compare model fitting results based on different age and gender samples. When covariates were controlled, the effects of social media use on depression and anxiety in the youth and female groups were significant at 1% level. The standardized effect coefficients were: youth depression 0.202 , youth anxiety 0.200 , female depression 0.196 , female anxiety 0.129 . The effects of social media use on depression in middle-age and above groups were significant at 5% level , but the effects on anxiety were not significant. The effects of male social media use on depression and anxiety were significant at 10% level, with standardized effect coefficients of 0.112 and 0.118 , respectively. The only covariate that had a significant impact on depression in the youth group were SRH, the standardized influence coefficient was 0.147 , and the other covariates had no significant correlation with depression in the youth group. Age and SRH had a significant positive effect on depression in middle-age and above groups, and the standardized effect coefficients were 0.134 , 0.371 . Only age and SRH were significantly affected by depression in female groups, and the standardized influence coefficients were 0.148 and 0.263 , respectively. Education, marriage, and SRH had significant positive effects on depression in male, with standardized influence coefficients of -0.114 , -0.128 , and 0.243 , respectively. The factors that had a significant effect on anxiety in the youth group were SRH, marriage, and income with standardization coefficients of 0.308 , -0.140 , 0.090 , respectively. Factors that have a significant impact on anxiety in middle-age and above group were SRH, education, marriage, and the number of people living together, with standardized influence coefficients of 0.184 , 0.117 , -0.141 , -0.138 , respectively. Other covariates had no significant correlation with anxiety in middle-age and above group. The only covariates that had a significant impact on anxiety in the female group were education and SRH, and the standardized effect coefficients were 0.103 , 0.211 , respectively. Marriage, the number of people living together, and SRH had significant effects on male's anxiety, with standardized influence coefficients of -0.217 , -0.163 , 0.270 . Other covariates have no significant correlation with male anxiety. --- Discussion Our study explores social media use and mental health problems among Chinese urban residents during the COVID-19 outbreak in January 2021. The purpose of our research is to reveal whether the mental health of residents during the second outbreak of the COVID-19 occurred after the pandemic was largely under control in China, and whether the state of social media's impact on the mental health of residents has changed in the wake of systematic social media governance in China. . /fpubh. . Our cross-sectional study shows that more than one-third of the population suffered from depression and more than one-eighth suffered from anxiety during the second outbreak in China in January 2021. According to the latest national sample study, the prevalence of depression in China is 6.9% . These findings are consistent with previous studies that public health emergencies can lead to public mental health problems . But it is gratifying to note that our findings are in striking contrast to those of some scholars on the mental health of residents during the first outbreak of the COVID-19 in China in 2019 . These indicate that mental health problems caused by the COVID-19 outbreak are gradually being cured in China. Our study also found that different social groups have different levels of mental health, with more mental health problems among groups of young people, female, low-income, unmarried, students, poor of SRH and more of shared housing, which require special attention. Social media is one of the main sources of information on COVID-19 . Our study analyzed that during the second round of the COVID-19 outbreak in China, the average Chinese relied more heavily on social media. 57.7% of participants used traditional media, but up to 80.1% of participants used social media instead. The frequency of use of social media is significantly higher than that of traditional media. Our findings are quite the opposite of previous conclusions about the effects of social media on mental health . During China's second COVID-19 outbreak in January 2021, increased use of social media significantly improved depression and anxiety among residents, outpacing the oftenwidely accepted factors of influence such as education, income, and age. For the first time, the study found that the relationship between social media use and residents' mental health has changed radically since the first COVID-19 outbreak in 2019, and one of the reasons behind may stem from China's effective management of the "information epidemic" of social media. During the first outbreak of COVID-19 in China in 2019, social media became an "information epidemic, " and false information and reports about COVID-19 bombarded social media, sparking unfounded fears among many netizens . In addition, many citizens expressed their negative emotions on social media, such as fear and tension, which spread rapidly on social networks . These behaviors can harm people's mental health. With the "information epidemic" seriously affecting people's mental health, China has quickly and decisively implemented a series of effective governance of the social media environment: make good use of social media. On the one hand, major social media set up "anti-epidemic" module, acting as "watch sentinel" to monitor and warn of the development of the COVID-19 epidemic, improving the speed of emergency response, so that authoritative information can be first posted on social media. At the same time, a special column has been set up to allow rumors to be detected in the first place, and social media has been used to greatly improve emergency management and service levels in the government and the public sectors. On the other hand, China vigorously promotes the positive energy of society in the social media. A series of promotion about patriotism, dedication, love, and other positive energy of the social environment has been carried out through specific films, television dramas, short videos, animations, interaction, and other forms, which vigorously consolidate the Chinese cohesion. Effectively regulate social media. Both strict law enforcement and moderate tolerance of some complaints are consolidated, so that social media can act as the smooth channels to understand people's feelings, to resolve grievances of the people, and as an important helper of the government to find and solve problem. For those who maliciously produce and disseminate rumors, legal responsibility will be investigated. Our study also found some interesting conclusions. First, our study found that the impact of traditional media in China is clearly decreasing: the frequency of traditional media use is much lower than that of social media use, and it is not significantly relevant to residents' mental health. What's more, we found that the relationship between social media use and mental health varied significantly among different groups, with women and young people having a stronger correlation between mental health and social media use, while men, middle-aged and above having a weaker correlation. We recognized that women and young people are such the groups as relatively suffer from serious mental health problems, and it is precisely such the groups whose mental health is more affected by social media. To deal with this situation, special attention and help from all levels of governments and relevant departments in China are needed for young people and female groups. The social media environment should continue to be optimized, especially to focus on more information on the care and encouragement of female and youth groups. There are some limitations to this study. First, although we used the "quasi-experimental" PSM to conduct causal verification and robustness analysis, and tried to analyze the causal relationship between social media and residents' mental health as much as possible, our study is based on cross-sectional data, and it's difficult to accurately clarify the casual relationship, and it requires a follow-up longitudinal study. In addition, the survey was conducted online, so there may be some bias in the representation of the sample, such as the low participation of older persons, which may affect the results of the assessment. But the advantage of web-based investigations is that they can get data quickly in the event of another emergency. Finally, although we control as many covariates as possible, we cannot rule out residual mixing due to unmeasured factors. --- Conclusion Our research showed that the mental health problems caused by the second outbreak of the COVID-19 in China in January 2021 are less serious than the first outbreak in 2019. Simultaneously, China's social media environment has been optimized. During the second outbreak, the relationship between social media use and residents' health has changed radically, with more social media use significantly relevant to fewer mental health problems, especially among women and young people. Our research found that during the large public health outbreaks, there exists heterogeneity not only in public mental health problems, but also in the degrees of correlation between social media and residents' mental health. Our conclusions provide a basis and clue for the subsequent precise intervention of people's mental health problems during public health events. We, therefore, call for the need to continuously optimize the social media environment, with a particular focus on more information on caring, mentoring and encouraging women, youth, low-income groups, and other relatively vulnerable groups. --- 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. --- --- --- 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.

Zheng Z, Sun N, Chen Y and Chen H ( ) Social media use and mental health of urban residents during China's second COVID-outbreak.

Introduction The COVID-19 pandemic is a newly global catastrophe that is beyond the scientific postulations of the epidemiological model, philosophical insights and sociological imaginations. This pandemic, a global health emergency and the greatest challenge that humankind has faced since the 2nd World War, emerged in Wuhan, Hubei province, China in December 2019, and was named by the World Health Organization as COVID-19 . The devastating outbreak defeated all of the historical calamities, mostly the natural disasters infringed in the scientific and technological era of the past century. Till date, COVID-19 cases have exceeded 8.1 million globally, with over 440,000 fatalities and more than 3.9 million recoveries, according to the latest tally from Johns Hopkins University . In this devastating situation, billions of people are staying at home to minimize the transmission of the virus and adopting preventive measures like remote office activities, international travel bans, mandatory lockdowns, and social distancing in the form of formal and informal quarantine systems . Bangladesh, a densely populated country in South Asia, officially declared its first identified COVID-19 case on 8th March 2020 and stands at 102292 confirmed patients, ranking 18th in the world with 1343 deaths and 40,164 recovered from COVID-19 until 18 June 2020 . The rising ratio of the infection and deaths has been alarming since it would cause uncounted deaths and massacres in overpopulated Bangladesh. Bangladesh witnessed tremendous crises in combating COVID-19 due to inadequate logistic support, health facilities, health and treatment management, administrative initiatives and social dilemmas. Moreover, there is a potential of heightened mass panic, stress and discrimination in the up-coming days which can be predicted from recent protests in different communities, the spread of rumors, falsehoods, non-scientific information, limitations in governance and growing discrimination towards certain groups of populations in Bangladesh . This pandemic crisis is not only associated with the crisis of health but also with the crisis of maintaining social relations, productions, distributions, marketing, job functioning as well as multifaceted services. The obligatory lockdown, social distancing, and maintaining isolation created enormous impacts on social relation, communications and networking that affected the livelihoods of the majority of people, but its psychological and socio-psychological outcomes need to be analyzed. The pressure of maintaining social distancing has been so strong that people have been forced to pass their time in a state of fear, anxiety, stress, and loneliness during physical contact, physical attachment, face to face communication and in activities of emergency. More importantly, providing care has become emotionally difficult for healthcare professionals as they often have complex and conflicting thoughts and feelings about balancing their roles as healthcare providers as well as parents, feeling professional responsibility but also fearing their association with coronavirus patients . This complicated situation has caused stigma towards these patients due to the virus's contagious nature. --- COVID-19 Pandemic in Bangladesh It was reported that an asymptomatic family member who travelled from the epidemic center of Wuhan was most likely responsible for a familial cluster of COVID-19 infections through human-to-human transmission via droplets as the main route of transmission within a susceptible population . Regarding the issue, an airways strike was established in order to inhibit the entrance of people from the infected countries. But the late decision created the situation to be out of control in Bangladesh and many other countries. On earlier March 2020, 142 Italy returnees were allowed to go home and were advised to maintain self-isolation. A compulsory 14-day quarantine period was not ensured under the governmental surveillance of formal the quarantine system and this case was the elementary cause of the entrance of the COVID-19 pandemic in Bangladesh. Since 7 March 2020, when Bangladesh announced its first confirmed cases of COVID-19, the number of infections in the country has grown significantly due to limited testing for coronavirus infections, that made it difficult to treat patients and to confirm isolations for which the fragile public health infrastructure was highly responsible . The inadequate testing facilities, late decision of airstrikes, a limited number of formal quarantines centers, scarcity of intensive care units, lower supply of testing kits, the administrative system of lower ratio detection and lack of special hospitals for coronavirus treatment are the prime reasons of spreading COVID-19 in Bangladesh. The lockdowns became weak, and social distancing rules were not followed because of rising outward movement for earning bread, soaring public gatherings in markets for purchasing and in mosques for praying, mostly in the month of fasting . Those unstructured reasons caused administrative mismanagement, weak regulatory mechanisms and emotional social relationships. Apart from the crises, the treatment of coronavirus patients has become a greater challenge since there is no special coronavirus hospital established in Bangladesh. Blaming the lack of PPE support, Bangladesh Doctors Foundation informed that 251 doctors were found to be tested positive to COVID-19 . As a result, all of the general public and private hospitals have been the main hub for spreading the virus, causing hospitals to be a place where nobody can feel safe, further heightening fear, anxiety, and stigma. --- Social Stigma as a Barrier to Well-Being Stigma is stated as a simplified, standardized image of the disgrace of certain people that are commonly held where stigmatized people present a threat to effective group functioning . The accelerating spread of COVID-19 and its upshots has led people to fear, panic, concern, and anxiety , and thus, it constitutes stigma as the socio-psychological disease. Among various significant factors, feelings of existential insecurity and strong aloofness with diffused fear , expectations about the responsibility for the infected , and labelling behavior with a psychiatric diagnosis are the prime reasons for growing social stigma upon the coronavirus patients. The social stigma in terms of discrimination, harassments, and hatred is at rife in social communities because it is hard to determine who is carrying the virus and who is not . Moreover, a disproportionate fear, rising over a lack of information, an abundance of misinformation and an absence of trust on the health system triggered the rise of the social stigma . The intensifying stigma culture is not confined into hatred and discrimination, rather it triggers emotional protest against the COVID-19 patients and even against the construction of a hospital for the treatment of said patients. The daily New Age reported that construction works, projected by the Akij Group and Gonoshasthaya Kendra, of a dedicated hospital for coronavirus infected patients came to a halt on the last week of March 2020 in Dhaka's Tejgaon industrial area after panicked locals protested against setting up the health facility there, fearing the spread of the virus in the neighborhood . Before that, the planning of treatment for the coronavirus patients in Regent hospital of Uttara, Dhaka and the initiative to set up a quarantine centre in Uttara's Diabari were cancelled after facing protests from scared people, which further created a massive health risks in Bangladesh . When the government of Bangladesh decides that the bodies of the patients who died from Covid-19 would be buried at Khilgaon-Taltola graveyard in Dhaka city, the Daily Star reported that the locals protested the decision, insisting on finding a "safer" place outside of Dhaka instead . Thus, rumor-prone stigma rises health risks and the insecurity of dead bodies . The heightened picture of social stigma is manifested when we see that the historically and socially discriminated and stigmatized people of the Bihari community living in the Geneva Camp of Dhaka city are rejected by the local hospital where coronavirus patients are regularly treated, and instead they are forced to self-quarantine in their crowded residences. . Alongside the organizational role of stigma, the residential stigma is newly created by COVID-19. As per the newspaper report, on mid-March, a female doctor working at Dhaka Medical College received an ultimatum from her neighboring apartment owners in Old Dhaka to leave the apartment or the jobotherwise, they threatened to throw her out of the building . Another 'community-stigma' manifested an astounding state of social grievance. As per example, 'The Business Standard' reported that-after knowing an 18-member family from an Upazila medical officer at Kutubpur Delpara of Fatullah in Narayanganj as coronavirus positive, local people attacked the house throwing brickbats and stones instead of showing any kind of sympathy towards the victims . The effects of the stigma associated with COVID-19 have been so catastrophic that in many cases the corona-affected patients would not get sympathetic and technical support from their relatives, close friends and even from their family members. As per media report, on 13 April, a 50-year old mother was thrown into a jungle of Shal in Sakhipur Upazila by her one son, two daughters, and sons-in-law after perceiving coronavirus symptoms, and they falsely assured their mother saying that 'Mother, stay in the jungle only for a night, then we will come back to rescue you' . This kind of apathetic stigma has caused the disappearance of any feelings toward people's own parents, family members, close friends and relatives. The increasing cases of apathetic stigma cracked all kinds of social relationships, producing unmeasurable frustration, alienation and estrangement in full-bloom. --- Conclusions With the outbreak of the COVID-19 pandemic, people who are infected and suspected with symptoms are labelled, stereotyped, and discriminated against because of a perceived link, and therefore, the rising stigmatization among the diverse community people increases unmeasurable miseries . This situation is highly fitting to the communities of Bangladesh. All of the traditional practices of social stigma have been submerged under the powerful force of COVID-19-related social stigma. For the cause of rising social stigma, the community well-being is threatened in terms of social security, health security, livelihood, social relationship, and even safety of the burial of dead bodies. The massive effect of social stigma associating with COVID-19 is so disastrous that the previous social norms, social values, and social relations, as well as social capitals, have been destroyed with an astounding manner. The prestigious social groups such as Bangladeshi immigrants, professional doctors, and administrative officers receive social disgrace and shaming from the local people. The most vulnerable situation is that COVID-19 newly formed 'rumor-prone stigma', 'organizational stigma', 'community-stigma', 'residential stigma' and 'apathetic stigma' deepening social distance, social isolation, social ostracism, and social discontents. For those multi-faceted forms of social stigma, the social bond in our communities which caused elation and is a sign of well-being is now shattered. The traditional norm of mechanical solidarity in rural communities, and the modern form of organic solidarity is seen to be fractured. In brief, this abnormal practice of social stigma is the prime responsibility of rising unbelievable deprivation of coronavirus treatment, treatment from other diseases, and the crisis of emotional and social capital. Thereby, the COVID-19 related social stigma transforms all of the social relations into one thing-'self', but 'self' exists without well-being, and thus society exists without cooperation, and all of the artificial social relations disappear. To check the pandemics and social stigma, strengthening and decentralization of the COVID-19 medical facilities including testing, tracing, formal quarantining, and special treatment for coronavirus in coronavirus hospitals for all should be ensured by allocating a large figure of the state budget and also by taking initiatives of publicprivate partnership for health management. Government and various voluntary organizations should form a Central Stigma Management Committee and districtwise subcommittee using the relevant experts and authentic guidelines that can motivate people of every community to support their family, locals and neighbors as a form of moral responsibility. --- Compliance with Ethical Standards --- Conflict of Interest The authors declare no conflict of interest. International Journal of Community Well-Being 4:315-321 Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

This commentary looks at the social stigma as a barrier to Covid-19 responses to community well-being in Bangladesh. The Covid-19 in Bangladesh particular the way the people respond this has many dimensions to view from sociological perspective. The main objective of this commentary is to analysis how this response is related to social stigma. Gathering information from the recent literature, results showed that there are number of causes around such stigma that include misinformation, feeling of insecurity, fear of responsibility, administrative malfunction, and lack of trust on treatment. These causes of stigma have number of forms such as humor-prone stigma, residential stigma, organizational stigma, community-stigma, and apathetical stigma. Results also show that there are many effects of stigma such as health-risks, harassment, discrimination, life-insecurity, psychological disorder, loss of social capital and emotional capital, shattering family bond and social solidarity that work as barrier to community well-being. This commentary recommends to overcome the barriers through strengthening and decentralization of the COVID-19 medical facilities including testing, tracing, formal quarantining, and special treatment for coronavirus in coronavirus hospitals by allocating a large figure of the state budget and also by taking initiatives of public-private partnership for health management.

INTRODUCTION One of the aims of this paper is to report on new scenarios, technologies, and applications related to the concept of the IoB-education and discuss consequences, challenges, risks and crises on the future of business revolution. Collaboration in level of scientists, research labs and universities have considerable effects on the way in which research is conducted and on the innovation process. Therefore the partnership between universities and research and innovation centers, in running joint programs and common courses, professor and student exchanges and doing collaborative research projects, has become one of the most important issues that should be considered in future science ecosystems which are moving toward the knowledge societies. The paper also aims to cover the impact of the global crises caused by contagion of Covid/Post-Covid-19 Era on educational business and discuss about SocioEdu Consequences caused by pandemic on our teaching and learning systems, including the large-scale deployment of IoT, the employment of AI, big data and machine learning for business forecasting, the use of cloud platforms for the managing of smart education, wearables, smart city, smart society, smart healthcare, smart welfare, smart SMEs, smart retail, smart supply chain, and the future of business based on digitalization culture called ''Work-welfare 5.0 in Urban 6.0'' as solutions. key question: How to design a future scenario to apply these theories, and models to evaluate the results for forecasting, preventing, and facing the consequences of the Covid-19 pandemic? key result: A SocioEdu Blue-Green sustainability and Digital Readiness and Recovery for today's SocioEdu consequences and tomorrow's social crises key impact: Make the world a better place for living with high SocioEdu Blue-Green sustainability impacts II. BACKGROUND A. The 5 th Wave/Tomorrow Age Theory and SocioEdu sustainability: The 5 th wave/tomorrow age theory is about proceeding of future of I4.0 as a symbol for west, Society 5.0 as a symbol for non-west and edge of tomorrow that has been invented and introduced by Prof. Dr. Hamid Doost Mohammadian for the first time in 2010 and had been evaluated and improved between 2017-2019. Based on this theory, education, has a very important vital role to reach social responsibility, CSR strategies, and environmentally friendly could improve quality of livability and life based on 7PS model to reach sustainable development. 30 September -02 October, 2020, Antigua Guatemala, Guatemala LWMOOCS VII -Learning With MOOCs 2020 Generally, the 5 th wave or tomorrow age theory is a tool to achieve one of the seven pillars, such as educational sustainability; that is a path to deal with today's challenges and tomorrow's crises as well as maintain the world for future. These theory can be as a readiness for sustainability with seven pillars with a focus on education and academic aspect based on being educational focus and get ready for the academic society and innovative ecosystems in the world beyond global tomorrow's crises such as contagion of Covid19 to make the word as a better place for living. Therefore, this theory is a way to create modern business which could deal with future concerns through HR competencies, implementing, developing, and applying high 4.0 technologies like IoT and IoT-Education. Technology development has led to new opportunities for business improvement in educational sector. The world academic economy has changed from an academic society and economy to a data and innovative academic society, ecosystem and economy, leading to the concept of the "Internet of Thing-Education". Human being strategies and high technologies are fundamental tools to implement in the 5 th wave theory readiness. --- B. i-Sustainability Plus Theory as a Solution for SocioEdu Consequences of Covid-19: i-Sustaianbility Plus Theory has been invented and introduced by Doost. H. in 2010 and has been evaluated and improved between 2017-2019, which is made of the trinity open innovation, sustainability and 4.0 smart high technologies e.g. digitization and smartness. This construct which is including the idea of sustainable smart education is probed as a new idea of academic society and innovative ecosystem in tomorrow's schools, universities, and societies. The idea of i-Sustaianbility Plus theory derived from the combination of real life, high technology, and virtual reality for which again digitization is a prerequisite. In recent decades, digitization, smartness, innovation and sustainability are remarkable drivers of sustainable development. Figure 3 is presented the i-Sustaianbility Plus theory, which is made of the trinity open innovation, seven pillars of sustainability and 4.0 smart high technologies such as IoT-education: --- C. Contagion of Covid/Post-Covid19 Era: It should be noted that the quality of actions in the teaching and learning ecosystem depends on the environment and equipment available to the students. In this new situation in which the students left campuses and removed from physical resources such as libraries, computer labs and common spaces, the financial and digital divide between their homes have become even more evident. Also removing from the labs reduces the performance of professors and researchers. On the other hand, since the crisis has been going on for a long time, we see that second priority activities such as holding large scientific conferences have been prepared to be run in virtual manner. In fact, after placing too much focus on daily operational needs, now long-term investment and movement has been emerged. There are fundamental changes occurred on the work of teachers, students, researches and scientists, and on the upcoming new challenges. By passing the hard time, and adapting with the new situation, now we are looking for the synergy between elements of learning and science ecosystem. This can be done by knowledge exchange and sharing best practices. Germany has one of the best reactions to the crisis of Covid- The main activities we consider in this project are holding conferences and workshops to study and share best practices and exchange innovative ideas between Germany and Iran in the field of the project. We are going to find innovative techniques, strategies and processes to increase readiness of academic society for post-covid-19 era by: 1. Sharing the best practice between entities and countries 2. Exchanging knowledge and experience 3. Establishing educational join programs and workshops 4. Establishing a professional network of the experts in this filed from two countries 5. Doing applied and developing research project like Erasmus + 6. Preparing advice for development in a Post-COVID-19 environment --- D. Seven Pillars of Sustainability model Based on the traditional model for sustainability, sustainability has three pillars but Prof. Dr. Doost Mohammadian believes that sustainability has more pillars than these three ones. Seven pillars are required to develop sustainability. Environment, economic, social, educational, cultural, technical, and political aspects are formed sustainability. These aspects make a puzzle that all of the segments directly or indirectly are related to each other. Figure 4 is presented seven aspects of sustainability and its' classification. To achieve educational sustainability all these seven parameters should improve approximately equable. Sustainability is occurred when the figure is more regular. Generally, high quality of livability and life, health and prosperity with social justice, being environmentally friendly and preserving the earth's capacity to support future life are the main aims of the sustainable development. --- E. 7PS package model, 7PS Pattern and DPIr Models Based on the 7PS model, it is possible to measure the sustainability compass with using 7PS DPIr Model . Table 1 is presented how to measure the educational sustainability which is made of the trinity of Impact , Probability and ratio . Figure 5 is presented how to monitor SocioEdu sustainability as readiness comprehensive plan to recognize the SocioEdu consequences caused by today's challenges and tomorrow's crises. 7PS package model is combination of 4 models related to seven pillars of sustainability with a focus on educational sustainability. --- Index Description Row Si Pi Ii ri Normal --- Sustainability for each pillars Probability for each pillars Impact for each pillars Normalized ratio for each pillars 1 2 3 4 Effective Educational Si = ∑ --- IV. METHODOLOGY: To address the aims and methodology of this research is to design a future scenario to apply these theories and evaluate the results. To answer the research question and achieve the aims of this study, it is important to pursue a sound methodology. The methodology of DRM will use as a tool for analyzing. This method is used in various pillars of sustainability such as: social sciences, business economy, technical, humanity sciences, and management. This methodology is improvement of the Kiwi and Kampenhood method which comes from the oldest faith of the world; Zarathustra with the three levels: Strategy-good thoughts, Tactic-good words, and Technique-good deeds It was selected as a methodology tool. In 2009, by using the dialectical thinking related to this methodology and designed a new model to improve and revise the method. DRM is based on three levels with seven questions and analysis had been done through these questions. --- V. DISCUSION AND FUTURE STUDIES According to the Matthias Horx as a researcher for future studies in Germany, and his colleagues at the ''Institute for Future Research'' have recently published a book entitled ''The World After Corona'', which has been widely reported in the German medias. Referring to the crisis caused by the outbreak of the disease, he writes: "These days I'm often asked, 'When will Corona's era end and we'll be back to normal? ''I say: never. There are some historical periods that change the course of the future. We call these periods a "deep crisis" We are now in this position and the result of his team research'' The world will change: The contagion of Covid-19 made us realize how much we do not understand. / Families, neighbors, and friends became closer Behaviors will change: Digital culture technology has taken its place in practice. / Most co-workers, who previously avoided remote video conferencing and preferred mission flights, now realize that it's a more practical and constructive approach. / Education stakeholders have learned a lot about Internet education, and working in online form Humanity will return: The medical staff helps, but along with advanced medical techniques, our social behavior is also crucial. / After the outbreak, we look back and remember how much humanity was really before contagion. The economy after the contagion crisis: Humans will also look at the economy to see how far the economy has been weakened, without dying. / Although we have experienced infrastructure damage and stock markets have fallen by 50 percent and many companies have gone bankrupt, we have never reached zero, the notion of economics is a living being that can sleep and even dream but continue to live. --- The main losers of the contagion of the Covid-19: Those who want to incite people against each other are not playing a role in future issues He has shown the result of a research concerning the most likely challenges, risks and crises in next year as below: The 31 risks have been grouped into five major categories: The theoretical basis for this approach is diverse, but the focus lies on "the 5 th wave and i-Sustaianability theories" which have implications on this priority: 1) SMEs with environmental responsibility including a focus on blue-green sustainability, 2) SMEs with social cohesion and 3) SMEs with economic efficiency . Another aim of this proposal is running a series of workshops across the globe in partnership with global and local networks, to help SMEs of all sectors understand how they can build back better in a post-COVID world. This pandemic throws a spotlight on the interdependences between business, nature, and society. If you didn't think that encroaching on natural ecosystems or protecting public health were relevant to your business, you almost certainly see the links now. The most important advantages of new technology are: Accelerating innovative processes Realizing new chances in markets Creating new ideas and using them into innovations Modeling products and services into market to test them in order to find out new results --- Developing businesses with maximum profit and effectiveness In addition, high SocioEdu sustainability that includes environmental responsibly, economic efficiency,and social cohesion, with business and economic efficiency. High educational technologies and social, can play an important role in achieving this theory. 30 September -02 October, 2020, Antigua Guatemala, Guatemala

The Covid-19 pandemic is changing our society on diverse dimensions (not only on an economic or social level). Before the 1970s, various businesses and economics could affect and improve technologies, and humans' life. Since we reached the 70ies, and with the beginning of the 3 rd wave or the post-industrial era, the new modern technologies such as Information Technology (IT), Internet of Things (IoT), Internet of Business (IoB), Internet of Energy (IoE), Internet of Manufacturing (IoM), Internet of Health (IoH), the emerging digitization, and innovative smartness in addition, issues such as knowhow, do-how, Artificial Intelligence (AI), and others could influence businesses, the economy, urban infrastructures, societies, social impacts and even human life. Wars, sanctions, sciences, technologies, IT, IoT, and smartness work together and have impacts on each other's, businesses, economy, and human life. Fundamentally, during humans' history inventions and innovations have been created through restrictions and challenges. Wars, sanctions, low sustainability, risks of contagion, recessions, climate pollution, environmental change, biodiversity collapse, urban sprawl, growing urbanizations, lack of infrastructure, gentrification, economic instabilities, and recently health crises are introduced as main restrictions. Wars and welfare integrate with human's life desirably or undesirably and make innovation. Improvement through such innovative digital technologies are so admirable because most of them have changed the world. Sustainability has an important role for societies and urbans to improve their quality of livability and life. The world academic economy has changed from an academic SocioEdu economy to a data and innovative academic SocioEdu, ecosystem and economy, leading to the concept of the "Internet of Business-Education". with many popular IoT applications involved (i.e. wearables, smart city, smart society, smart healthcare, smart welfare, smart SMEs, smart retail, smart supply chain). This pandemic throws a spotlight on the interdependencies among business, nature, society, economy, and education. It may be tempting right now, when it is hard to see beyond the next few weeks, to dismiss the 17 Sustainable Development Goals (SDGs) as a distraction. But they have been described as a "crowd-sourced purchase order from the future" precisely because they offer a tremendous business opportunity. Based on SDGs all developed and developing countries in a global partnership have to go with CSR strategies and Hybrid Businesses for improving social cohesion and welfare, health, education, inequality, blue-green environmental sustainability, and economic efficiency. In this article we will discuss how the 5th Wave/Tomorrow Age and i-Sustainability Plus theories could help us for a readiness to forecast, prevent and face to the SocioEdu Consequences of Covid-19.

Introduction Computer vision applications have become ingrained in numerous aspects of everyday life, and problematically, so have the societal biases they contain. For example, gender and racial biases are prevalent in image tagging [62,7] and image search [39,52]; visual recognition models have disparate error rates across demographics and geographic regions [16,24]. The perpetuation and amplification of social biases precipitate the need for a deeper exploration of these systems and of the bias propagation pathways. We focus on the task of image captioning: the process of generating a textual description of an image [69,50,76,48,3,33]. This task serves as an important testbed for visual reasoning and can improve accessibility of digital images for people who are blind or low vision. In this work, we assess the pathways for bias propagation: from the images, to the manual captions, and finally to the automatically generated captions. We focus our attention on studying the Common Objects in Context [47,19] dataset; it is a widely used image captioning benchmark [32], thus making any biases especially problematic [22]. We collect both skin color and perceived gender annotations on 28,315 of the people in the COCO 2014 validation dataset after obtaining IRB approval. This data allows us to analyze disparities in image captioning across different demographics. Concretely, we observe: • The dataset is heavily skewed towards lighter-skinned and male individuals. 1 Further, darkerskinned females are especially underrepresented, appearing 23.1x less than lighter-skinned males. • There are racial terms in the manual captions. The racial descriptors are not learned by the older captioning systems [59,50], but are learned by the newer transformer-based models [67] -although the slurs do not yet appear to be learned. • Image captioning systems perform slightly better on images of lighter-skinned people. This is consistent with disparate accuracies on e.g., pedestrian detection [73] and facial recognition [16]. • There are visual differences in the depictions of lighter and darker-skinned individuals. For example, lighterskinned people tend to be pictured more with indoor and furniture objects, whereas darker-skinned people tend to be more with outdoor and vehicle objects. • Even after controlling for visual appearance, the captions still differ in word choices used to describe images with lighter versus darker-skinned individuals. This is particularly apparent in the manual captions and in modern transformer-based systems. Our work lays the foundation for studying bias propagation in image captioning on the popular COCO dataset. Data and code is freely available for research purposes at https://princetonvisualai.github. io/imagecaptioning-bias/. --- Related Work Presence of dataset bias. Our work follows a long line of literature identifying, analyzing, and mitigating bias in machine learning systems. One key facet of this discussion is the bias in datasets used to train models. Under the framework of representational harms [14,6], there is commonly a lack of representation [16,74] and stereotyped portrayal [17,62,54,65] of certain marginalized demographic groups. Along with many ethical concerns [56,66], these dataset biases are problematic because they can propagate into models [15,17]. In this work we analyze the biases present in a commonly-used image captioning benchmark, COCO [47,19], using our new crowdsourced annotations. Mitigating dataset bias. The root causes of dataset bias are complex: they stem from bias in image search engines [52], data collection practices [74,24], and real-world disparities. Proposed solutions to dataset bias include new data collection approaches [38,35], manual data cleanup [74,75], synthetic data generation [58,20,60] -or, in extreme cases, even withdrawing the dataset after insurmountable biases have been identified [13]. Researchers have advocated for increased transparency of datasets [27,35], including developing tools to steer researcher intervention [8,70]. Our work does not aim to mitigate dataset bias but instead to articulate its impact on downstream image captioning models. Algorithmic bias mitigation. In tandem with efforts to reform data collection, a variety of algorithmic bias mitigation techniques have been proposed; see e.g., Hutchinson and Mitchell. [34] for an overview. This work goes along with others that unveil biases present in existing algorithms [53,49,4]. One important theme is bias amplification [78,71,72], or social biases in the data getting amplified in the trained models. In this vein, we study how bias in manual image captions propagates into automated captioning systems. Image captioning models. Image captioning models are increasingly being developed as a more complex way of labeling images [69,50,76,48,3,33]. Recent work has discovered biases in these systems, but often with respect to gender [31,12,64]; the study of racial biases in captioning has been limited to analyzing bias in the manual captions [54,65]. Racial bias has been identified in other automated systems [9] ; here we expand this work to studying racial biases in image cap- tioning. This spurs the important question of whether race should be included in generated image captions at all. Prior works [63,51] find that, in certain contexts, people who are blind or low vision want racial descriptors to be included. Further, this motivates the need to understand how people prefer their identities labeled by an automated captioning system, a question studied extensively by Bennett et al. [10]. --- Crowdsourcing Demographic Annotations --- Annotation process Dataset. To study bias in image captioning systems, we collect annotations on COCO [47,19], a large-scale dataset containing images, labels, segmentations, and 5 humanannotated captions per image. COCO is a widely used image captioning benchmark. We focus on the 40,504 images of the COCO 2014 validation set, and look for person instances with sufficiently large bounding boxes such that there is a reasonable expectation of being able to infer gender and skin color. This results in 15,762 images and 28,315 person instances. Annotation setup. Using Amazon's Mechanical Turk , we crowdsource race and gender labels. In our interface , we present workers with a person instance in a COCO image and ask them to provide the skin color using the Fitzpatrick Skin Type scale [25], ranging from 1 to 6 , and the binary gender expression. We also give workers the option of marking "unsure" for either. Each instance is annotated three times. We compensate the workers at a rate of $10 / hr. Inferring race and gender. Race and gender annotations are fundamentally imperfect [29,41,61]. First, the annotated labels may differ from the person's identity. Second, the labels are discretized . Further, the labels are for social constructs and thus subjective and influenced by the annotators' perceptions. We follow prior work [16,73] in formulating our annotation process; we use phenotypic skin color as a proxy for race because of its visual saliency over other conceptualizations of race. However, as noted by Hanna et al. [29], we are actualizing a particular static conceptualization of observed race here. By operationalizing race this way, we miss differences that may appear in other operationalizations, such as racial identity. Quality control. To ensure annotation quality to the extent possible, we limit the task to workers who have completed over 1,000 tasks with a 98% acceptance rate. We also construct 57 gold standard images where the gender and lightor-dark labels were agreed-upon by five independent annotators, including one of the authors. We inject 5 of these images randomly in a task with 50 images, and only allow workers who have correctly labeled these images to submit. --- Gender annotations We start by analyzing the collected gender annotations, looking at distributions at both the instance and image level. Instance-level annotations. We analyze the gender annotations of the 28,315 person instances. To determine the label for a person, we use the majority over the three annotations. If majority is not achieved, or there are contradictory gender labels, the instance is labelled as no consensus. We observe that contradictory gender labels are most common when the person is a child, has obscured facial features, or possesses features that contradict social gender stereotypes . Analyzing the distribution, we see that males make up 47.4% of the instances compared to females who only comprise 23.7% . Most of the remaining instances were annotated unsure , and a consensus was unable to be reached for only 2.2% of instances. Image-level annotations. To analyze the dataset at the granularity of images, which is what the captions refer to, we map individual instance annotations to the image . We use the annotations given to the largest bounding box, under the assumption that captions will mainly refer to the largest person in the image [11]. The only exception is if the second largest bounding box contains an individual of the opposite gender, and is more than half the size of the largest bounding box. In this case, we categorize the image as both. The image-level distribution closely mirrors that of the instance-level . Again, there are more than twice as many male images as female images . Comparing collected gender annotations with automatically derived ones [78]. Previously, works looking at gender bias in COCO have used gender labels derived from the manual captions: "[if] any of the captions mention the word man or woman we mark it, removing any images that mention both genders." [78] We compare our annotations with theirs. They label 5,413 images: our labels agree with theirs on 66.3% and disagree on 1.4%; the remaining 32.3% we determine cannot be reliably labeled with one gender, e.g., because the person is too small or there are multiple people of different genders in the image. We successfully label 10,780 images; they only label 3,591 of these correctly . This is consistent with the argument of Jacobs and Wallach [37]: gender is operationalized differently in caption-derived versus human-collected annotations. --- Skin color annotations For the skin color annotations, we follow a similar process as with our gender annotations. The only difference is that we add a method for dividing skin color into the broader categories of lighter and darker. Using these new categories, we similarly analyze the skin color distribution at both the instance and image level. Instance-level skin color distribution. Using the same schema as in Sec. 3.2, we obtain instance-level annotations for skin color. The top two most frequently occurring Fitzpatrick Skin Types are 2 and 1 . In contrast, Fitzpatrick Skin Types 5 and 6 comprise only 1.9% and 1.7% of the instances, respectively. This underrepresentation of darker-skinned individuals is an example of representational harm in and of itself. We also include a broader skin color breakdown consisting of two categories: lighter and darker. Following previous work [16], we define the lighter category as all instances rated 1-3 on the Fitzpatrick scale and darker as containing 4-6. We also assign some of the instances that were previously uncategorized by skin color to these broader categories. Using this skin color breakdown, 61.0% of the instances are lighter individuals, whereas only 8.1% are darker individuals. The amount of no consensus instances decreases from 15.4% to 13.9% when using this breakdown. Image-level skin color distribution. At the image-level, we categorize skin color as lighter and darker, employing the same consensus method as for gender in Sec. 3.2. Of the images, 64.6% are part of the lighter category and 7.0% are part of darker, meaning there are 9.2x more lighter-skinned images than darker-skinned. --- Intersectional analysis. We analyze the skin color and gen-Figure 2: The results of our crowdsourced demographic annotations on the COCO 2014 validation dataset, as well as the self-disclosed demographics of the annotators. Left column: distribution of perceived skin color and gender expression of the 28,315 people instances. Middle column: distribution after collapsing individual annotations into image-level annotations . Lighter-skinned people and people who are male make up the majority of their respective categories. Right column: self-reported demographics of AMT workers. der labels in tandem. Within lighter images, males are overrepresented at 52.8% compared to females at 25.7%. However, this difference is even starker when looking at darker images, where males comprise 65.1% of the images while females only make up 20.6%, reflecting the unique intersectional underrepresentation faced by darkerskinned females, as noted by Buolamwini and Gebru [16]. In fact, of the 15,762 images annotated, only 226 of them are of darker-skinned females. Worker information. AMT workers were asked to optionally disclose their own race and gender identity. Of the workers asked, 97.9% provided their gender and 97.3% provided their race. As seen in Fig. 2, the annotators are predominantly white and male . Prior work has found that annotators describe in-group versus out-group members differently [54]. Thus, there may be a concern that the skew in worker demographics could influence our collected labels. To understand whether a worker's demographics influences their selection of labels, we explore disagreements in annotations. We do so by comparing the mean difference in annotation when the pair of workers are of the same self-reported demographic group versus when they are of differing groups. If workers from different groups label images differently, we would expect pairs from distinct groups to have a greater disagreement than pairs from the same group. However, we find for skin tone there is not a substantial difference in the disagreement between pairs of the same racial group and different groups . For gender, the mean difference for same gender pairs and different gender pairs is similar as well. This indicates that there is not a systematic difference between how workers of different self-reported demographic groups label images, suggesting our collected labels would be similar even if the workers came from a different demographic composition. --- Experiments We now discuss the findings from our experiments on understanding what kinds of biases propagate in image captioning systems. First, we examine racial terms and disparate performance . We then analyze bias in terms of representation, i.e., differences between the lighter and darker images and corresponding captions. To do this we first consider the images in Sec. 4.3, before controlling for these visual differences and studying the captions in Sec. 4.4. --- Models. We examine the captions generated by six image captioning models: FC [59] is a simple sequence encoder that takes in image features encoded by a CNN; Att2in [59] is similar but images are encoded using spatial features; DiscCap [50] further adds a loss term to encourage discriminability; Transformer [67], AoANet [33], and Oscar [45] are transformer-based models representing the current state-of-the-art. In our analysis we particularly focus on contrasting Att2in vs DiscCap, since they differ only in the added discriminability loss, and the older vs the newer models. We train the models on the COCO 2014 training set using proposed hyperparameters from the respective papers . Oscar is further pre-trained on a public corpus of text-image pairs Data. Our racial analysis is performed on 10,969 images of the COCO 2014 validation set which were definitively labeled as either lighter or darker . --- Captions contain racial descriptors We begin by analyzing the presence of racial descriptors and offensive language in the manual as well as automatically generated captions. Manual captions. Prior works [54,65] show that people are more likely to use racial descriptors when describing non-white individuals. We observe this pattern in humanannotated captions by conducting a keyword search of the captions in the COCO 2014 training set using a precompiled list of racial descriptors . For ambiguous terms that can be used in a non-racial context, we manually inspect the captions. Assuming the training distribution mirrors that of the validation, for the manual captions, annotators used racial descriptors to describe individuals who appear to be white 0.03% of the time versus 0.54% of the time for individuals who appear to be Black. Furthermore, in 26.9% of the instances when a racial descriptor for a white individual is used, the annotator is also mentioning an individual of a different race in the caption as well . We see this as a manifestation of the belief that "white" is the norm, and race is only salient when there is a deviation or explicit difference between multiple people. In addition to looking for racial descriptors, we check for the presence of slurs and offensive language using a precompiled list of profane words [77]. There are 1,691 instances of profane language, occurring in 0.40% of the sentences in the COCO 2014 training set. We find alarming occurrences not only of racial slurs but also of homophobic and sexist language as well, similar to the NSFW discoveries by Prabhu and Birhane [13]. Automated captions. Racial descriptors are not found in the automated captions generated by FC, Att2In, DiscCap, AoANet, or Oscar. While this may be attributed to the fact that racial descriptors are uncommon in the training set, we disprove the idea that this is wholly the reason. To do so, we observe that other words which occur at similar rates Figure 3: Examples of images for which the Transformer model [67] assigns racial or cultural descriptors to the caption. While in the first image the descriptor of "Asian" is present in the human-annotated caption, neither of the descriptors, "Indian" nor "Mexican," are applicable in the latter images. are in fact still present in the model-generated captions. For example, the word "Japanese" occurs 69 times in the training set and 0 times in AoANet-generated captions while other descriptors, such as "uncooked" and "soaked", which appear 88 and 61 times in the training set, occur 2 and 6 times in the generated captions respectively. While it is rare, we find that racial and cultural descriptors as well as offensive language do propagate into the captions generated by the newer transformer-based models. For Transformer, AoANet, and Oscar, we find instances of offensive language. In addition, there are racial descriptors in 2 of the captions generated by Transformer and 12 cultural descriptors. Furthermore, for 10 of the 14 images, the model uses these descriptors when the human captions do not contain any racial or cultural descriptors . This leads to the worry that models may replicate offensive language or exploit spurious correlations to assign descriptors in a stereotypical and harmful way. --- Performance differs slightly between lighter and darker images We next evaluate whether image captioning models produce captions of different qualities on images with lighterskinned people than darker-skinned people. To do so, we first assess the differences in BLEU [55], CIDEr [68] and SPICE [2] scores between captions on lighter and darker images. Both BLEU and CIDEr rely on n-gram matching with BLEU measuring precision and CIDEr the similarity between the generated caption and the "consensus" of manual captions. SPICE, however, focuses more on semantics, capturing how accurately a generated caption describes the image's scene graph . From these results , we make two key observations. First, according to both BLEU and CIDEr, the Table 1: The differences in captioning performance as measured by BLEU [55], CIDEr [68], and SPICE [2] multiplied by 100 on the COCO 2014 validation dataset. Error bars represent 95% confidence intervals across random seeds used to train 5 models per architecture. --- BLEU CIDEr SPICE FC [59] 0.5 ± 0.5 -0. This suggests that the way models are choosing to describe the images may be better-suited for the majority group. In fact, we see there is a slight positive correlation between the performance of the model and the differences in performance between the two groups with an R2 of 0.343 . Second, there are no noticeable differences with SPICE, indicating that the captions identify key visual concepts equally accurately across both groups. Nonetheless, it is important to note that negative results do not indicate something is bias-free, but merely that our particular experiment did not uncover strong biases. --- Visual appearance differs between lighter and darker images The analyses so far only consider issues in the captions themselves, irrespective of the image. We now explore how the visual depictions of people of different groups differ. We analyze simple image layout statistics, apply the RE-VISE [70] tool for discovering bias in datasets, and consider differences in visual appearance of the image content. We split our skin-tone-labeled image dataset of 10,969 images into 9,609 images for training and 1,360 for testing. 2 We use area under the ROC curve as our metric on a balanced test set, so random guessing would have an AUC of 50%. We bootstrap over 1,000 Figure 4: Regressing model performance, as measured by CIDEr [68], against difference in performance suggests that as performance increases, the difference may correspondingly increase as well . The horizontal and vertical error bars in the graph represent the 95% confidence intervals for the performance and differences, respectively resamples and report a 95% confidence interval. Image layout statistics. We consider the following simple image layout statistics as our features: number of people in the image, largest person bounding box size, distance of the largest bounding box from the center of the image, and gender one-hot coded. We train logistic regression models using LBFGS through the sklearn package [57] to predict whether the input corresponds to the lighter or darker label. An ability to classify serves as a signal for how distinguishable the input features of the two groups are. We use a balanced class weight and run five-fold cross-validation to tune the L2 regularization hyperparameter . Our two best performing models are trained on the distance from center and the distance plus the gender. Distance alone achieves an AUC of 56.6 ± 5.2; adding gender increases the AUC to 57.8 ± 4.9. Distance is predictive because darker-skinned individuals tend to be further from the image center than lighter-skinned individuals; this is troubling since the "important" parts of an image tend to be more centered [11]. Gender is a useful feature since from Sec. 3.3 we know that the gender distribution differs between the two groups. REVISE [70] bias discovery. We next apply the REvealing VIsual biaSEs tool. 3 Using REVISE we discovered that darker-skinned people appear more frequently with outdoor objects, and lighter-skinned people appear more frequently with indoor objects . Specifically, objects like sink, potted plant, and toothbrush all appear with lighter-skinned people over 13x as much as with darker-skinned people, despite lighter-skinned people only appearing in 7x as many images as darker-skinned peo-Figure 5: Images with people of lighter and darker skin tones co-occur with object categories at different frequencies. Whereas the former tend to be pictured with object categories that are indoor, the latter tend to be pictured with object categories that are more likely to be outdoors. ple. Although at the moment the differences in object cooccurrences do not appear to have noticeable downstream effects , these differences may lead to discrepancies in performance as certain objects become more easily identifiable for different skin tone groups. Visual appearance. Finally, we use image classification models for a detailed examination of how the content of the images differs between different skin tones. To ensure that the skin color of the pictured individual does not affect the model's prediction, we use COCO's object-level segmentations to mask all the people objects. We fill in these masks with the average color pixel in the image. Using the masked images, we fine-tune a pre-trained ResNet-101 [30] over five epochs using the Adam optimizer [42] and a batch size of 64. We oversample the darker images to account for the imbalanced class sizes. During training, the learning rate is initialized to be 0.01 and decays by a factor of 0.1 after three epochs. The model achieves an AUC of 55.4 ± 4.9, indicating that there is a slight learnable difference between the scenes of lighter and darker images. --- Captions describe people differently based on skin tone Finally, we consider how both manual and automatic captions differ when describing lighter versus darker images. To do so, we first control for the visual differences, in order to disentangle the issues coming from the image content versus from the words used in the caption. We do so by finding images that are as similar as possible in content, and differ only by the skin color of the people pictured, i.e., constructing counterfactuals within the realm of our existing dataset. Concretely, for each darker image, we find the corresponding lighter image that minimizes the Euclidean distance between the extracted ResNet-34 fea-Table 2: Three bias analyses on manual and automated captions of images for which visual content has been controlled. For the first column the VADER sentiment score [36] is multiplied by 100. For the last two columns, the number is AUC×100 for classification ability, where higher numbers indicate a greater ability to distinguish between the two groups. Error bars represent 95% confidence intervals across random seeds used to train 5 models per architecture. --- Sentiment Embedding Vocab The leftmost images represent an example pair from the most similar top 20% of pairs, and the rightmost represent the bottom 20%. We pick 40% as our threshold for controlled images to include. tures [30] of the masked images using the Gale-Shapley algorithm [26] for stable matching . After examining the results, we select the top 40% most similar image pairs. The resulting dataset has 876 images. When needed, we use 700 for training and 176 for testing ; otherwise we compute statistics over the whole dataset. As expected, a visual classifier trained on these images achieves an AUC of only 44.7 ± 9.3, failing to differentiate between the two groups. In the following analyses, we use the same six models and training setup as in previous experiments. However, we use the dataset, introduced above, which consists of 876 unmasked images for evaluation. This data thus allows us to examine whether human-annotated and model-generated captions diverge even when visual differences are controlled. --- Sentiment Analysis For our first line of inquiry, we use the Valence Aware Dictionary and Sentiment Reasoner [36] to perform sentiment analysis on the human-annotated captions. Limitations include that sentiment analysis tools have been shown to encode societal biases themselves [43,23], and may not generalize well to out-of-distribution machinegenerated text. VADER returns a compound polarity score from -1 to 1 . Scores less than -0.05 are considered negative; scores greater than 0.05 positive. We find that human-annotated captions describing lighter images have a mean compound score of 0.073 ± 0.01 whereas those describing darker images have a mean compound score of 0.059 ± 0.01. The difference in compound scores is statistically significant , with captions describing lighter images being more positive. We find that automated captioning systems do not appear to amplify the difference in sentiment scores between the two groups . The lack of difference is largely due to the fact that automated captions tend to be more neutral than the human-annotated ones, thus removing most of the sentiment. In fact, the compound scores were all less than 0.03, excluding scores for captions generated by Transformer . --- Sentence embedding differences For our next analysis, we use sentence embeddings from the Universal Sentence Encoder [18] to compare how the semantic content of captions differs between lighter and darker images. To note, racial descriptors in the captions are not removed for this experiment. We train a multilayer perceptron classifier on the embeddings and run five-fold cross validation to tune the learning rate and number of epochs . We find that the classifier can differentiate between the captions with an AUC of 68.9 ± 3.5, indicating a learnable difference in the resulting caption content despite the visual content being indistinguishable. We see in Tbl. 2 that the ability to differentiate based on embeddings drops in the generated captions, especially for the more advanced Transformer model to 54.2±3.1, which is almost random. Although humans appear to be assigning different content to similar images with people of different skin tones, automated captioning models do not appear to uphold this trend, at least with respect to the particular sentence embeddings we use. --- Vocabulary differences Finally, we consider word choice in the captions. We use a logistic regression model and a vocabulary of the 100 most commonly used words in the COCO 2014 training set. Our features are size 100 binary indicators of whether a particular word is present in a caption. The classifier achieves an AUC of 61.8 ± 3.8 on human captions. Beyond the differential use of racial descriptors we already observed in Sec. 4.1, this suggests annotators use different vocabularies to describe images even with similar visual content . The ability to distinguish between lighter and darker images further increases when automated captions are used. Particularly, in Tbl. 2 we see from Att2in to Dis-cCap and FC to Transformer, the AUCs slightly increases from 62.8±1.5 to 63.0±3.2 and 65.9±4.2 to 66.0±1.4, respectively. From FC to AoANet, there is a greater increase in AUC from 65.9 ± 4.2 to 68.0 ± 1.8. We do note that, for Oscar, the ability to differentiate based on vocabulary decreases compared to FC as the AUC drops from 65.9 ± 4.2 to 64.4 ± 2.8. This may be due to the fact that Oscar is pre-trained on a larger corpus of data; the greater dataset diversity may help diminish the differences between the vocabularies used. Overall, this leads us to believe that more advanced models are more likely to employ different word choices when describing different groups of people. Interpreting these results relative to that of the previous section in which we found that the semantic content of generated captions did not differ much between different groups, we consider whether different words are being used despite caption content being similar. As an example, the sentences "Apples are good." and "Apples are great." may map to similar sentence embeddings, but the specific word choice employed is different. In this vein, we find, for instance, that on AoANet's captions, the average coefficent of the word "road" is 0.226 higher than that of the word "street" , even though upon manual inspection the images being described are similar . While differences in the usage of words, such as "road" and "street," are relatively innocuous, these subtle differences in vocabulary may become more problematic when we consider how certain words like "articulate" have developed a different meaning when applied to Black people [21,1]. Thus, in future work, it is important to consider not only the semantic differences captured in the sentence embeddings but also the specific words being employed. --- Discussion and Conclusion In this work, we seek to understand not only what racial biases are present in the COCO image captioning dataset, but also how these biases propagates into models trained on them. We annotate skin color and gender expression of people in the images, and consider various forms of bias such as those in the form of differentiability between dif-ferent groups. We find instances of bias in the dataset and the automated image captioning models. However, we are careful to note that cases in which we did not find bias do not mean there are not any, merely that our particular experiments did not uncover them. By looking at the models that seem to be most indicative of where the image captioning space is progressing, we can see that the bias appears to be increasing. For researchers, this serves as a reminder to be cognizant that these biases already exist and a warning to be careful about the increasing bias that is likely to come with advancements in image captioning technology. Based on these analyses, we propose directions for mitigating the biases found in captioning systems. First, from our findings in Sec. 3.2 and 4.1, we see that human annotators make assumptions about the demographics of people pictured or use different language when describing people of different skin tone groups. To mitigate this, dataset collectors can provide more explicit instructions for annotators . In addition, we also find that ground-truth captions contain profane language . In line with existing mitigation efforts [74,13], manual captions containing slurs or other offensive concepts should be removed from the dataset. Additionally, in Fig. 2 we see that only 7.0% of the dataset contained images of people with darker skin tones, i.e., 1096 images. We need to collect more diverse datasets such that we can measure disaggregated statistics and compare metrics such as the difference in SPICE scores with the knowledge that our measurements do not suffer from a high sampling bias. Finally, from our analysis of generated captions , we note that Oscar exhibits less bias compared to the other transformer-based models. This suggests the greater dataset diversity from pre-training the model may help reduce the amount of bias that propagates into the automated captions. 40.0 ± 0.4 120.0 ± 0.5 23.0 ± 0.3 --- D. Vocabulary differences coefficients In Sec. 4.4 we explore the different word choices in the captions describing lighter and darker images. We provide the most predictive words for lighter and darker across the manual captions and the automatically generated captions in Tbl. 5. Table 4: The differences in captioning performance as measured by BLEU [55], METEOR [5], ROUGE, [46], CIDEr [68], and SPICE [2] multiplied by 100 on the COCO 2014 validation dataset. Error bars represent 95% confidence intervals across random seeds used to train 5 models per architecture BLEU ∆ METEOR ∆ ROUGE ∆ CIDEr ∆ SPICE ∆ FC [59] 0.5 ± 0.5 -0.1 ± 0.2 0.0 ± 0.5 -0. The most predictive words of lighter and darker images for a logistic regression trained on manual and automated captions of images for which visual content has been controlled. Lower coefficients are more predictive of lighter. For the automated captions, the coefficients are averaged across the 5 models per architecture. Human FC [59] Att2in [59] DiscCap [50] Transformer [67] AoANet [33] Oscar [ --- Appendix A. Comparing collected gender annotations with automatically derived ones We explore extending the schema introduced in Sec. 3.2 for deriving gender labels from captions in three ways: 1) only labeling images where there is a person who has a bounding box greater than 5,500 pixels, 2) expanding the list of gendered words beyond "man" and "woman", and 3) having different cutoffs for how many captions need to mention a gender for the image to be labeled. We call the use of the gendered set {man, woman} "few," and that of our expanded set "many." Our expanded set "many" consists of the following words: ["male", "boy", "man", "gentleman", "boys", "men", "males", "gentlemen"] and ["female", "girl", "woman", "lady", "girls", "women", "females", "ladies"]. Our results in Fig. 7 show that while these extensions significantly increase both the number of images correctly labeled and the accuracy of labeled images, all methods are inaccurate and/or incomplete. The gender labels derived from captions remain highly imperfect, as expected, cautioning against automated means of gender derivation [40]. --- B. Racial descriptors When searching for descriptors of race and ethnicity in Sec. 4.1, we first convert the captions to lowercase. We then use the following keywords ["white", "Caucasian", "Black", "African", "Asian", "Latino", "Latina", "Latinx", Figure 7: Comparison of various ways of automatically deriving image-level gender annotations from existing image captions. "Few" and "Many" gendered words refers to the size of the set of gendered words considered, and numbers refer to required captions that mention a gendered word, higher numbers limiting the images that can be labeled. Percentage over each bar indicates accuracy. All methods are imperfect and noisy, cautioning against the use of automatically-derived gender annotations. "Hispanic", "Native", and "Indigenous"] -also in lowercase -to query the captions. --- C. Caption performance In Sec. 4.2 we assess the differences in caption performance for BLEU [55], CIDER [68], and SPICE [2] when evaluated on the COCO 2014 validation set. We extend this analysis by providing the overall scores across the four image captioning models and looking at two additional automated image captioning metrics. To start, we look at the performance for our four models. As seen in Tbl. 3, Oscar has the best performance across all metrics. Further, we see that the newer transformer-based models outperform older models across all metrics as well. We also report the differences in performance between lighter and darker images for two commonly used image captioning metrics -METEOR [5] and ROUGE [46] . Similar to the results for BLEU and CIDER, the differences for METEOR and ROUGE are greater for Att2in, Transformer, and Oscar. AoANet also shows some slight differences in performance for ME-TEOR and ROUGE. This supports our observation that the better performing captioning models also tend to show greater discrepancies in performance between lighter and darker images.

Image captioning is an important task for benchmarking visual reasoning and for enabling accessibility for people with vision impairments. However, as in many machine learning settings, social biases can influence image captioning in undesirable ways. In this work, we study bias propagation pathways within image captioning, focusing specifically on the COCO dataset. Prior work has analyzed gender bias in captions using automatically-derived gender labels; here we examine racial and intersectional biases using manual annotations. Our first contribution is in annotating the perceived gender and skin color of 28,315 of the depicted people after obtaining IRB approval. Using these annotations, we compare racial biases present in both manual and automatically-generated image captions. We demonstrate differences in caption performance, sentiment, and word choice between images of lighter versus darker-skinned people. Further, we find the magnitude of these differences to be greater in modern captioning systems compared to older ones, thus leading to concerns that without proper consideration and mitigation these differences will only become increasingly prevalent. Code and data is available at https://princetonvisualai. github.io/imagecaptioning-bias/.

BACKGROUND Sex ratios at birth , calculated as males:females , are widely used as a measure of gender bias within a population. [1][2][3] Landmark work conducted some four decades ago in India and China highlighted imbalances in M:F birth ratios above the normative 105:100, an artefact of deliberate sex selection resulting in excess prenatal female mortality estimated at millions. 2 4 5 Widely referred to as the 'missing women' phenomenon, skewed M:F ratios are unequivocally accepted as reflecting son preference and undervaluation of girls and women; as Sen 5 noted, they 'tell us, quietly, a terrible story of inequality and neglect' . Extensions to this scholarship more recently has focused on migrants of Asian heritage residing in 'western' societies. Burgeoning work in Canada, [6][7][8][9][10][11][12][13][14] the UK, [15][16][17][18] the USA, [19][20][21][22][23][24] Sweden, 25 Norway, 26 Australia 27 and Spain, 28 has documented male bias in sex ratios largely among their Indian, Chinese and Korean migrant communities. Overall, analyses of sex ratios in individual country contexts estimate anywhere from 2000 at a minimum to 4000 maximum missing women between the 1980s and 2000s. Skewed M:F ratios among migrant Asian communities largely present among firstgeneration migrants, especially where the mother is overseas-born 8 10 13 15 27 although --- Strengths and limitations of this study ► This study uses whole of population data. ► Two independent data sets, the Longitudinal Census of New Zealand and Birth Register Cohort Series, were used for corroboration. ► The study uses administrative data which is not specifically collected for the purposes of this research that may affect the quality of the analysis. ► The use of proxies may affect the accuracy of the analysis. ► The study is limited by its exclusive use of live birth ratios and future research may include stillbirths and abortion data, should reliable data be available. Open access recent research in Canada shows continuation of these trends among second-generation South Asian women. 6 Studies point to imbalances in M:F ratios at higher parities especially where the first two children are girls. 13 15 20 22-24 27 29 30 The practice is found among affluent, urban migrants, ruling out economic need as the exclusive driver of sex selection. 17 18 Instead, residual cultural norms are pivotal in the drive for male children; traditional gender practices around property inheritance, responsibilities of elder care, and continuity of the family name, which privilege men in most Asian societies, are largely followed even after generations of migration to the west. 9 19 31 Religion is implicated in son preference although the practices are neither uniform nor similar across communities. Sex selective abortions are more likely to be found among Sikh migrant communities, 12 16 whereas Abrahamic religions , given their strictures on abortion, are more likely to have larger families until they attain the family composition of their preference. 13 17 Although sex selective abortion is a predominant method, it is not the sole form of prenatal son preference. Other methods include larger family sizes and the application of the 'male preference stopping rule' or the practice of stopping pregnancies when the desired male child/children numbers have been achieved 21 and selective in-vitro fertilisation practices. 32 There is a glaring lack of comparative analyses of SRBs in Aotearoa/New Zealand despite the significant increase in the Asian migrant populations in the last 30 years. New Zealand is a demographically multiethnic and 'superdiverse' society although its politically foundations are rooted in biculturalism and Te Tiriti o Waitangi signed between the Crown and Māori in 1840. According to the 2018 Census, Asian ethnicities make up around 15.1% of the total New Zealand population, a growth of approximately 10% since the 1990s. 33 34 Asians are currently the largest ethnic minority population and comprise citizens, permanent residents, work-visa holders, refugees and international students. Scholarship on Asian women's lives highlights continuities in traditional gender norms. 35 In comparison to other ethnic groups, rates of induced abortion are higher among Asian women. 36 37 In 2019, amendments to the Abortion Legislation Bill decriminalised abortion relaxing the hitherto stringent conditions for termination although Section 20F specifically notes that, '[t]his parliament opposes the performance of abortions being sought solely because of a preference for the fetus to be a of a particular sex', requiring a review to assess evidence of any abuse of the legislation within 5 years. 38 This paper, an analysis of historical and current sex ratios within and among Asian migrant populations, aims to determine the influence, if any, of son favouring bias on SRBs, or in other forms as part of prenatal fertility and family size decision-making. --- METHODS --- Data sources The Statistics New Zealand Integrated Data Infrastructure is a secure database containing deidentified microdata on people and households, collected by government agencies and non-government organisations. Our study used the Census, Ministry of Health, Department of Internal Affairs and Ministry of Business, Innovation and Employment datasets. --- Historical census series: 1976-2013 Our first analysis aimed to describe the changes in ethnic sex ratios over time. This analysis used Census datasets from 1976, 1981, 1986, 1991, 1996, 2001, 2006 and 2013. At the time of analysis, Census 2018 was not yet available in the IDI. From these datasets, we extracted age, sex and ethnicity data. Age was defined as age on census day. Sex was recorded as a binary variable, man or woman. Ethnicity was coded according to the ethnic groups listed in the Health Information Standards Organisation's Ethnicity Data Protocols, the standard for collection, recording and outputting ethnicity in the health and disability sector in New Zealand. Of note are ethnicities that were coded for, historically, that are no longer coded for in more recent census. These included classifications such as West Indies, Negro, Micronesia and conglomerate ethnicities such as 'Ceylonese/ Singhalese/Sri Lankan', 'Syrian/Lebanese/Arab'. In the cases where there is no direct match for a historical code in the Protocols, the contemporary name of the region or name of the broad geographic area was imputed as the ethnicity code. Imputation was mostly required for Middle-Eastern, Latin American and African and Pacific ethnicities. Given that the historical census series is a measure of the total New Zealand population, we summarised these data as the number of males born per 100 women. To represent the range of sex ratios that are likely to arise based on random variation , we calculated 95% CIs based on binomial probability. --- Birth register cohort Our birth register cohort comprised all live births between 2003 and 2018, inclusive, recorded in the DIA Births dataset. The study period was determined by the data availability of the Ministry of Health Maternity dataset. Ethnicity for the mother was obtained by linking an encrypted unique identifier associated with the birth and data from the Ministry of Health Demographics dataset. We chose a small number of covariates based on international literature. These were: visa group, which is a proxy for generational status in New Zealand. It is assumed second and third generation migrants are likely to have resident status and those holding visas will be first generation/recent migrants. Visa status was derived from the MBIE Decisions data table ; parity, which was derived from the births dataset and includes a record of Open access the number of siblings a newborn has; maternal age, which was estimated based on the birth year of the mother and birth year of the newborn; and area level deprivation. Area level deprivation was derived using the latest notified address and linking this to an Index of Multiple Deprivation decile. 39 The IMD measures deprivation of a data zone in terms of employment, income, crime, housing health, education and access to essential services, and ranking each area in New Zealand from most to least deprived. The data zone is a geographic scale that is designed to capture whole neighbourhoods while preserving socioeconomic homogeneity, making it a suitable proxy for socioeconomic position. Descriptive statistics for the New Zealand birth cohort 2003-2018 are presented as counts and percentages or means with SD and medians with IQR. Logistic regression was used to compare odds of a male child relative to a female child and adjust for other potentially influential factors. The European/Pākeha population was used as a reference population as per other studies on SRB. Finally, the probability of having a third child within the 2003-2018 birth cohort was examined in the context of the sex of the first two children among families of 3 or more children using logistic regression. Analysis was conducted for the four permutations of female and male siblings . --- Patient and public involvement No patient involved. --- RESULTS --- Historical census series 1976-2013 The New Zealand population aged 0-5 years ranges from just over 300 000 up to 360 000 throughout the 8 censuses between 1976 and 2013 inclusive. The Indian, Chinese, Asian and MELAA population has grown steadily as a proportion of the total over these years. The Chinese population with about 1800 0-5 year olds in 1976 growing to 13 000 in 2013. The Indian population from 1400 0-5 year olds in 1976 to 13 400 in 2013. The number of men for every 100 women presented by ethnicity is shown in figure 1. The decreasing magnitude of the 95% CIs and declining variability of the male to female ratio for the Asian and MELAA populations reflects this steady growth. There is no evidence of excess female mortality, among the migrant population in New Zealand, in this robust data series. --- Birth register cohort 2003-2018 There were around 877 101 births in New Zealand from 2003 to 2018 . Around 5% of these births were to mothers with a recent migrant and non-Pacific background. Parity is low with a median parity for the cohort of 1 and mean of 1.7 with relatively little variation among different ethnicities. There were no significant associations between ethnicity and odds of a male child with or without adjustment for visa group, parity, maternal age and deprivation. When examining family sex composition and family size , there was a clear preference for mixed sex regardless of ethnicity with odds of having a third child significantly lower versus two males MF versus MM, OR 0.82 and FM versus MM, OR 0.83 . However, there was a significant interaction between family sex composition and ethnicity and therefore different ethnic groups were modelled separately. Indian families, OR 1.51 and Chinese families OR 1.22 were significantly more likely to have a third child if the first two children were female compared with both male . European families were marginally less likely to have a third child if the first two children were male, OR 0.95 and the significant preference for mixed sex compared with both males was maintained throughout the separate ethnic models. Finally, the sex ratio for parity 3 children of families with two female first born was not significantly skewed for Indian (459 males/408 females, 52.9% (Wilson 95% CI Figure 1 Number of males for every 100 women aged 0-5 years from NZ Census series from 1976 to 2013 with 95% CIs, expected sex ratios at 5 years of age remain around 105 men to 100 women for most countries and particularly developed countries due to reductions in infant mortality. 45 --- DISCUSSION Our study results point to marked differences in M:F ratios among New Zealand's Asian minority and migrant population groups relative to comparator countries like Canada, the USA and Australia. Varied measures of SRBs and child sex ratios in these countries consistently demonstrated a male surplus-female deficit in Indian and Chinese populations, resulting in significant missing women in the overall population. In contrast, SRB for New Zealand Indian and Chinese ethnic populations do not show similar male-favouring bias for the period under study, in the multiple data sets, that is, Census and Birth Data. M:F deviations from the SRB norm of 105:100 norm were not outside of what would be expected due to random variation, or more recent estimations specifically for New Zealand at 1.058. 40 . Most significantly, there was also no evidence of sex ratio skew at birth order >2/ higher parities as has been seen in several other contexts. Deprivation and migration did not have a significant influence on the sex ratio either. In all these measures, Asian sex ratios are unremarkable and no different to the estimates for the reference group, that is, European/Pākeha. The lack of evidence for excess female mortality allows us to conclude that if it occurs at all, prenatal sex-selective Open access termination is not a widespread practice among the Asian populations of New Zealand. The absence of bias in sex ratios may be attributed to a range of factors. Prominently, New Zealand, unlike neighbouring Australia, the USA or Canada, had stringent restrictions around pregnancy terminations, which may have been a deterrent to the procurement of sex-selective abortions. Also unlike in Canada and the USA, where the sex of the baby may be revealed during regular ultrasound screening during pregnancy, there are stricter stipulations around such disclosures in New Zealand. Although every interaction between practitioners and their clients cannot be monitored, the evidence does not seem to suggest significant breaches in disclosures. Furthermore, ultrasounds confirming the sex of the baby are usually conducted around 18 weeks but by law abortions beyond 20 weeks are illegal. Arguably legislative restrictions are likely to affect domestic access to termination; however, there is also no evidence to suggest that Asian women travelled overseas to procure sex-selective abortions. Prohibitive costs associated with travel from New Zealand may constrain the return to home countries for this purpose but, conversely, there is no evidence to suggest that higher-income migrants have used this pathway. More positively, acculturation of Asian migrant populations and alignment with the more liberal gender norms in the host country might have an influence on the reshaping of attitudes to the girl child and would be in Open access line with results noted elsewhere. 13 16 Broader social positioning as minority populations might have also impacted on the perception and valuation of children as assets and social capital, regardless of sex. Research has highlighted myriad ways in which 1.5 and second generation migrant children support families through bringing in secondary income, supporting parents through language and cultural interpretation, 41 42 and deepening roots in the community. 41 For many first-generation migrants, children also facilitate in the process of legalising their residency status. Until 2006, any child-regardless of sex-born in New Zealand was automatically accorded citizenship. Compared with other minority groups in New Zealand, such as Māori and Pacific Island communities, or overseas such as Punjabi Sikh community in Canada, the majority of New Zealand's Asian and ethnic community are relatively recent arrivals and in varied stages of settlement. Under these circumstances, the benefits of children, both boys and girls, may outweigh any perceived cultural disadvantages. The lack of sex-selective abortion specifically does not dismiss son preference wholly. Indeed, what has emerged from the latter part of our analysis is that desire for male children continues to be germane among Indian and Chinese couples but is practiced as extended family sizes and planned sex composition. Male preference is manifest in 'soft' prenatal decision-making placing it in the ethical crossroads between gender bias and sex discrimination, on the one hand, or a matter of choice and family balancing, on the other. 16 32 43 Study limitations Methodologically, the strengths of the study include the use of whole of population data. However, these are administrative data not collected for the specific purposes of examining son preference and missing women. Therefore, there is likely to be imprecision in some of the measures, for example, visa group as a proxy for duration of family residency in New Zealand. However, the analyses also reveal significant gaps in prenatal sex-specific data. The analysis also used live birth sex ratios as the sole indicator of gender inequity in Asian and ethnic communities. Live birth sex ratios are open to critique and needs to be triangulated with other data sets to confirm existence of sex-selective fertility practices. 32 Further work could focus on intervals between births, sex of stillbirths, and further disaggregation by religion, income, and region. New Zealand also currently does not capture sex-specific abortion data, especially for abortions after 14 weeks when sex is known; this too would have been a source of additional data for triangulation. Similarly, there are also comparisons to be made against the inordinate perinatal stillbirths among the Indian population in New Zealand. 44 While as of now there is no suggestion that this gender-related, a sex-differentiated analysis would further enhance our understanding of the physiological and social implications of gender and culture during pregnancy. Finally, the data presented here are a snapshot of time whereas the migrant community in New Zealand is continually in transition. In the years to come, we are likely to have a growing proportion of migrant populations who have been born in New Zealand or have come to the country as children, and who make up the next generation of reproductive age women. There is also a rise in inter-racial/interethnic marriages, all of which are likely to inform a new generation of gender norms in many of these communities. --- CONCLUSION Our study, an exploration of M:F sex ratio imbalances in Asian immigrant populations in New Zealand, failed to reveal any skewness towards male children, or present any evidence of potential sex-selective abortion. The data, however, demonstrated extended family sizes particularly where the first two children were female. Based on these findings, the study draws the following conclusions. The first is to underscore the 'New Zealand anomaly' in Asian ethnicity sex ratios disputing sexselection and missing women as a widespread phenomenon in the country. Second, we conclude that gender bias and son preference are prevalent through softer cultural practices that are not captured in sex ratio imbalances. These study findings come at an apposite moment in New Zealand's legal history. The Abortion Legislation Bill in 2020 was enacted in the face of some concerns that any relaxations of the law may evoke sex selective abortion. Our results, which highlight an absence of this practice, provide a baseline for future reviews to assess the impact, if any, of the decriminalisation reforms. Instead, we argue that there should be scrutiny on the more subtle forms of son preference in the prenatal stages including family balancing or the less explored in-vitro fertilisation practices. Equally, son preference can also be demonstrated after birth, for example, in the opportunities available to girl children. These are areas for further research. Most importantly, these findings suggest that culture is continually being adapted to local conditions, be it fertility regulatory frameworks or migration histories. While education programmes in the community around daughter valuation may be beneficial, son preference and its effects are likely to be ameliorated through creating social systems-for example, in education, employment, community structures, leadership and politics-that are equitable to girls and women of colour and present the best possibilities for elimination of son preference as a cultural practice. Twitter Janine Paynter @JaninePaynter --- Data availability statement Data may be obtained from a third party and are not publicly available. Access to these data is available from the Statistics New Zealand Integrated Data Infrastructure and the longitudinal Census series for authentic research purposes. See https://www. stats. govt. nz/ integrated-data/ integrated-data- --- Open access conducted background literature review and assisted in final preparation of the manuscript. RS-K is the author of the Introduction, Discussion and Conclusion section, and JP with assistance from AC authored the Methodology and Results Section. All authors approved the final manuscript. RS-K and JP are the guarantors for the study. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. Disclaimer Statistics New Zealand Disclaimer: The results in this paper are not official statistics They have been created for research purposes from the Integrated Data Infrastructure , managed by Statistics New Zealand. The opinions, findings, recommendations, and conclusions expressed in this paper are those of the author, not Statistics NZ. Access to the anonymised data used in this study was provided by Statistics NZ under the security and confidentiality provisions of the Statistics Act 1975. Only people authorised by the Statistics Act 1975 are allowed to see data about a particular person, household, business, or organisation, and the results in this paper have been confidentialised to protect these groups from identification and to keep their data safe. Careful consideration has been given to the privacy, security, and confidentiality issues associated with using administrative and survey data in the IDI. Further detail can be found in the Privacy impact assessment for the Integrated Data Infrastructure available from www. stats. govt. nz. Competing interests None declared. Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. --- Patient consent for publication Not applicable. Ethics approval This study was approved by the University of Auckland ethics committee on 1 July 2019 and formal notification that it was out of scope for HDEC review, required for use of Ministry of Health data, on 24 May 2019. Provenance and peer review Not commissioned; externally peer reviewed.

Objectives Recent research from the UK, USA, Australia and Canada point to male-favouring sex ratios at birth (SRB) among their Asian minority populations, attributed to son preference and sex-selective abortion within these cultural groups. The present study conducts a similar investigation of SRBs among New Zealand's Asian minority and migrant populations, who comprise 15% of the population.The study focused on Asian populations of New Zealand and comparisons were made with NZ European, Māori, Pacific Island and Middle-Eastern, Latin American and African groups. Secondary data were obtained from the New Zealand historical census series between 1976 and 2013 and a retrospective birth cohort in New Zealand was created using the Stats NZ Integrated Data Infrastructure from 2003 to 2018. Primary and secondary outcome measures The primary outcome measure was SRBs and sex ratios between the ages 0 and 5 by ethnicity. A logistic regression was conducted and adjusted for selected variables of interest including visa group, parity, maternal age and deprivation. Finally, associations between family size, ethnicity and family sex composition were examined in a subset of this cohort (families with two or three children). Results There was no evidence of 'missing women' or gender bias as indicated by a deviation from the biological norm in New Zealand's Asian population. However, Indian and Chinese families were significantly more likely to have a third child if their first two children were female compared with two male children. Conclusion The analyses did not reveal male-favouring sex ratios and any conclusive evidence of sex-selective abortion among Indian and Chinese populations. Based on these data, we conclude that in comparison to other western countries, New Zealand's Asian migrant populations present as an anomaly. The larger family sizes for Indian and Chinese populations where the first two children were girls suggested potentially 'soft' practices of son preference.

Introduction Research on the role of Information and Communication Technologies in significant events is gaining traction [1]. Specifically, during a pandemic, ineffective communication between authorities and citizens can hinder the prompt adoption of preventive measures, impairing community recovery. How information is disseminated both between officials and citizens and among citizens requires further research [1], especially given the explosion of risk campaign messages, user-generated content, and complex dynamics of information dissemination are enabled by ICTs and social media [2,3]. Therefore, it is vital to understand how to encourage the sharing of official information among citizens, especially information generated by agencies that are staffed with science personnel to review health-related data in a scientific manner. ICTs have afforded convenient access to and dissemination of both official and unofficial information during the pandemic [4]. Information from an authoritative source , Centers for Disease Control and Prevention , National Institute of Health , and Food and Drug Administration ) is defined as official information in this study. A better understanding of official information sharing allows more timely and proactive responses for enhancing the efficiency of crisis management. On the other hand, unofficial information includes pandemic information from unofficial sources such as online forums, ordinary citizens, or alternative news. Although unofficial information is less likely to be verified than official information, these sources do not necessarily contain false information, and similarly, an official source does not necessarily mean that the information is accurate. Since only approximately 20% of U.S. news consumers seek the latest information about COVID-19 from CDC, NIH, and FDA [5], this study aimed to uncover the mechanisms behind the sharing of official versus unofficial information. Such an understanding would allow health authorities to facilitate the dissemination of authoritative information and enhance the effectiveness of their crisis communication. Given that health crisis communication is critical in times of a pandemic, this study explored the dynamics of audiences' predispositions, affective experiences, information seeking, and sharing of communication sources by adopting the O-S-O-R model [6]. The findings can shed light on how to improve communication practices, taking both top-down practices and peer communications into account simultaneously. While models predicting sharing of information do exist, this study considered the mechanisms underlying the sharing of unofficial information alongside the sharing of official information. More importantly, there has never been a time when health communication has been so politicized, such that Pew data have shown that Americans form divergent attitudes towards organizations such as the WHO and CDC, with Democrats trusting CDC and WHO and Republicans distrusting both organizations [7,8]. Our study examined how individuals' predispositions, types of COVID-19 information-seeking behaviors, and emotions toward the pandemic, taken together, affect their sharing behaviors of official and unofficial information about the pandemic, adopting the O-S-O-R model. The findings can offer insights that will help health organizations to meet the information needs of users by promoting official content on the risks of the pandemic. --- Crisis Information Sharing: Official vs. Unofficial Information Information sharing refers to the "practice of giving a defined set of people access to news content via social media platforms, as by posting or recommending it" [9] During the COVID-19 pandemic, information exchanges and expressions of opinion online have made social media important channels for information sharing. Often, governmental agencies provide important guidance for the public and news media regarding appropriate responses [10], and citizens typically rely on government instructions to decide on preventive behaviors [11]. In this pandemic, the global community has depended on the WHO to coordinate the response. In the United States, several federal health agencies representing the national government's official voice have been responsible for containing the pandemic, including the CDC, NIH, and FDA. Although Vraga and Bode [12] stated, "expert consensus can sometimes provide clearer boundaries between what is accurate and inaccurate" , "experts" and "evidence" can be subjective and difficult to define. Expert consensus is not always available, even among health authorities. During the initial stage of the COVID-19 pandemic, for example, health and science experts held different opinions about whether and when one should wear a facemask. Nevertheless, we can expect a wide consensus in the scientific community to underpin solutions provided by major health authorities in comparison to information not released by these identified, official sources. In fact, Helsloot and Grosnendaal [13] showed that information published by governments becomes buried under citizens' tweets during impactful emergencies. To enhance the effectiveness of authorities' communication with citizens and facilitate the spread of official, authoritative information, this study examined the antecedents and mechanisms accounting for citizens' engagement with official information sources. Such an understanding could even provide important implications for media literacy programs by providing media users with knowledge and efficacy to verify information based on source authoritativeness [14]. In fact, while substantial scholarship has been devoted to studying rumor transmission during disasters [15], less is known about the transmission of unofficial information and the withholding of official information in a health-related crisis context. --- The O-S-O-R Model The O-S-O-R model is an effective framework for explicating complex social cognition processes as it recognizes an individual's preexisting orientation and the mediating effects of media content exposure on personal-psychological factors [16]. The model has been widely employed in political communication contexts and recently applied to the health communication context [17,18]. Political communication literature guided by the model often theorizes the individual's political orientation as the O 1 variable and political participation or efficacy as the response variable. Previous health communication studies using the O-S-O-R model have examined audience perceptions and knowledge of a health issue as the preorientation variable, attitudes toward the issue or the health practitioners and agencies involved as the O 2 variable, and the desired behavioral change with regard to the issue as the response variable. Adopting the O-S-O-R model, this study examined how audiences' political orientation and trust in Trump affected their COVID-19 information seeking from news media or peers and their emotions toward the pandemic , which further led to their information-sharing behaviors . In other words, multiple predictors are expected to act on information-sharing behaviors simultaneously . --- Initial Orientations The first orientation, O , is often conceptualized as the preexisting structural and motivational characteristics that media users possess [19,20]. This study included two situational characteristics of Americans related to the COVID-19 pandemic: political orientation and trust in Donald Trump, the president of the United States from 2017 to 2021. Because demographics are likely to influence media behaviors [21], other orientation variables such as age, gender, education level, and income were included as control variables in our model. --- Political Orientation Political orientation is often treated as an antecedent in O-S-O-R models. Given the politicized nature of the COVID-19 pandemic [22], political orientation is likely to influence individuals' evaluations of the health crisis and, in turn, their informationseeking and -sharing behaviors. During the crisis, Republicans have been found to believe that the death toll was lower and social distancing was less necessary than Democrats [22]. Although most official agencies called for social distancing measures, such as cancelling public events and working from home [23], Republicans were found to believe less in these measures than Democrats [22]. Overall, Republicans were much more likely than Democrats to believe that the media were exaggerating the risks of COVID-19 [24]. Research has consistently confirmed that politically conservative Americans tend not to perceive the COVID-19 pandemic as a health priority, and thus are less motivated to seek information from liberal news media than conservative news media. This is because liberal news media typically portray the pandemic as a critical public health threat and suggest the importance of preventive actions. Since conservatives tend to believe that the media has exaggerated the risks of COVID-19 [25], they were anticipated to seek less information from attitude-conflicting sources and more information from alternative information sources, such as people from their surroundings . Hence, we hypothesized the following: H1: Political conservatism relates negatively to seeking information from liberal relative to conservative news media news media) and positively to seeking information from peers. --- Trust toward Donald Trump Trust in the sitting U.S. president, which is closely related to political orientation, can also play a role in information-seeking behaviors. Given that some conservative politicians downplayed the threat of COVID-19 [26], Americans who trusted Trump might not consider the disease a health priority and thus might turn to like-minded information to avoid exposure to information that does not align with their beliefs. In fact, the observed partisan differences in distancing were associated with subsequently higher COVID-19 infection and fatality growth rates in pro-Trump counties. Further, support for Trump was found to relate positively to individuals' beliefs on whether the COVID-19 threat was exaggerated [27]. Together with the fact that liberal news media are likely to disapprove of the Trump administration's handling of the COVID-19 outbreak, as well as the tendency for individuals to seek information consistent with their personal beliefs [28], those who trust Trump might turn from liberal news sources to alternative ones and believe in information that supports their personal political attitudes [29], and vice versa among those who distrust Trump. --- H2: Trust toward Donald Trump relates negatively to seeking information from liberal relative to conservative news media and positively to seeking information from peers. --- Stimuli Information seeking is a key correlator of information sharing [30]; information sources have been found to generate different levels of intention to seek and share information during disasters [31,32], and our research treated information seeking both from liberal news media and from peers as stimuli . According to Glenski et al. [33], an article headline alone may prompt an individual to share and endorse a news article. Just as news consumption can foster online discussions and encourage civic engagement [34], we expected liberal media consumption to enhance people's knowledge on COVID-19 risks as well as preventive measures [23], encouraging them to engage further in the issue, including sharing information. As shown in a Lee et al. [35] study, exposure to stomach cancer information from the media was found to impact the processing of that information. Hence, we expected issue-specific media use to play a role in the model predicting information-sharing behaviors. Given that liberal news media are likely to criticize the work done by Donald Trump, liberal media consumers were expected to be exposed to more official information published by CDC, NIH, and FDA [36]. Being exposed to more information on COVID-19, liberal media consumers might be more aware of the potential risks of COVID-19. Similarly, the more they talked to people around them about the pandemic, the more they would be informed about COVID-19 risks and influences. Hence, information seeking, in general, is likely to expose readers to more information, official or not, and such information-seeking behavior was expected to nurture the circulation of both official and unofficial information. This expectation resembles findings on the positive effect of individuals' search for information and their participatory behaviors [37], such as sharing information on their social networks [38]. In the current study, we specifically expected media use to impact information sharing via two affective routes, anxiety and anger. --- Affective Orientations and Responses The second orientation, O , refers to the cognitive or affective state of the audience which leads to the audience's responses to the media message, such as sharing the message [6]. The risk and crisis communication literature has identified two affective factors in the health-related crises that were examined in this study as the O 2 variables: anxiety and anger. They can determine the type of information shared. Even though anxiety and anger are both regarded as negative emotions, they have been found to drive different attitudes and behaviors [39]. Research has shown anger to be a form of aversion, closely related to emotions such as disgust and contempt [40]. Anxiety, on the other hand, has been shown to be more closely related to emotions such as depression [40]. According to Marcus et al. [39], anxiety is likely to be driven by one's surveillance system when threatening circumstances are identified, while anger "arises when familiar threats impede our way" . The two are therefore distinct emotions and were anticipated to invoke different information-sharing behaviors. The rationale for including these two specific emotions in our model is that anxiety and anger are commonly found to associate with information-seeking intentions and behaviors [41,42]. Moreover, since the two emotions have been shown to invoke divergent outcomes, with anxiety motivating avoidance and anger motivating aggression [43], it was likely that people would develop diverse information sharing behaviors during COVID-19. The majority of citizens did not have first-hand experiences with COVID-19 but did consume pandemic information through the consumption of information from news media and peers. While liberal news media outlets tend to cover COVID-19 as a serious public health threat, seeking information from peers is likely to result in similar affective reactions, as learning more about a pandemic at face value is expected to increase related risk perceptions. --- Anxiety Anxiety is often associated with uncertainty and a lack of personal control [44], which is to be expected during a pandemic. According to Valentino et al. [45], findings on the motivational effects of anxiety are mixed. In political communication literature, Berger [46] found a positive relationship between anxiety and information sharing, but Lerner and Keltner [47] found anxiety to be associated with withdrawal, which discouraged engagement. In risk-communication literature, Jin et al. [48] found anxiety to be associated positively with disaster information-sharing behaviors, with individuals seeking not only to express their emotions but also to cope with their stress by communicating with people they know intimately. Indeed, people often experience anxiety when they face an imminent threat [49], leading them to adopt protective behaviors [50], and scholars have generally found a close relationship between risk perception and anxiety [51]. In the pandemic context, anxiety should drive sharing of both official and unofficial information, as both allow people to cope with their stress, connect with those who might provide beneficial information and social support, and determine useful protective behaviors for the potentially threatening circumstances. As anxiety should not prompt differences in information sharing, we assumed that anxiety is positively related to both official and unofficial information sharing. Further, people with more exposure to information tend to have more opportunities to share; hence, we hypothesized: H3: Information seeking through liberal relative to conservative news media and peers relates positively to anxiety. H4: Anxiety relates positively to sharing both official and unofficial information. --- Anger Anger is often experienced when there is a target [50] or an individual or organization to blame [49]. In the COVID-19 pandemic, for example, the public has blamed the government for doing too little to prevent the crisis and/or to mitigate its negative impact. Scholars have also found anger to be an approach emotion associated with mobilization and behavioral actions [50]. According to Hasell and Weeks [52], an easy way to channel anger is to share information; if people are angry with a target, they are likely to share negative information to discredit or punish that target. Health crisis literature shows that an individual's anger at a health provider can drive his or her acceptance and dissemination of unofficial information [53]. For example, Han et al. [54] found that in South Korea, the public's anger was positively associated with their acceptance of COVID-19 rumors, while in China, anger-inducing rumors about the pandemic triggered people's dissemination of the rumors on social media [55]. In this study, we expected that if people were angry with the government's response to the pandemic, they would share unofficial but not official information about the pandemic with others to challenge the government's action. --- Mediating Mechanisms The study of mediation is important to the assessment of direct effects with considerations of contextual and situational factors, as the mere inclusion of direct effects is often insufficient to fully explain human behaviors [16,56]. Hence, the proposed model recognizes the mediating effects of an individual's pre-existing orientation on the consequences of media content exposure [16]. In particular, anxiety and anger can differentially mediate the effects of people's predispositions on their distinct information-sharing behaviors. For example, due to the attributional nature of anger [57], those of certain political orientations can share unofficial information out of anger . Moreover, those of certain political orientations may be driven by anxiety to share official information [58]. Indeed, liberal media users were found to relate positively to preventive behavior engagement [59]. Given that no prior studies have used the O-S-O-R model to distinguish between official and unofficial information sharing, a research question was posed to address the mediations among the different components for official versus unofficial information sharing. --- RQ1: To what extent do affective orientations and stimuli mediate the relationship between audiences' predispositions and responses ? --- Materials and Methods This research utilized survey data collected from 21 to 26 April 2020, within the three months after the WHO and the Trump administration declared COVID-19 a public health emergency [60]. During this time, people were recommended to wear face masks and perform social distancing, and states and territories were starting to issue mandatory stay-at-home orders [60]. An online panel of respondents were recruited through Qualtrics . To ensure a sample that closely resembled the demographic distribution reported by the U.S. Census Bureau, quota sampling in terms of gender, age, and education was performed. Given that the recruitment of female participants aged 65 or above and holding less than a high school degree was difficult, the quotas for participants holding less than a high school degree were reduced to 6% from 13% and distributed equally across the other education groups. A sample of 856 was recruited. The sample was comparable with the U.S. national population in terms of gender , age , and education . The average household income reported by the sample was between $50,001 and $60,000. --- Measurement Most items used in this research were adopted from previous studies on crisis and risk communication [48,61,62]. Items for sharing of official and unofficial information were constructed based on the presence of various sources in the COVID-19 pandemic. --- Political Conservatism On a seven-point scale, respondents were asked whether they considered themselves to be liberal , conservative or somewhere in between . --- Trust toward Donald Trump This was measured by the question, "How much do you trust President Donald Trump?" on a seven-point scale from 1 to 7 . --- Information Seeking from Liberal Relative to Conservative News Media Information seeking from liberal news media was measured by asking respondents, "How often have you looked for information regarding COVID-19 from the following sources? New York Times, ABC News, CBS news, and NBC news." The scores of the four items were averaged to measure liberal news media information seeking . Information seeking from conservative news media was measured by asking respondents, "How often have you looked for information regarding COVID-19 from Fox News?" . Dividing the liberal media consumption score by the conservative media consumption score, we derived the relative score . --- Peers' Information Seeking Information seeking from peers was measured by asking respondents, "How often do you seek information regarding COVID-19 from family, friends, coworkers?" The mean of these items was calculated to represent peers' information seeking . --- Anxiety To measure anxiety, respondents were asked to what extent COVID-19 made them feel the following emotions: anxious, worried, and concerned from 1 to 7 . The average score of the three items was calculated . --- Anger To measure anger, respondents were asked to indicate the extent to which they disagreed or agreed with the following statements with respect to their emotions toward the United States governmental emergency responses in the COVID-19 pandemic: "I feel angry", "I feel irritated", and "I feel annoyed". The average of the three items was obtained to form the anger measure . --- Sharing of Official Information Sharing of official sources was measured on a seven-point scale ranging from 1 "never" to 7 "frequently" by asking respondents how often they had shared information regarding COVID-19 from the following sources: the Centers for Disease Control and Prevention , the National Institutes of Health , and the Food and Drug Administration . The average of all four items was calculated . --- Sharing of Unofficial Information Sharing of unofficial sources was measured on a seven-point scale ranging from 1 "never" to 7 "frequently". Respondents indicated how often they had shared COVID-19 information to other people that was attributed to an unfamiliar source, not attributed to a certain source, not attributed to a credible source, not issued by an official source, and unofficial, from someone on the Internet. The average of all five items was calculated . 2.1.9. Covariates Demographic covariates included age, gender, education level, race/ethnicity , and income level. Additional covariates included issue involvement, i.e., the extent to which respondents perceived the pandemic as important and relevant ; and social media information seeking, i.e., the tendency that respondents sought COVID-19 information from social media platforms including Facebook, Instagram, Twitter, TikTok, and Snapchat . --- Analytical Schemes We tested our path model using the R Lavaan Package [63]. A construct with multiple indicators was identified by a composite based on the average scores of all items. To adjust for the measurement error, the error variance of the multi-item construct was fixed at times the indicator's variance [64]. Note that the covariates were regressed on all endogenous variables. In the path model , the exogenous variables were political conservatism and trust toward Donald Trump. The first layer of endogenous variables included information seeking from liberal relative to conservative news media and peers' information seeking. The second layer of endogenous variables included perceived anxiety and anger. The last layer of endogenous variables included the sharing of official information and the sharing of unofficial information. Maximum likelihood estimation was used. The path model was evaluated based on the standard cut-off values for the model-data fit indices [65]. Furthermore, the significance of indirect effects proposed in RQ1 was tested through bootstrapping . Unstandardized coefficients and their standard errors are reported in the subsequent section. Chi-square = 31.91, CFI = 0.99, RMSEA = 0.07, 90% CI RMSEA = [0.04, 0.09], SRRM = 0.017. Covariates, including gender, age, education level, and income level, race/ethnicity, issue involvement, social media information seeking were regressed on all exogenous variables. --- Results --- Initial Orientations The overall model-data fit was good [65], = 31.91, SRMR = 0.017, RMSEA = 0.07, 90% CI RMSEA = [0.04, 0.09], and CFI = 0.99 . The variance explained by the predictors was 0.42 and 0.40 for the sharing of official versus unofficial information, respectively. While H1 assumes that political conservatism relates negatively to seeking information from the liberal versus conservative media and positively to seeking information from peers, our results showed that political conservatism was negatively associated with information seeking from the liberal versus conservative media , supporting H1a; however, political conservatism was not associated with peers' information seeking, disconfirming H1b. H2 hypothesizes that trust toward Donald Trump relates negatively to seeking information from the liberal versus conservative media and positively to seeking information from peers. There was a positive association between trust toward Donald Trump and peers' information seeking and a negative association between trust toward Donald Trump and seeking information from the liberal versus conservative news media , supporting both H2a and H2b. --- Stimuli , Affective Orientations , and Response With respect to whether information seeking through liberal versus conservative media and peers relates positively to anxiety and whether anxiety relates positively to sharing both official and unofficial information , our results showed support for both H3a and H3b. While information seeking through liberal versus conservative media was positively associated with anxiety, information seeking from peers also related positively to anxiety . Furthermore, anxiety was found to associate with sharing of official information . Hence, H4a but not H4b was supported. Subsequently, the relationships between liberal versus conservative media and peers and anger , and the relationships between anger and sharing official and unofficial information were tested. While information seeking from both liberal versus conservative media and peers did not relate to anger, political conservatism and trust in Donald Trump were directly associated with anger without going through the two information-seeking variables. Further, our results also showed that anger was associated with sharing unofficial information , supporting H6b but not H6a. --- Mediations RQ1 asks the extent to which information-seeking behaviors affect sharing of official and unofficial information through various mediating mechanisms. Regarding sharing of unofficial information, our results supported the mediating role of anger in the effects of audiences' predispositions on the sharing of unofficial information. Both political conservatism and trust toward Donald Trump affected the sharing of unofficial information through anger: indirect effect = -0.02, SE = 0.005, p < 0.01 for political conservatism and indirect effect = -0.03, SE = 0.007, p < 0.001 for trust toward Donald Trump. For the sharing of official information, our results supported the mediating role of anxiety in the effects of information seeking on the sharing of official information. Namely, information seeking from both liberal versus conservative media and peers led to anxiety, which in turn affected the sharing of official information. Last, the full sequence of O-S-O-R was examined by testing the indirect route from political conservatism to liberal media information seeking, to anxiety, and ultimately to the sharing of official information. The result was significant . Namely, the more conservative a respondent was, the less information the person sought from liberal versus conservative media, the lower level of anxiety the person experienced, and the less they shared official information. --- Discussion This study examined the mechanisms driving the sharing of both official and unofficial information in a pandemic. The consideration of both sharing of official and unofficial information in a single model distinguishes this research from other models predicting information sharing. By illuminating the different affective mechanisms as well as identifying the different initial orientations and stimuli that drive the sharing of two types of information via the O-S-O-R model [6], our results contribute to the understanding of how scholars and practitioners can promote official content to enhance the effectiveness of communication in times of a politicized health crisis. --- Affective Mechanisms Recognizing the mediating effects of affective orientations on the consequences of information sharing from both official and unofficial sources, anxiety and anger were found to trigger distinct information-sharing behaviors. This finding supports the extant theorizing on emotions and information behaviors in crises or issues involving risk implications [51]. While both affective orientations are negative in nature, anxiety affected the sharing of official information, while anger toward the U.S. governmental response affected the sharing of unofficial information. These results support the literature regarding how the two distinct emotions can drive different behavioral outcomes [39]. While the literature on the effects of anxiety on information sharing has mixed results [45], our findings add support to the positive relationship between anxiety and official information sharing [46] in a pandemic, in which anxiety does not necessarily mean a full withdrawal of engagement. The lack of full information and personal control over the pandemic might have led people to cope with their stress by relying on professional health authorities. As suggested by Marcus et al. [39], threatening circumstances are likely to drive anxiety. When faced with a threat such as new to humankind such as COVID-19, individuals who experience a lost in personal control [44] share information as a protective behavior [50] when they are motivated to solve the problem [67]. Our findings add support to the positive relationship between anxiety and information sharing [46,48]. More importantly, this positive relationship was found to rely only on official information rather than unofficial information as a strategy. As people cope with anxiety about a new pandemic that could be lethal, they tend to communicate with others they know via sharing information from authoritative, official sources even though the pandemic situation is highly politicized. These sources allow the intake of information from the major health authorities responsible for containing the pandemic in the United States. Moreover, when there is a target to blame [49,50], in this case the government, anger can trigger the sharing of unofficial information for two reasons. First, such a sharing behavior can represent a means to express anger [52]. Second, from a functional perspective, such a behavior can supply any information not already in the authorities' discourse for the angry public to resolve the undesired situation [54]. Behaviors to advocate for information missing from the mainstream discourse make emotion an "approach emotion", which is associated with mobilizational actions [50]. When one is angry at a crisis manager , such anger can drive both the acceptance as well as dissemination of unofficial information [53]. Our findings echo this line of health crisis literature [52], such that people who are angry with the government's response to the pandemic tended to share negative, often unofficial, information to discredit and challenge the government's action. In short, while anxiety drives the sharing of more credible, official information, encouraging more top-down flow of crisis information, anger drives the sharing of unofficial information, encouraging more information among citizens, or even adding new crisis discourses to the public domain. --- Initial Orientations As anger is likely to be triggered when there is someone or something to blame in a crisis, interventions to reduce the sharing of unofficial information could be difficult. This is particularly true when people are already exposed to information that conflicts with their predispositions, such as political orientation and attitudes toward the target of blame. Overall, audiences' predispositions play a significant role in people's informationseeking behaviors and, in turn, their perceptions of the pandemic and information-sharing behaviors related to COVID-19. Given that some conservative politicians downplayed the threat of COVID-19 [26], Americans who trusted Trump might not consider the disease a health priority and thus might turn less to liberal information to avoid exposure to information that does not align with their beliefs. These findings not only align with studies which found exposure to traditional news media to be associated with fewer misperceptions regarding COVID-19 [68], but also add support to the selective exposure hypothesis [28]. --- Stimuli While most studies have found information seeking to be a key predictor of information sharing [30,69], our findings found a link between the use of liberal relative to conservative news media and official information sharing via anxiety. The consumption of liberal news was likely to create exposure to more official information published by CDC, NIH, and FDA, as well as higher awareness of the potential risks of COVID-19. According to Duffy et al. [4], people who perceive a higher level of risk are expected to share more information as they will likely perceive information, official or not, to be useful and relevant to their friends and family members. As liberal media consumption was found to be associated with higher levels of anxiety [59], the increase in related risk perceptions could trigger anxiety among readers. By sharing information to others, one can protect people in one's social networks and combat the pandemic by encouraging others to cooperate with the official response measures. It is therefore not surprising to find that liberal news media consumption triggered information sharing via anxiety. Our findings contribute to the literature on the O-S-O-R model and health crisis communication. --- Implications In general, the O-S-O-R model suggests that information sharing should be studied without lumping different kinds of information into one sharing variable. By integrating the risk communication and political communication literature, our findings provide two theoretical implications. First, when facing a pandemic that is severe and politicized, people engage in different affective mechanisms to share distinct types of information. Second, given that the literature on online information engagement has so far probed the concept of online information sharing as a unified construct, examining the sharing of official versus unofficial information as distinct constructs is necessary and valuable. Our study also revealed several practical insights for pandemic communication management. First, to facilitate the positive implications of official information on social media, communication professionals should educate the public about how to form accurate risk perceptions in a complex and saturated information environment. Communication professionals can also provide key opinion leaders with credible information and encourage them to share such beneficial information with their own networks [70]. Second, an appropriate amount of anxiety may benefit public health by fighting COVID-19 on the societal level due to people's increased sharing of official health information, even though it is unwise to purposefully elicit anxiety. Communication managers can inform individuals on ongoing pandemic management via accurate and transparent information to maintain a certain level of anxiety, which in turn drives official information sharing. Such an insight could be useful in understanding the seeking and sharing of pandemic information during the pandemic, including the vaccine information that came later in 2021. Finally, crisis specialists and government communication managers should be aware of and act to minimize the anger resulting from poor or indifferent crisis actions and communication to be experienced by audiences in order to discourage unofficial information sharing, especially malicious rumor mongering. --- Limitations Several limitations of this study need to be addressed. First, there might be other channels of information seeking that could be included in the current model, such as health websites. Although information seeking from social media was controlled for and liberal news media and peers did cover the majority of channels for seeking information, future studies should take more information-seeking platforms into account. Second, the data were collected in the United States during the COVID-19 pandemic. The extent to which the findings from this specific context can apply to other scenarios, crises, and countries demands further investigation. Third, while a model was generated with variables placed sequentially in line with extant theorizing, this study utilized a cross-sectional design, and causation cannot be guaranteed. Studies with panel data should be implemented to complement the current findings. Finally, respondents may not have accurately reported their frequency of official and unofficial information sharing. In fact, they might not be willing to report sharing unofficial due to social desirability. Future research should gather actual behavioral data to complement current findings. --- Conclusions Despite the limitations, this study examined the different mechanisms accounting for the sharing of unofficial and official information during a pandemic. Borrowing from the O-S-O-R model, our findings highlight not only the different roles of anger and anxiety but also the importance of audiences' predispositions in driving information sharing in cyberspace. In particular, anger toward the U.S. government's outbreak response was found to associate with unofficial information sharing, while anxiety was associated with official information sharing. Our finding on the relationship between anger and unofficial information sharing potentially contributes to the research on misinformation consumption and sharing by highlighting the roles of particular discrete emotions in driving unofficial information sharing. The findings also reveal the need to investigate different kinds of information sharing given the mix of government-public communication and citizen-tocitizen communication in the current media ecosystem. --- Author Contributions: Conceptualization, S.J.T., X.Z. and Y.-R.R.C.; methodology, S.J.T., X.Z.; formal analysis, X.Z.; data curation, S.J.T. and X.Z.; writing-original draft preparation, S.J.T., X.Z. and Y.-R.R.C.; writing-review and editing, S.J.T., X.Z. and Y.-R.R.C.; visualization, X.Z. supervision, S.J.T.; project administration, S.J.T. and X.Z. All authors have read and agreed to the published version of the manuscript. ---

The COVID-19 disease outbreak has seen mixed information flows comprising top-down communication from health authorities to the public and citizen-to-citizen communication. This study aimed to identify mechanisms underlying the sharing of official versus unofficial information during the outbreak. Survey findings based on a nationally representative U.S. sample (N = 856) showed that individuals' predispositions affected their information consumption and affective experiences, leading to distinct types of information-sharing behaviors. While anger toward the U.S. government's outbreak response was directly associated with unofficial information sharing, anxiety was directly associated with official information sharing. These findings enhance our understanding of the propagation of different kinds of pandemic information and provide implications for public education on information verification based on source authoritativeness.

Human mental life is profoundly situated. The human mind, for better or worse, has always been inextricably intertwined with the plasticity of our changing sociomaterial environments. This process of metaplasticity applies to human ontogeny as it does to human evolution. [1][2][3] Different academic disciplines, constrained by the theoretical assumptions, the time scales and the analytical units that they use, are able to observe, and would probably emphasise, different aspects of the relationship between mind and matter. Still, the underlying message is the same: our everyday material habits and forms of material engagement are inextricably linked with our cognitive and emotional livesthe major dynamic being the plasticity of the brain linked to the affordances of our body and the action possibilities offered by the things we make and use. We cannot understand human intelligence in isolation from its changing material and social surroundings. Material things matter and need to be taken seriously. 4,5 This realisation of the interrelatedness between human experience, affect and materiality has, potentially, major implications for how we approach and understand important issues of mental health. Yet, those implications are not well understood. The important question concerning the influence of materiality on mental health and human well-being has not been adequately addressed and requires work across disciplinary boundaries. Although the agency of things [6][7][8] and the vitality of matter 9 have been central themes of research in the humanities and the social sciences, research in neuropsychiatry has, by and large, side-tracked the thorny issue of material engagement. Given what we now know about the importance of materiality in human social and cognitive life, [10][11][12] this form of epistemic neglect is unacceptable and needs to be overcome. I do not wish, and there is no space in this brief editorial, to discuss the reasons for that blind spot. Instead, I want to suggest a way forward. Suffice it to say that the main reason psychiatry avoided investigating the full variation of the interactions between cognition and material culture, and their effects on mental health, is not because it sees those interactions as unimportant or uninteresting. Rather, it is because some traditional assumptions about the mind make it hard to study them. 13,14 Those assumptions are changing, replaced by the more productive idea that the boundaries of the mind 'should not be set a priori, but should be responsive to the nature of the phenomena under study'. 15 The boundaries of the human mind have always been a problem, but perhaps never so problematic as they are today. [16][17][18][19][20] There is unrealised potential here for creating new approaches to care and intervention that can complement existing practices and broaden, or in some cases challenge, prevailing assumptions in mental health. However, in order for this potential to be realised, a more nuanced, casespecific and evidence-based understanding of the material dimensions of human mental health is needed. 21 How can we measure the effects that different material ecologies, 12 practices and technologies have on human mental life? What constitutes evidence in that context? How can we best interpret existing evidence and use it to effect changes that could help improve our emotional and mental wellbeing? Largely, we still lack effective ways of answering those questions. Combined expertise and a range of specialisations are needed to develop both the theoretical and practical means that will allow us to make progress with these challenging issues. --- The material engagement approach The aim of this essay is to propose one possible path to this end, a path that could lead to better integration of in-depth anthropological and philosophical qualitative analysis with the quantitative methods and measures of psychiatry and neurophysiology. Specifically, I propose a material engagement approach combining methods from anthropology and embodied cognitive science that allows us to observe, to record and to compare the actual processes by which different materials and forms of material culture affect the people who experience and use them. [22][23][24] The material engagement approach provides a theoretical framework tailored to facilitate a heightened responsiveness to the details of situated action and the affordance of the materials involved . 25,26 The distinctive feature of the material engagement approach is the commitment to a view of thinking as a process that is distributed, enacted and situated, as well as assembled, from a variety of non-localisable mental resources spanning the boundaries of the individual brain and body. Material engagement theory as an explanatory path is based on three interrelated working hypotheses, which can be summarised as follows. [22][23][24] The hypothesis of extended mind, which explores the constitutive intertwining of cognition with material culture; the hypothesis of enactive signification, which explores the nature of the material sign not as a representational mechanism but as a semiotic conflation and cohabitation through matter that enacts and brings forth the world; and, finally, the hypothesis of material agency, which explores agency not as a human property but as the emergent product of situated activity . Taken together, these hypotheses enable us to restate the problem of cognitive boundaries, offering a crossdisciplinary relational basis for understanding the material ecology of mental health that integrates the 'materiality turn' in humanities 27 and social sciences with new embodied and enactive trends in philosophy and cognitive science. 28 Traditional anthropological participant-observation combined with visual methods and object interviews can be used for the study of everyday naturally occurring situated material activities and somatic modes of attention. 29,30 The creative methods of art and medical humanities are also needed to investigate how different forms of material engagement achieve their effects. [31][32][33] Some of the key issues include: the role of arts and creative industries in psychotherapy; 33,34 the development of evidence-based understandings of the therapeutic effects and qualities of different art materials and digital media; the role of museums in mental health [35][36][37][38] as a space for creative interventions; and the cross-cultural, 39 social and gender 40 dimensions of mental health. Such a cross-disciplinary anthropological understanding will allow us to harness the power of creative industries more effectively and to develop new tools that will not just meet current needs but also help to generate new ideas on how to design a landscape of creative affordances for future experimentation. There is currently a lot of emphasis on the ways technological innovation can be fostered to benefit older people. [41][42][43] Still, there has been very little work trying to understand how these new innovations fit or compare with more traditional material ecologies and creative habits. 44 Doing so could prove beneficial for all stakeholders involved. To illustrate those claims, I will focus on the example of dementia. --- The case of dementia Dementia is one of the biggest mental health challenges modern society faces. 45 The profound changes in people's lives brought about by dementia and by long-term caregiving are well known. This process of self-transformation is often described as a diminishment of or loss of self. 46,47 All people who live with dementia are affected. It is common to see dementia solely as the product of brain pathology and to leave people's bodily interactions with the material world unattended. But perhaps it is also important to pay closer attention and try to understand better how dementia extends beyond the brain of the individual. 20 The gradual loss of self is not solely the product of neurological factors. An impoverished material environment and the gradual loss of connection with significant others that provide a familiar, stable and comforting self-referential space also contribute to a diminishing capacity for memory, interaction and communication. One important question in this regard concerns how the evocative nature of material culture and its various cognitive, emotional and affective dimensions can be harnessed to guide the design of new care strategies and material settings that promote patients' agency, autonomy and ability in social interaction and communication. Are there ways to alter the person's material environment that can result in positive changes in embodied self-conceptualisation? 48,49 One way to answer that question could be by examining the changes in materiality experienced during periods of major transition, for instance, during the transition from home to 'care home'. This is a challenging period, often marked by uncertainty, anxiety, confusion and distress, which can be explained in terms of changing cognitive ecology and the stripping away of personal possessions that help to constitute one's personal memories and self-identity. 50 The philosopher Daniel Dennett, in his book Kinds of Minds , describes this cognitive impairment when elderly people are in hospital, bereft of the usual contextual cues and affordances of their familiar home environments. 51 More recently, Shaun Gallagher, 26 adopting an enactive perspective, proposes an affordance-based approach to therapy that places special focus on the changing physical, social and cultural environments. He argues that a variety of mental health problems and neurological and psychiatric disorders can be understood in terms of changes in a subject's 'affordance space'which is the lived space defined relative to the capacities of a specific agent. Significant objects and personal possessions that have been acquired and accumulated over the life course constitute material memories and material anchors of selfidentity. The powerful affective linkage of material possessions to one's identity explains the positive association between memorabilia and mood, as well as their observed influence on how elderly individuals achieve a sense of continuity. To lose these objects, often abruptly and unwillingly, owing to changes in residential circumstances, such as the transition from independent living to a nursing home, can be a 'major cause of loss and grieving for elders'. 52 Understanding the cognitive life of things and the importance of object memories can allow us to develop strategies of care for the things that can actively enhance or even create a sense of continuity, stability, and comfort, especially during those transitional stages when one moves from home to a long-term care facility. 53 Interestingly, in many care homes and hospices nowadays, residents are allowed to bring their own furniture and personal objects. One could track what they choose to take with them and how they actually use it. 54 Are there any preferred types of objects? Do the objects carried along retain their original meaning and usage or do they become something different? Do individuals always keep what they want to remember? Do they always remember what they keep? What is the evidence that personal objects can facilitate patients' transition to their new and often alienating surroundings ? Can we provide informed guidance on how objects can be used more efficiently and appropriately in personal memory practices? There have been a number of studies looking at how old people moving into long-term care engage with their new material surroundings, 55 but few of them have focused specifically on how dementia and other cognitive impairments may affect residents' ability to think and to feel with, through and about things. For instance, Buse and Twigg explored the potential of material objects to evoke narratives and elicit memories through the sensory and material dimensions of dress. The study focused on three main dimensions, i.e., kept clothes, discarded clothes and absent clothes, and used ethnographic methods to explore everyday experiences of clothing and their influence on maintenance of personal identity and biographical continuity. The study nicely illustrates the significance of dress, often unnoticed, in supporting personal identity in dementia care. 56,57 Combining traditional anthropological methods of participant observation with in-depth phenomenological analysis could help to answer questions of how different people experience this radical shift in their material ecology. Ethnographic attention to patients' changing bodily engagements, habits and material attachments during their transition to institutional care could provide a thorough first-hand examination of that. Cognitive 58 and visual ethnography 59 offer particularly apt methods to record and analyse everyday episodes of situated action and distributed intelligence, paying attention to the details of specific contexts. To investigate those processes, an analytical vocabulary cutting across brain-body-world divisions is required. Focusing primarily on non-verbal aspects of the dyadic or triadic interactions between patients, carers and things, one could analyse the sequential organisation of those interactions using a combination of pictures, written commentaries and graphic elements to illustrate features of material engagement such as movement, direction, eye-gaze, gesture, attention and action coordination. It is important to recognise that every person's experience of dementia is different. People with dementia have their own unique life histories and actively make meaning of their lives through their attachments to and interactions with specific places, objects and people. People should be encouraged to spend time with the objects of their attention, to observe closely and with all their senses, and to draw what they observed or find other ways to voice their opinion. 60,61 One could look specifically at self-narratives evoked and constructed through shared discussions of artefacts and objects. Also of interest is to explore, more specifically, the role of biographical objects as memory scaffolds and affective probes of autobiographical memory through their attachments to family histories and self-identity, offering perhaps a complementary pathway to non-verbal facilitation of communication, promotion of independence, and rediscovery of personal identity. Communication is one of the major problems in dementia care, and participatory ethnography offers an appropriate means of interaction. To study people with dementia, it is essential to give them a voice. Material engagement interventions facilitate and support complex interactions, offering alternative non-verbal avenues of expression and communication. People with dementia may understand, do and express with and through things far more than they can express with language. What would be important to clarify in this connection is the extent to which memory loss in dementia and other neurodegenerative diseases might be partially compensated by the use of selected biographically salient objects and material scaffolds for self-narrative and autobiographical memory. For instance, studies on bodily and object ownership have shown that owned objects can trigger a strong memory trace and are often treated as psychological extensions of self. 62 One could explore the effects that such strong associations between self and owned objects may have in the context of dementia. Another promising avenue of research could be to investigate the cognitive mechanisms activated when people with dementia think with and through things in an embodied, performative manner, and to compare them with the cognitive mechanisms activated when they think about things in the internalist representational manner. Understanding those processes is important, because whereas most forms of dementia seem to affect the latter capacities for thinking about the world , the former enactive capacities remain largely intact and could be re-used and developed to compensate for the lost ones. For instance, memory interventions in the form of memory books, photo albums and so-called memory boxes have repeatedly demonstrated their effectiveness as external memory aids and powerful communicative means. 63 The precise cognitive function of those external memory prompts and cues remains largely unknown, but it has been suggested that they work precisely because they do not require conscious cognitive effort in order to trigger retrieval of related semantic information from memory storage, effecting positive changes in the conversational behaviours of people with a wide range of severity of cognitive impairment. 64 There is little doubt that things and material objects evoke and trigger memories through associations with important life experiences and events, although exactly how that happens and how best those forms of material memory should be described is less understood. Beyond the widely recognised memories of objects, we should be considering also the possibility of object memories. Objects embody memories in their forms, life histories and uses that can provide powerful media for recollection, selfexpression and self-identification. The mundane things we own constitute memories of our past, material traces available to be re-enacted through our bodily engagements with the material world. In that sense, things provide a durable network of material signification, transposable dispositions and bodily habits which can be harnessed to enhance cognitive abilities or compensate for memory loss when biological memory is damaged. Like a blind's person stick, mundane things and media can help people to re-orient and re-structure their cognitive landscape in a variety of ways. 65 A material engagement approach could help us to explore whether there are ways we could structure and adapt the material ecology of any given environment in order to create scaffolds, material anchors 66 and action affordances 26,67 able to effect a more efficient utilisation of brain networks or stimulate the person's ability to recruit alternate brain networks that could enhance cognitive reserve. 68 A broader but related question in this regard could be whether different forms of material engagement can dynamically influence and mediate the relationship between brain damage and its clinical outcome. Here, one could refer to similar findings on collective remembering and cognitive reserve in old couples. 69 --- Conclusion In this essay, taking the example of dementia, I sketched the outline of a material engagement approach to mental health. I proposed that a collaboration of archaeology, anthropology, philosophy and neuropsychiatry could help us understand how people with dementia are actually engaging with their local material environments and the effects that the latter has on people's agency, cognitive abilities and selfhood across the full dementia care pathway. Such a collaboration could also form the basis of a longitudinal mixed method exploratory investigation of the possibilities for action and interaction that different objects and forms of material culture provide for people with dementia in common everyday situations and in different care settings . Ultimately, the objective is to broaden recognition of the material aspects of dementia. That is, understanding better the ways by which different forms of material engagement can help people with dementia to: remember and/or forget; enact, express their agency and preserve their selfidentity; and communicate and interact with others, especially family members and care providers. Hopefully, this will stimulate further cross-disciplinary research and broader engagement with material culture studies, which, in the context of current dementia and mental health research, are often seen as marginal and of limited value . Achieving these aims requires concerted action from a wide variety of organisations, collaboration and cooperation from the public. This exchange between practitioners, researchers and the public would generate further awareness, promoting new evidence-based cross-disciplinary research on the role of creative practices and material culture in mental health. It could also create transferable insights into the influence of creative practices in mental rehabilitation and in psychotherapy that could be used as alternative/additional solutions to traditional treatments of mental illness across the life course. The benefits of the proposed material engagement approach could be broad, offering important insights and practical recommendations for the well-being and long-term care of the aging population, with dementia or not. Understanding how different forms of material engagement affect people's cognitive abilities and selfhood across the full dementia care pathway would also lead to specific evidence-based recommendations for designing interventions, material scaffolds and assistive techniques that could help people maintain and improve their mental well-being. This could also inform and facilitate safe and responsible design of healing environments, offer suggestions on how their effects should be measured, and suggest policy changes to provide a mental healthcare environment fit for the future. Fostering deeper understanding of the ecology of mental health and developing simple interventions enabling people to improve their psychological well-being is increasingly important in today's environment, with mental health leaders seeking innovative and cost-effective ways to optimise the quality of care. Concerted action and cross-disciplinary research is required to ensure that all gaps in research are filled. Understanding how different forms of material engagement affect people's mental health presents us with such a challenge. --- About the author Lambros Malafouris is a senior research fellow at Keble College and the Institute of Archaeology, University of Oxford, UK.

Consensus is growing, in many areas of the humanities and social sciences, that aspects of the material world we live in have causal efficacy on our mindsthe major dynamic being the plasticity of the brain linked to the affordances of our bodily engagements with things. The implications of that on how we approach and understand important mental health issues have not been adequately addressed. This paper proposes a material engagement approach to the study of the processes by which different forms of materiality achieve their effects. Focusing on the example of dementia, I propose that a collaboration between archaeology, anthropology, philosophy and psychiatry could help us to fill this gap in our knowledge, allowing us to understand the exact effects of everyday objects, personal possessions and forms of material engagement on people with dementia.

This study attempts to explicate the "upstream" health protective effects of productive activities by studying a marker of biological risk in older adults. Specifically, we examine the link between productive activities and C-reactive protein , a measure of inflammation. CRP is a useful biomarker for this investigation because it typically rises in later life, and high levels of CRP are linked to cardiovascular diseases including heart attack . The main aims of the article are twofold: Identify the association between the number of roles for productive activities and inflammation; and systematically examine whether the relationship between productive activity and CRP varies by the frequency of engagement within each activity. --- Productive Activities and Health in Later Life Considerable evidence supports the claim that productive activity is beneficial to older adults' health. Research on successful aging pioneered by Rowe and Kahn points to the importance of engaging in productive activities to maintain health and well-being. Productive activity is a significant element of maintaining health and functioning, distinct from social relations or general social embeddedness. Role theory has often been used to explicate the health benefits of productive activities: volunteering or attending organized meetings provides individuals with a unique opportunity to hold a meaningful social role and remain similarly active in later life . Because most Americans subscribe to a busy ethic and strive to engage in meaningful roles , productive activities become a viable alternative for adults who typically experience multiple role transitions in later life. Previous research has used somewhat distinct definitions of productive activities to predict health outcomes. Volunteering and paid employment are most commonly recognized as productive activities . Studies show that volunteering is associated with well-being, whether the outcome considered is depressive symptoms , self-rated health , or hypertension . Employment has similar health-promoting effects in that employed men and women show virtually no decline in self-rated health and physical functioning; however, healthier people are also more likely to be employed . Caregiving is also widely recognized as an important form of productive activity . The contribution of middle-aged and older caregivers generates enormous benefits for the nation and for those served. Despite a large body of research documenting caregivers' depression and worse physical health , some studies demonstrate that caregiving behaviors may improve adults' well-being . Moen, Robison, and Dempster-McClain , however, found no direct effects of concurrent or past caregiving on health. Although studied less frequently, community engagement is another component of productive activity that may have health benefits. Community engagement includes a wide range of activities such as voting, attending community meetings, and contributing financially to or holding a leadership position in a community organization . Some of these activities require a considerable investment of resources, but Mendes De Leon and coworkers argue for the importance of tapping activities in which older adults are commonly involved. Meeting attendance is one of the more prevalent forms of community engagement for older adults, even though it requires a commitment of time and skills . Few studies, however, have examined whether meeting attendance has a salutary effect on health in later life. Health benefits from such roles in later life may accrue from the meaning attached to them, including status, influence, and emotional gratification ). When the types of productive activity are profiled across the life course, younger adults typically engage in roles that are related to their other statuses , whereas older adults are motivated by more intrinsic reasons such as their values in helping others or their interests in local community . Thus, the meanings and values associated with nonobligatory roles may be more salutary to health than is the case for obligatory roles. If the person feels generative by contributing to the wellbeing of others, the benefits are manifest in lower risk of impairment in activities of daily living and mortality . Several theoretical and empirical studies also suggest that productive activities may be especially beneficial for older adults. Role accumulation theory suggests that multiple role occupancy is relevant in older ages because reduction rather than accumulation of roles becomes more commonplace in later life . Empirical studies confirm this thesis. For instance, Van Willigen used longitudinal data to demonstrate that older volunteers are more likely than younger adults to experience additional psychological health benefits for each hour of participation. Also, Musick and Wilson reported that volunteering had a negative effect on depressive symptoms for older adults but not for those less than 65 years old. Because older adults may be more likely than younger persons to derive a psychological benefit from productive activity, one wonders whether the effect extends to physical health phenomena such as inflammatory processes. --- Advancing the Literature The literature on the health benefits of productive activities in older adults has yielded several notable discoveries. Perhaps most impressive is the breadth of outcomes considered. Previous research demonstrates that productive activities among older adults are associated with better self-rated health , fewer depressive symptoms , and greater life satisfaction . Recent discoveries also reveal that productive activities reduce the risk of hypertension in Americans and the risks of institutionalization and mortality in Finnish older adults . As Rowe and Kahn argue, productive activities have distinctive health benefits because they provide not only a sense of belonging to a group with whom one shares common interests and activities but also a unique chance to hold social roles that give purpose and meaning to life . Building on the contributions of previous research demonstrating the health benefits of productive activities, we articulate three ways to advance the current body of research. First, most studies of productive activities rely on self-reported measures of physical and mental health. Although self-reported measures are widely accepted as valid, investigations are warranted that examine the influence of productive activities on biological measures, which are harbingers of subsequent health problems. Indeed, the power of using biomarkers in health research is the early detection of bodily risks that might not yet register clinically. CRP, produced in the liver, is a useful biomarker because it indicates systemic levels of inflammation and is linked to risk of heart attack, stroke, and mortality . CRP is related to some modifiable cardiovascular risk factors such as smoking, physical activity, and obesity but is also inversely related to social integration: Maintaining ties to a larger community and having close and supportive relationships are related lowering chronic inflammation . Accordingly, CRP presents an important and novel way to explore links between productive activities and physical health in older adults. Productive activities may, in fact, lower older adults' inflammation through multiple pathways, particularly when they select the role. One potential pathway involves stimulating healthier responses to environmental challenges or stressors. Indeed, an observational study showed that social integration is related to quicker cardiovascular recovery from a day's exertions . Perhaps productive activities are similarly beneficial by improving bodily response to stressors, but we are unaware of any studies that examine the link between productive activity and CRP. Second, though the majority of existing research examines a single indicator of productive activity, some studies point to the importance of considering a more complex picture of individuals who remain active . Because concurrent engagement in multiple activities is common, failure to consider multiple roles may be misleading because the purported salubrious effects of productivity activities might be due to a related role that is not measured. Third, though multiple roles may be beneficial to health, the level of involvement in each role may be consequential to the purported health benefits . Scholarship on volunteering also points to how burnout or overcommitment can lead to termination of the volunteer role or unfavorable health outcomes . Such role overload can manifest itself on inflammation in two ways. First, high level of time commitment in productive activities may offset the purported benefits . Second, and more generally, it raises the question of whether the health benefits of productive activities accrue primarily by regular engagement in one role, modest engagement in multiple roles, or some combination of both. A longitudinal study of older adults included multiple roles, suggesting that both the number of roles and the amount spent engaging in each role are significant predictors of happiness . The current investigation examines both the number of productive activities performed by older adults and the frequency of participation within each type of activity in predicting levels of CRP. Drawn from the extant literature, we specify four hypotheses to guide the analysis. First, we hypothesize that older adults reporting more roles of productive activity will have lower levels of inflammation, measured by C-reactive protein. Second, we hypothesize that the frequency of participation within a role is negatively related to CRP. Testing the first two hypotheses leads us to specify a third that may be referred to as a role overload hypothesis: The health benefits of productive activities accrue until the person reaches very high levels of participation in multiple roles . Fourth, given that other studies reveal that older adults are more likely than younger adults to reap health benefits from productive activity, we hypothesize that the effect of productive activities on CRP is stronger for adults 70 or older than for those who have not yet reached 70 . --- Method --- Sample The study used data from the National Social Life, Health, and Aging Project , a nationally representative, population-based sampling during 2005-2006 of 3,005 adults aged 57-85 in the United States. In order to ensure representation of particular groups, the sample includes an oversampling by age, race, and gender. Additional information on the NSHAP sample is available in a report by O'Muircheartaigh, Eckman, and Smith . A random 83% sample of the 3,005 respondents was selected to undergo whole blood spot collection, and 1,939 individuals provided usable samples. The reasons for eliminating cases include refusal, equipment problem, and insufficient volume of blood. As is standard in the literature, 136 additional cases with very high CRP levels , which may be indicative of inflammatory response to acute conditions such as cold, flu, or injury, were dropped from the analyses . Regardless, sensitivity tests demonstrated no meaningful differences in the pattern of findings when including the highest CRP values. Finally, 13 cases with missing information for race or emphysema were excluded because the number of these cases is small. The final analytic sample size is 1,790. --- Measurement CRP Concentration.-During the interview, a blood sample was collected via a fingerstick and disposable lancet and applied to filter paper for transport and storage. The blood spot assays were conducted at the Laboratory for Human Biology Research at Northwestern University . As shown in Table 1, mean CRP was 2.08 . Given the highly skewed distribution of CRP, values were transformed for the statistical analyses . Productive Activities.-Four productive activities were examined: employment, volunteering, attending meetings, and caregiving. We created a binary variable for each role and summed them to create number of roles . The frequency of activity was assessed by items asking respondents how often they engage in each activity. For employment, respondents reported the number of hours they typically work during 1 week. We defined those who work 40 hr or more a week as full-time workers. Additional analyses examined alternative thresholds for full-time employment, but the conclusions were unchanged. For volunteering, respondents were asked how often they volunteered for religious, charitable, political, health-related, or other organizations in the past 12 months. Responses ranged from never to several times a week . Using the same response categories as for volunteering, respondents were also asked about their frequency of attending meetings of any organized groups. Consistent with prior studies, meeting attendance is used as a measure of community engagement-a component of productive activity . Finally, respondents were asked whether they are currently assisting an adult who needs help with day-to-day activities because of age or disability. If answered yes, they were then probed how many days per week they typically spend caring for this person. The original response ranged from 0 to 7 days. However, because only eight individuals answered that they spend 5 days on caregiving, we combined 5 and 6 days. The final response categories ranged from 0 to 6, consistent with volunteering and attending meetings. Additional Covariates.-In addition to productive activities, several covariates were included because of their association with chronic inflammation . Age is coded in years, and sex is dichotomized with 1 indicating female. Race was divided into a series of binary variables with non-Hispanic White serving as the reference group. Other race consisted of Hispanic Americans, Native Americans, Asian or Pacific Islander, and those identifying themselves as multiracial. Marital status was also dichotomized with 1 indicating married or cohabiting with a partner. To tap socioeconomic resources, four categories of educational attainment and a binary variable for low wealth were included in the analyses. For wealth, participants were asked to estimate their net worth including all of their investment, properties, and other financial assets minus debt. Then a binary variable was created differentiating respondents in the lowest 20% of the household net worth from those in the top 80%. For lifestyle variables known to be related to CRP, we use self-reported information. Participants who reported as currently smoking cigarettes, pipes, cigars, or chewing tobacco were classified as current tobacco users. Physical activity was measured with an item probing respondents' engagement in physical activities such as walking, dancing, or exercise . Obese individuals were identified with body mass index values ≥ 30 kg/m 2 . The analysis adjusts for physical and mental health conditions that may be related to CRP. For mental health, we also included depressive symp toms based on the 11-item Center for Epidemiologic Studies Depression Scale . Clinically relevant chronic health conditions include dummy variables for whether the respondent was ever diagnosed by a physician for emphysema, asthma, and diabetes. Given the documented inverse relationship between lipid-lowering treatment and CRP , lipid medication use was also included in the analyses, measured with a binary variable. Supplementary analyses considered additional covariates such as income, antihypertensive medicine use, and self-reported physical and mental health. These were omitted then from the final analyses, however, because they were nonsignificant in multivariate specifications. --- Analyses Analyses used ordinary least-squares regression to model the natural log of CRP. We specified separate models to examine number of productive activities and frequency within each activity, as well as their joint effects. In addition to an additive model with the independent variables, multiplicative models were estimated to test interactions between the number of roles and frequency of productive activity. We also tested the robustness of our results in two supplementary analyses: OLS regression predicting raw CRP values and ordinal logistic regression with ordinal CRP as others have done . Because CRP has clinical cutoffs that lend themselves to logistic regression , we used the three-category classification of CRP as the outcome variable in supplementary analyses. The results from the ordinal logistic regression are provided in the Supplementary Material, but we focus on the OLS results in the text. We handled missing values on variables by employing multiple imputation for 498 cases, 294 of which were missing information on volunteering variable. We imputed five values for each missing observation, averaged the coefficients and calculated new standard errors that account for the variation across multiple potential imputed values . We used all the information available to impute for missing data on independent variables but did not impute for the outcome measure . --- Results Table 1 displays the descriptive statistics for each variable, and the mean CRP level of 2.08 is regarded as average risk . Most NSHAP respondents engaged in two to three productive activities, with the largest proportion engaging in meeting attendance and the lowest proportion in caregiving . Volunteers comprised 62.5% of the sample; 31.9% of the sample was employed. Note that more than one third of the entire sample has a body mass index ≥ 30. Turning to Table 2, the Model 1 results reveal that each role of productive activity is associated with lower CRP even after adjusting for all control variables. Women and Black respondents generally had higher levels of CRP, and higher educational attainment was associated with lower CRP. Of the lifestyle factors, tobacco use and obesity were associated with higher CRP, and physical activity was associated with lower CRP. Given the gender differences, we estimated supplementary models stratified by gender but the substantive results were similar. In order to further examine possible confounding effects of lipid medication on productive activity and CRP, we conducted supplementary analyses with and without the lipid medication variable, and the conclusions were unchanged. In Model 2, we examined the frequency of engaging in each productive activity while adjusting for the covariates. Of the four types, only frequent volunteering is associated with lower CRP . The other three types of productive activity were not associated with CRP. The effects of other control variables were similar to the previous model. The final OLS model includes both number of roles and frequency of each activity. The results shown in Model 3 reveal that the frequency of volunteering is associated with lower CRP, but number of roles is no longer significant. The results lend partial support for Hypothesis 2 stating that frequency of participation within a role is negatively related to CRP. As described earlier, we also estimated the effects of productive activities on three categories of CRP ; the results are presented in the Supplementary Material. The pattern of results from the ordinal logistic regression is very similar to those obtained with OLS. As shown in Model 3 in the Supplementary Material, a unit difference in volunteering is associated with an 8% reduction in the odds of being in a higher category of inflammation . Our next step was to complete additional analyses to examine the role overload hypothesis. We tested two specifications in the OLS analyses: A quadratic term for overall frequency of productive activities and an interaction between the number of roles and the sum of activity frequency . Results from both tests give no evidence that high levels of engagement in multiple roles are detrimental to health in older adults . Finally, we tested for the hypothesis that the benefits of productive activity are stronger for adults 70 years or older . In doing so, we focused on volunteering frequency because the results from previous analyses revealed that volunteering was the most consistent predictor of CRP. We first tested the interaction of age groups and volunteer frequency on CRP. The result revealed that the model with the interaction term was a better fit than the model excluding the product term . Thus, we proceeded to further investigate the age differences. Stratifying the sample into 58-69 years and 70-85 years, we discovered that the frequency of volunteering has distinct effects on CRP in the two groups. Although frequency of volunteering is associated with lower CRP, the effect is actually stronger for those 70 or older. As shown in Figure 1, among people with low levels of volunteering, CRP levels differed little by age group. By contrast, among persons engaged in frequent volunteering, CRP was lower among the respondents 70 or older. People who are 70-85 years of age and very active as volunteers exhibited lower CRP than persons aged 58-69 who were not very active as volunteers. The test of Hypothesis 4 focused on volunteering-the most consistent predictor of CRP-but we conducted supplementary age-stratified analyses to determine whether the other forms of productive activity were related to CRP in either age group. Even in the subsample < 70, the other forms of productive activity were not related to CRP. --- Discussion Using nationally representative data from older adults, the current study showed that older people engaged in more roles of productive activities report lower CRP . However, when both frequency of participation and number of roles were considered simultaneously, frequency of participation in volunteering manifested the most robust association with CRP . The findings are consistent with previous research showing that volunteer activity improves various health outcomes . Volunteering is associated with better health, possibly through emotional gratification , feelings of generativity , and increases in social integration . Regularly engaging in volunteering has a special way of getting The study extends the research on health benefits of volunteering and clarifies the need for investigating not only self-reported measures of health but also biomeasures. Doing so will aid further understanding of the mechanisms between productive activities and health. The findings of this study point to a specific type of productive activity-volunteering-as an agent for maintaining lower levels of chronic inflammation. The analysis also yielded an intriguing finding related to age differences in the effects of productive activities: frequent volunteering is particularly beneficial on CRP among adults 70 years or older . NSHAP septuagenarians and octogenarians who were frequent volunteers had lower CRP levels than comparably active volunteers who were 58-69 years old. This finding shows that frequent volunteering benefits older adults in terms of having a "younger" biological profile. Although our findings are based on CRP as the outcome, the conclusion is consistent with findings of others that older people may reap more positive health benefits from volunteering than is the case for adults of younger ages . Because it might be the case that younger and healthier individuals are more likely to be selected into volunteering, we examined mean CRP levels in different age groups using a two-sample t-test. The results showed that mean CRP levels do not differ across these groups. This study examined the effect of multiple roles of productive activity including volunteering, employment, meeting attendance, and caregiving. The results, however, showed clearly that volunteering is the type of productive activity most likely to have health benefits. We wondered whether the emphasis on volunteering in prior studies may have neglected health benefits from other forms of productive activity, but the results from this study suggest otherwise. Perhaps some prior studies examined other forms of social activities but, like the present investigation, found volunteering to be the most consequential to health. Whatever the case, our findings add to the accumulated body of research demonstrating that volunteering is distinctive in its health-giving properties . We also hypothesized that role overload may nullify the purported health from productive activity . We found no evidence for the role overload hypothesis even if people were very active in many roles . Although our findings are consistent with much of the literature regarding the overall benefits of productive activity, we could not replicate the adverse effect on health due to a high time commitment to volunteering . Perhaps the role overload thesis applies to other health outcomes, but not to CRP, or it may be due to measurement issues described subsequently. This study is limited in important ways, and one may be related to our inability to replicate the adverse effects of very high time commitments in productive activities. Unlike other studies that capture actual hours of productive activity, the maximum value for productive activity in NSHAP is "several times a week." In comparison to accumulated hours in an activity, this is a crude measure, perhaps resulting in a ceiling effect. We thought that seven categories of frequency in most of the productive activities, coupled with four roles, would be sufficient to tap intense engagement in multiple productive activities, but it would be preferable to have a more detailed measurement of frequency within each role. Second, the measurement of productive activity in the NSHAP is notably limited. When volunteer activity is considered, the number of organizations and types of activities may have significant implications on volunteers' health, but these are not assessed in the survey. We used the available measures to examine the research question but welcome future research with more complete measurement of productive activity. Community engagement is an important component of productive activity, but our findings are limited by using one indicator: meeting attendance. It would be very helpful for future studies to tap other indicators of community engagement such as voting, leadership roles, and financial contributions to community organizations. Studies that include a more comprehensive measurement of productive activity may prove especially useful for explicating the mechanisms by which it is related to health. Third, a notable limitation of this study is the cross-sectional nature of the data, raising the concern that healthy individuals might be selected into participation in productive activities. Although the measurement of CRP obviously occurred after the productive activities reported in the interview, it is still possible that health selection is operant. To address this possible concern, we accounted for health lifestyle factors, chronic conditions, and depressive symptoms-and the benefits of productive activity on CRP remained despite the inclusion of these variables and socioeconomic resources. Although we think the risk of reverse causality invalidating the results is quite low, longitudinal tracking of productive activities and CRP would be especially helpful to check on the conclusions presented herein. Finally, the sample comprised community-dwelling adults, and the results cannot be generalized to those living in assisted living, nursing homes, or other types of congregate facilities. Future research needs to investigate whether older adults living in these environments reap the same health benefits from productive activities as was observed for community-dwelling adults in the NSHAP. In sum, the study not only provides empirical support to the generalization that volunteer activity is beneficial to health, but it contributes to the literature by revealing that such health benefits include lower inflammation. We are unaware of any previous study to report that productive activity, especially volunteering, is associated with lower CRP. Given that CRP generally rises with age, the benefits of volunteering to counteract that trend are noteworthy. Moreover, we uncovered that the benefit of frequent volunteering was actually stronger for the oldest members of the sample, suggesting the health utility of sustained engagement in volunteer activities. --- Supplementary Material Supplementary material can be found at: http://gerontologist.oxfordjournals.org.

The study investigates whether productive activities by older adults reduce bodily inflammation, as indicated by C-reactive protein (CRP), a biomeasure associated with the risk of cardiovascular diseases. Design and Methods: The study uses a representative survey of adults aged 57-85 from the National Social Life, Health, and Aging Project (N = 1,790). Linear regression models were used to analyze the effects of multiple roles (employment, volunteering, attending meetings, and caregiving) and the frequency of activity within each role on log values of CRP concentration (mg/L) drawn from assayed blood samples. Results: Number of roles for productive activities was associated with lower levels of CRP net of chronic conditions, lifestyle factors, and socioeconomic resources. When specific types of activity were examined, volunteering manifested the strongest association with lower levels of inflammation, particularly in the 70+ group. There was no evidence that frequent engagement in volunteer activity was associated with heightened inflammation. Implications: Productive activities-and frequent volunteering in particular-may protect individuals from inflammation that is associated with increased risk of hypertension and cardiovascular disease.

I N T R O D U C T I O N Adolescence is defined as the age group of 10-19 years and is characterized by rapid changes in the physical, biological and hormonal characters of the individual resulting into her/his psychological, behavioural and sexual maturation [1]. In the year 2009, there were 1.2 billion adolescents in the world and a vast majority of these, around 88%, lived in the developing nations. India accounts for the largest national population of adolescents, 243 million, in the world [2]. Adolescence is characterized by a remarkable upswing in the prevalence of many of the threats to children, including violence and abuse [3]. In 1999, the World Health Organisation Consultation on Child Abuse Prevention defined child abuse or maltreatment as 'all forms of physical and/or emotional ill-treatment, sexual abuse, neglect or negligent treatment or commercial or other exploitation, resulting in actual or potential harm to the child's health, survival, development or dignity in the context of a relationship of responsibility, trust or power' [4]. This broad-spectrum definition covers maltreatment by parents and caregivers also. UNICEF reported that an average of 86% children aged 2-14 years experienced some form of violent disciplinary practice at home [5]. Also, across the world, approximately 20% women and 5-10% men report sexual abuse during childhood, while 25-50% of all children report being physically V C The Author [2016]. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected] abused [6]. In the USA alone, almost six reports of child maltreatment are received by the state and local agencies every minute [7]. Moreover, across much of the world, families and societies treat girls and boys unequally, with girls disproportionately facing privation, lack of opportunity and lower levels of investment in their health [8][9], nutrition [10] and education [11]. In India too, a study in 2005 by the Ministry of Women and Child Development, Government of India , reported that two out of every three children were physically abused, more than half were sexually abused, almost half were emotionally abused and 7 out of every 10 girl child respondents reported neglect of one form or the other by family members [12]. Child abuse is a violation of basic human rights. According to WHO, in the year 2000, there were an estimated 57 000 deaths attributed to homicide among children aged less than 15 [6]. But ironically, a large number of child deaths are not routinely investigated and hence the real figures about number of fatalities due to child abuse are difficult to establish, even in the most developed parts of the world [13][14][15][16][17]. Moreover, the data on non-fatal child abuse and neglect are also questionable with regards to completeness and usefulness come in wake of the fact that in many countries, there are no legal or social provisions in place to record, search for and eventually respond to child maltreatment [18]. In addition, there is profoundly diverse cultural understanding and equally variable legal definition of child abuse and neglect between countries [19]. In India, the acceptance of child rights as primary inviolable rights by the law and lesser so by the society is a recent development, highlighting that the understanding of the problem is new and still in its infancy [12]. While many studies report the prevalence of child abuse in India, only few have tried to explore the social factors that may have a role in predisposing children to the various forms of abuse [12,[20][21][22][23][24], and the various barriers to self-disclosure of such incidents to support groups like family, friends and child protection services [20]. The authors conducted a study to determine the prevalence and identify the determinants and disclosure barriers of child maltreatment among adolescent girls and also examine the various family characteristics that put the girl child at risk of various types of abuse, mainly physical, sexual and emotional abuse and neglect. The study was carried out in an urbanized area of Delhi, India. The present article describes the findings of the study pertaining to prevalence and determinants of child maltreatment, including family characteristics, among its study participants. --- M A T E R I A L S A N D M E T H O D S The study was conducted in Najafgarh, Delhi, which is one of the field practice areas of the Department of Community Medicine, Vardhman Mahavir Medical College and Safdarjung Hospital, New Delhi, India. Najafgarh is a semi-urban area that comes under the South-West district of Delhi and has a total population of close to 0.9 million with a literacy rate of approximately 83% [25]. As the purpose of the study was to determine prevalence and determinants of child abuse and girl child neglect in adolescent girls in their family or close living environment, it was realized that responses of study participants may be influenced by collection of data in families. Hence, it was planned to conduct the study in schools, place where children can be relatively freely enquired about incidents of maltreatment. There are four government girls' senior secondary schools in the town area of Najafgarh and the study was conducted in all these four schools. The required sample size of study participants was calculated using the formula 4PQ/d 2 where P ¼ prevalence, Q ¼ 1 À P and d ¼ relative precision. Here, prevalence was taken as 0.34 based on a previous study on child abuse and neglect carried out in 2005 by the MWCD, GOI [12], and relative precision was taken as 10%. Considering a noresponse rate of 10%, the sample size came out to be 855. The combined student strength of the four schools was approximately 6000, and hence, stratified random sampling was done to select study participants. One section of each of the classes 7th-12th from each of the four schools was randomly selected, and all the students present in the class on the day of data collection were included in the study. Permission was taken from the office of the Directorate of Education, South-West district of Delhi, and the school principals, and written consent was also taken from the adolescent girls before including them in the study. A self-report, pre-tested and semi-structured questionnaire in the Hindi language was used for the purpose of the study. It was modified from various standard questionnaires [12,26,27] and included questions on the socio-demographic profile and family characteristics of the study population, and child maltreatment-physical abuse, sexual abuse, emotional abuse and neglect. The questionnaire was initially in the English language. It was translated to the local Hindi language by the Hinditranslation section of Vardhman Mahavir Medical College and Safadarjung Hospital , New Delhi. The Hindi version was then retranslated to English by two separate individuals who did not know anything about the study, and matched with the original English version. There was no difference in the two English versions that were obtained. The questionnaire was validated during pre-testing done in August 2011 among 48 female students of Class 10 of a government co-educational senior secondary school located in Najafgarh. Subsequent to this, the questionnaire was modified. The questions-what is the frequency of abuse, whether matter of sexual abuse was reported to police and whether the family motivates the respondent on her achievements-were added to the final questionnaire. The study did not specifically enquire about maltreatment within schools, either by teachers or fellow students in order to prevent conflict of interest with school authorities. Clearance for conducting the study was taken from the Institutional Ethical Committee of VMMC and SJH, New Delhi. Confidentiality was ensured at all stages of data collection and analysis. Data were collected between April 2012 and March 2013. The data collection session was followed by an interactive session in which the queries of the study participants on what can be done if they face child maltreatment were answered. The primary investigator explained the rights of children in India and the helpline and emergency services that are available in wake of child maltreatment. The study participants were also allowed individual interaction with the primary investigator if they wished to discuss any of their issues. Three study participants were brought to Vardhman Mahavir Medical College and Safdarjung Hospital, New Delhi, for consultation with a psychologist. --- O P E R A T I O N A L D E F I N I T I O N S The definitions of child maltreatment used in the present study have been adopted from the 1999 WHO Consultation on Child Abuse Prevention [4]. The definitions of family, father, mother and other family attributes were taken from the 2011 training manual of the national census of India [28] and from the Oxford dictionary [29], and the socioeconomic status scale used for estimation of socioeconomic class of the families of the study participants was the revised Kuppuswamy scale [30]. All the operational definitions used in the study have been given in Supplementary Appendix A of the present article. The period in which incidents of abuse were enquired about was different for different forms of abuse. This was done keeping in mind the ability of an individual to recall the circumstances and frequency of abuse. While incidents of certain forms of physical abuse may be forgotten , others like 'tied with rope' or 'burned/scalded' may remain in the memory for long durations. Incidents of sexual abuse also remain in the memory of victims for long periods or even a lifetime. Recall for incidents of emotional abuse may be shorter because this form of abuse is mainly verbal. Neglect is a process and hence it was enquired for over the lifetime of the study subjects [31]. The period for which the various forms of abuse and neglect were enquired for has also been mentioned in the Supplementary Appendix A of the present article. All data entry was done on MS Excel and a master sheet was created. A total number of 174 variables were identified and suitable coding was done for each one of these. After entry of every 10 questionnaires, one was picked randomly and data entry was verified. An independent person verified data entry of two randomly chosen forms after entry of every 25 questionnaires. The data were analysed using SPSS version 21. All the variables were analysed using descriptive statistics to calculate frequencies, mean, range and standard deviation. Multivariate logistic regression analysis was done for selected variables to calculate adjusted odds ratio. --- R E S U L T S A total of 1060 adolescent girls became a part of the study. The response rate was 99.3%. The mean age of the study subjects was 14.5 years . Majority of the study subjects were in their mid-adolescence , while 28.9% and 11.8% were in early and late adolescence, respectively. Majority of the study subjects belonged to nuclear families. The mean number of family members was 6.1 , and a total of 78.5% of the study subjects belonged to families with five to eight members. The father figure in the lives of majority of the study subjects was their biological father, followed by step-father and grandfather . Similarly, the mother figure in the lives of majority of the study subjects was their biological mother, while stepmother was the mother figure in the life of 1 study subject. In a total of 41 study subjects, there was no father figure, and in a total of 6 study subjects, there was no mother figure. There were 3 study subjects who had lost both their parents, two of whom lived in an orphanage, while the third was looked after by her elder sister. The head in the families of majority of study subjects was their father, followed by the mother , grandfather , grandmother , paternal uncle , brother and sister . There were 2 study subjects who lived in an orphanage and, hence, this question did not apply to them. The fathers of 62.0% of the study subjects had high education, while mothers of 65.9% of study subjects had received no or low education. Also, fathers of 99.4% of study subjects were employed, while mothers of 87.4% of the study subjects were housewives. Majority of the study subjects belonged to families of low socioeconomic status . Out of the total 1060 study subjects, 312 had not faced any form of maltreatment in their lives, while the remaining 748 had faced some or the other form of maltreatment. The most common form of maltreatment faced by the study subjects was physical abuse , followed by neglect . Among all the 452 study subjects 3). Among all the 282 study subjects who had faced sexual abuse, 263 had faced sexual abuse during situations ever, 101 had faced some form of severe sexual abuse ever and 7 had been victims of sexual assault ever. The most frequent perpetrators of severe sexual abuse were the relatives, friends or neighbours of the study subjects . Out of all the 402 study subjects who had faced emotional abuse, the most common form was comparison of the study subjects with their siblings or other children and demeaning in relation to them, and this was seen in 292 cases. In 150 Prevalence and Determinants of Child Maltreatment 231 cases, the families had wished that the study subject 'was either not born' or 'was dead'. The most frequent perpetrator of emotional abuse of all forms was the mother of the study subjects. The frequency of emotional abuse was mostly between 1 and 5 times in the past 6 months, with the median frequency being 2 . Out of a total of 425 study subjects who reportedly were neglected by their families, most of the study subjects said that they were made to do household work by unwillingly compromising on their studies. Also, 142 of the neglected study subjects reported that their families did not motivate them despite their achievements in academics or sports or other co-curricular activities. Of all the 386 study subjects who were forced to do household work, 75 were made to do so 'almost every day' . Out of the total 1060 study subjects, 367 reported presence of family-related predictors of maltreatment. A total of 257 study subjects reported that there were excessive arguments between their parents. Also, 80 study subjects reported domestic violence. The fathers of 94 and mothers of 47 study subjects were themselves victims of maltreatment in their childhood . On multivariate logistic regression analysis, no or low education of father was found to significantly increase the odds of physical abuse in the study subjects. The odds of physical abuse were significantly lower in the study subjects whose mothers were housewives. The odds of sexual abuse increased in the mid and late adolescence and this was found to be significant. Also, odds of emotional abuse were significantly higher in the study subjects aged 17-19 years and those whose fathers had received no or low education. The odds of neglect were significantly lower in the study subjects whose mothers were housewives. The presence of family predictors of maltreatment significantly increased the odds of all forms of maltreatment in the study subjects . The presence of family-related predictors of maltreatment was associated with all forms of maltreatment in the study subjects and this was statistically significant . On multivariate regression analysis, excessive arguments between parents, domestic violence, imprisonment of family member, physical impairment in study subject, history of maltreatment in father during childhood and history of maltreatment in mother during childhood-all significantly increased the odds of physical abuse in the study subjects. The odds of sexual abuse in the study subjects were significantly increased if there were excessive arguments between parents, imprisonment of family member, history of maltreatment in father during childhood and history of maltreatment in mother during childhood. Also, odds of emotional abuse were significantly increased in the study subjects if there were excessive arguments between parents, domestic violence, imprisonment of family member, history of maltreatment in father during childhood and history of maltreatment in mother during childhood. The odds of neglect were significantly higher in the study subjects if there were excessive arguments between parents, mental impairment in family member, history of maltreatment in father during childhood and history of maltreatment in mother during childhood . --- D I S C U S S I O N In this study, we found that 7 out of every 10 adolescent girls had faced some form of maltreatment in their life. A total of 42.6% adolescent girls were physically abused, 26.6% sexually abused, 37.9% emotional abused and 40.1% reported neglect. This is similar to the findings of Wijma [32], Sarkar [21] and Patel and Andrew [23]. The study on child abuse in India in 2007 by the MWCD, GOI had reported higher prevalence of maltreatment [12]. The difference may be because the above-said study had used a different research methodology and also because it was carried out in 13 states of India among a heterogeneous group of adolescent population. The most common form of physical abuse faced by the adolescent girls in this study was slapping or hitting, which is similar to the findings of the study on child abuse in India in 2007 by the MWCD, GOI [12], the WorldSAFE study [24] and Sarkar [21]. The present study found no association of age and type of family with physical abuse, whereas the previous studies reported greater prevalence in younger children aged 5-12 years and those belonging to nuclear families. Physical abuse was reported more by adolescent girls from families of low socioeconomic status, whereas Deb and Modak reported it more in adolescents from high-income families [20]. While the present study found that odds of physical abuse increased when the father of the study subjects had received no or low education but no association with the education of the mother, and the odds decreased when the mother was a housewife, suitable literature to compare these findings was not found. Similar to the present study, Deb and Modak also reported that physical abuse was more common in families with disturbed environment [20]. Moreover, adolescent girls with physical impairment were more likely to be physically abused and this as a risk factor was also documented in the 1995 report of the L'Institut Roeher Institute in Toronto [33]. The prevalence of sexual assault in the present study was found to be 0.7%, whereas the study on child abuse in India in 2007 by the MWCD, GOI [12] reported a prevalence rate of 5.4% and Patel and Andrew reported it as 6% [23]. The difference could be because of restriction of the definition of sexual assault in the present study as penetration of vagina, whereas in the other two studies quoted above, it included penetration of anus as well as oral sex also. Our study reported that the perpetrators of severe sexual abuse were relatives, friends or neighbours and this finding is similar to the findings of the study on child abuse in India in 2007 by the MWCD, GOI [12], Deb and Modak [20] and Patel and Andrew [23]. In our study, we also found that odds of being sexually abused were greater in older adolescents, as was also reported by the study on child abuse in India in 2007 by the MWCD, GOI [12] and Patel and Andrew [23]. Our study also reported greater odds of facing sexual abuse in adolescent girls belonging to families in which there were excessive arguments between parents, as was seen in studies by Deb and Modak [20] and Patel and Andrew [23]. The present study also found increased odds of sexual abuse in adolescent girls who belonged to families with history of imprisonment of family member and maltreatment of either parent during childhood. In our study, comparison with siblings or other children and demeaning in relation to them was the most common form of emotional abuse faced by the study subjects, and this is similar to the findings of the study on child abuse in India in 2007 by the MWCD, GOI [12]. The most common perpetrator of emotional abuse reported by our study was the mother, and the results were similar in the above-mentioned study. A history of excessive arguments between parents and domestic violence were associated with increased odds of emotional abuse in the present study, and this is similar to the finding of Deb and Modak [20]. A history of imprisonment of family member and maltreatment in either parent during childhood were also found to be associated with increased odds of emotional abuse. In the present study, almost one-third adolescent girls were made to do household work by unwillingly compromising on their studies. Neglect was more in families with no or low education of parents and working mothers. This may possibly be because parents are less aware of needs of children and also working mothers may be less able to devote time to them. Our study highlights that a victim of one form of abuse also has increased odds of being a victim of other forms of abuse as well, supported by the findings of Patel and Andrew [23]. Our study had much strength. Stratified random sampling was done and data were collected away from the family environment to avoid fear of revealing true information. The questionnaire used was developed from previously standardized and validated questionnaires and covered most aspects of child maltreatment. The study had certain limitations too. The results cannot be generalized because adolescents living in other environments, like those staying at home, on streets or in shelter homes, or studying in private schools were not studied. Also, recall bias and under-or over-reporting may be a problem. Moreover, child maltreatment in schools was not enquired about to avoid any conflict with the consenting authorities. --- C O N C L U S I O N Child maltreatment can have drastic effects on the psychological and physical health of a child, but the signs may be subtle . It is important to sensitize the school and health authorities to identify these signs and enquire about maltreatment from children and adolescents. Parental and family counselling to shun violent methods of discipline should also be given a priority. The local nongovernmental organizations should be involved in providing protection and support to abused children. The mass media too should highlight the ill-effects as well as the legal repercussions of child maltreatment. Child safety and protection authorities should provide support to abused children and children at risk of maltreatment on receiving calls on helpline numbers. Further studies to investigate maltreatment of other groups of adolescents and also the various predictors or risk factors for maltreatment should be taken up to better understand the various aspects of child maltreatment. --- S U P P L E M E N T A R Y D A T A Supplementary data are available at Journal of Tropical Pediatrics online. --- A C K N O W L E D G E M E N T S Office of the Deputy Director, Directorate of Education, South-West district of Delhi and the school principals who consented to conduct of the study in schools under their jurisdiction.

The study examines family characteristics that put adolescent girls at increased risk of abuse, mainly physical, sexual and emotional abuse and neglect. Stratified random sampling was done among classes 7th to 12th of government girls' schools of a semi-urban area of Delhi, and a total of 1060 adolescent girls participated. Majority were in mid-adolescence. Approximately 70% study subjects faced at least one form of maltreatment. Physical abuse was faced by 42.6%, sexual abuse by 26.6%, emotional abuse by 37.9% and neglect by 40.1% of study subjects. The most frequent perpetrator of physical and emotional abuse was mother, and of sexual abuse were friends, relatives or neighbours. No or low education of father increased odds of physical and emotional abuse, while odds of physical abuse and neglect were lower if mothers were housewives. Excessive arguments between parents and history of maltreatment in parents increased odds of child maltreatment in study subjects. K E Y W O R D S : child maltreatment, child abuse, sexual abuse, physical abuse, psychological/ emotional abuse, neglect.

Introduction --- B ehavioural risk factors, such as smoking, drinking too much alcohol, nutritional choices or physical inactivity, often acquired in childhood, can condition health status in adult life. 1 Specifically, physical inactivity, a diet rich in red meat, low in fibre and low consumption of fruits and vegetables, smoking and a high body mass index have been shown to be associated with a worse health status of individuals. [2][3][4] However, these factors can be modified throughout life, reducing the risk of several pathologies and improving overall health. 5 People who have these behaviours often take less care of themselves, and therefore also participate less in screening programmes and other preventive activities. 6 This is often because certain social determinants may condition their behaviour. In fact, socio-economic status and social inequalities are related to risk factors, indeed, the higher the socioeconomic level, the greater the presence of risk factors. 7 Lower socio-economic status and educational level, worse employment conditions and place of residence are some of the social inequalities that can lead to not undergoing preventive activities such as regular health check-ups or participation in screening programmes. [7][8][9][10] The implementation of colorectal cancer screening programmes is widely recommended, in fact, both the United States Preventive Services Task Force and the Council of the European Union recognize its potential and recommend organized programmes. 11,12 Furthermore, cancer screening programmes can reduce cancer-specific and all-cause mortality. 13 Despite this, CRC screening programmes are not implemented equally around the world, depending on the CRC incidence, economic resources and healthcare structure. 14 Usually, they differ in the screening method-Faecal Occult Blood Test , optical sigmoidoscopy, optical colonoscopy or computed tomography colonoscopy-the organizational characteristics-population-based or opportunistic-and the target population. Moreover, high participation rates are essential in screening programmes in order to be cost effective and to achieve health benefits. [15][16][17] Therefore, knowing how both social inequalities and participation in preventive activities influence participation CRC screening specifically could help improving participation rates. The goal of this study is to identify the association between nonparticipation in CRC screening programmes and social inequalities and risk factors or participation in preventive activities. --- Methods --- Data sources A systematic review of the literature was conducted between June and September of 2019 using the following databases: CINHAL, Medline, Scopus, Social Sciences Citation Index, Embase and PsycINFO. --- Search strategy The search strategy combined a wide range of Medical Subject Headings and free text terms related to screening, CRC, participation and social inequalities. This search was limited to articles published from 2000 to June 2019. The search strategy used in Medline is given in Supplementary table S1. --- Screening and review process Studies were included according to this criteria: studies focusing on risk factors, participation in preventive activities , inequalities according to the Dahlgreen and Whithead model 18 and participation in any type of CRC screening programmes with any type of screening method ; primary studies; studies in which 45-75-year-old people at average -risk of CRC were offered to screening; published in peerreviewed journals between 2000 and 2019 in English, French, Portuguese and Spanish; where relationships between social inequalities and risk factors or participation in preventive activities and participation in CRC screening were analyzed and whose quality was fair or good. Two authors screened all titles and abstracts of the identified references, with a third in case of discrepancy. Then, the whole text of the selected studies was analyzed for eligibility according to inclusion and exclusion criteria. Disagreements were resolved by consensus. Reviewers eliminated articles not related to the study as shown in Figure 1. This systematic review was performed according to the Preferred Reporting Items for Systematic Reviews and MetaAnalyses guidelines. 19 Two independent authors extracted data, including the name of the first author, year of publication, country and region, methodology, exposure and main results. The description of the included studies is shown in table 1. Variables are presented as percentages for participation as odds ratios and 95% confidence intervals. --- Study methodological quality assessment Methodological assessment was carried out independently in parallel by three researchers according to the Quality Assessment Tools of the National Heart, Lung and Blood Institute for quantitative studies to judge each study in terms of "good," "fair" or "poor" quality. 20 If the ratings differed, then reviewers discussed the article and a final decision was made in an attempt to reach consensus. --- Results --- Literature search In total, 1893 studies were identified in the bibliographic search and after removing duplicates, 736 were screened for title and abstract. From those, 144 full-text articles were assessed for eligibility and 135 were excluded according to the following reasons: the study did not analyze the relationship between preventive activities or risk factors and social inequalities individually with participation in CRC screening programmes, did not study 45-75-year-old population, studied high-risk population or the methodological quality was poor. After evaluating the full text, nine articles were included in the systematic review. The list of excluded articles is shown in Supplementary table S2. --- Quality and characteristics of the selected studies All studies included in this review were descriptive and observational retrospective cross-sectional except one, 21 which was a randomized multicentre population-based controlled trial. Five of the studies were conducted in organized screening programmes, [21][22][23][24][25] whereas the rest were opportunistic, [26][27][28][29] using different tests. In four of them, only FOBT was used, 22,24,25,27 while in three of them, 21,28,29 colonoscopy and/or sigmoidoscopy were offered to the people invited, in addition to FOBT as a screening test. According to the age of the invited people, four studies included 50-74/75-year-old population [23][24][25][26]29 and the rest studied people from 50 to 65 22 or 69 27 whereas in only one study the population was 55-64. 21 Finally, in one study the test used was the guaiac FOBT , 23 and in another one the colonoscopy exclusively. 26 In all the studies, the analyzed data were self-reported and, in addition, in three of them, 23,25,29 national survey data were added and in two studies medical records. 26,27 Regarding the gender of the persons included in the studies, it should be noted that only women were included in two of them. 22,28 Three of the studies were conducted in France, 22,23,25 three in the United States, 26,28,29 and the other three in Greece, 27 Italy 21 and Norway. 24 Of the nine studies finally included, according to the aforementioned Quality Assessment Tools for quantitative studies, 20 five were of good quality [22][23][24][25]29 and four of fair quality . 21,[26][27][28] The results of the quality assessment are shown in Supplementary tables S3 and S4. The strengths of the observational included are the clarity of the objective and description and selection of the study population, the different levels of exposure and their measurement regarding the factors studied and the measurement of the results. On the other hand, the main weaknesses were that the sample size was not adequately justified. Due to the characteristics of the studies, criteria 10 and 13 were not applicable for descriptive studies. The major limitation of the intervention study was the absence of blinding in the measurement of the outcome variable. We eliminated two studies due to their low methodological quality, however, both include variables that have also been collected in the studies included in this review. --- Synthesis of evidence The main results of the included studies are described below according to the factors that may be related to participation in CRC screening programmes grouped into social inequalities, participation in preventive health activities and risk factors. It is necessary to emphasize that there is a high variability in the studied variables among the selected studies. The factors analyzed in each study are shown in Table 2. --- Social inequalities Gender was analyzed in seven of the nine studies 21,[23][24][25][26][27]29 and in the other two, only women were included. 22,28 In studies that analyzed both genders, the percentage of women was between 51% and 62%. Five studies analyzed whether gender was related to participation in screening, 21,23,24,26,29 and it was only in the Norwegian study that this relationship was established, 24 concluding that women participated more than men. In addition, five studies analyzed all possible factors related to participation by gender. [23][24][25]27,29 For women, having a family doctor, 27 district, 23 additional health insurance, 23 age, 23,24 smoking status, 24,25 BMI, 24 physical activity, 24 consultations with a General Practitioner , 25 no consultation with a specialist, 25 alcohol, 25 first degree relative with CRC, 25 regular screening using mammography 23,25 and having a Pap smear 23 were the analyzed variables. As far as men are concerned, obesity, 26 age, 23 health insurance, 23 district, 23 additional health insurance, 23,25 having consulted a medical specialist, 23,25 smoking status, [23][24][25] occupation, 24 non-native ethnic background, 24 BMI, 24 guideline adherence, 29 endoscopic screening method, 29 social class, 25 number of consultations with a GP 25 and physical activity were analyzed. 21 Only in two studies was the relationship between all these factors established without any stratification by gender. Thus, whereas Anderson et al. 26 reported an association between obesity, having a higher level of education and living with a partner with participation, Senore et al. 21 did so with having a family history of CRC, physical activity and smoking. When only women were included in the studies, participation was associated with BMI, 22 having consulted with a gynaecologist in the past 12 months, 22 annual income, state of health involving medical visits and being retired. 28 Ethnicity has been analyzed in three studies, 24,26,28 and only Knudsen et al. 24 established that having a non-native ethnic background was associated with a higher probability of participation than natives for men in their study in Norway with FOBT, the association was not significant in women. Anderson et al. 26 reported the association between nonparticipation in the colonoscopy-based opportunistic screening programme and educational level, thus having high school diploma is associated with greater participation in both genders. Neither Fon Sing et al. 23 found an association, although in this case the screening test was FOBTg. It should be noted that no association was found in any of the studies in which the screening test was FOBT. 21,22,24,27,28 Regarding the employment situation, Katz et al. 28 noted that being retired or being a volunteer is associated with participation in three recommended cancer screening programmes with reference to the unemployed or disabled. On the other hand, Senore et al. 21 established that, for both genders, employed people participated in a programme based on FOBT and sigmoidoscopy more than those who were unemployed. Having health insurance has also been analyzed 22,23,25,27,28 and an association between having additional health insurance and a higher probability of participating in CRC screening was only found in the studies by Fon Sing et al. and Sicsic et al. 23,25 using FOBTg and FOBT, respectively. The place where people live has also been considered, 23,24,27 but only Fon Sing et al. 23 found an association: those living in a district where the pilot programme was implemented have a higher probability of participating. Although the relationship between marital status and participation was considered in four studies, it was not established in any of them. 21,[23][24][25] Income was analyzed in two studies, 23,28 and a relationship was only established with participation in Katz et al. 28 Those who had the highest household income were more likely to participate than those who had the lowest income in an opportunistic screening programme with FOBT, colonoscopy or sigmoidoscopy. --- Participation in other preventive activities Uptake in other preventive activities or use of the health system's resources, mainly those related to the various screening programmes 22,23,26,27 or visits to the GP 22,25,28 or specialist 22 have been studied. However, although Anderson et al. 26 studied the relation between having taken a past CRC screening test, they found no association with participation in colonoscopy-based CRC screening. Dimitrakaki et al. 27 took into account participation in prostate and breast cancer screening programmes and did not observe any association with participation in CRC screening. However, Fon Sing et al. 23 concluded that having a mammogram or Pap smear test was associated with a higher probability of participating in CRC screening than those who had never taken it, or had not taken it in the last 2 or 3 years, respectively. On the other hand, concerning visits to the GP, 22,25,28 gynaecologist 22 or specialist, 23,25 different results were reported. Bertaut et al. 22 observed a relationship between participation and visiting the gynaecologist but not with visits to the GP or gastroenterologist. On the contrary, visiting the GP 25,28 or specialist 23,25 increased the probability of participating, while it was only established among men in the Fon Sing et al. 23 study. --- Risk factors The relationship between obesity and participation had been analyzed in four studies. 22,24,26,29 Seibert et al. 29 and Knudsen et al. 24 discontinuation was analysed in two of them. 24,25 Two studies analyzed discontinuation and in both they found an association with BMI. Knudsen et al. 24 found that both men and women with a BMI over 35 are more likely to discontinue than people with normal weight 24 while Seibert et al. 29 only found an association in men with a BMI over 40 Only Anderson et al. 26 analyzed the influence of obesity on participation and noted that people with a BMI greater than 30 are at greater risk of not participating in screening. The results of the influence of physical activity are diverse, while Bertaut et al. 22 found no association Knudsen et al. 24 established a relationship with discontinuing CRC screening although it was only stated in women in the stratified analysis. In turn, Senore et al. 21 concluded that physical activity is only related to participation in men. In both studies, the higher the level of physical activity, the greater the likelihood of participating. Tobacco has been widely studied, [21][22][23][24][25][26][27][28] although only four studies established an association with participation. 21,[23][24][25] Two studies concluded that being a smoker decreased the probability of participating, 21,24 while Sicsic et al. 25 only established that relationship in women and Fon Sing et al. 23 in men. On the contrary, only four studies analyzed the relationship between participation and alcohol consumption [22][23][24][25] although it was only established in one of them, in fact, safe consumption 30 was associated with a greater probability of participating. 25 Moreover, although having a family history of CRC had been analyzed in three studies, 21,22,26 but only among women, a greater probability of participating was established. 21 Regarding general health status, although it had been analyzed in four studies, only Senore et al. 21 established an association. In fact, worse health status was associated with a higher probability of nonparticipation. Finally, it must be noted that diabetes and acetylsalicylic acid consumption, 26 eating fruits and vegetables regularly 22,24 and dental status 23 had also been studied but none of the studies established a statistically significant relationship between these factors and participation. --- Discussion The literature related to health risk factors and non-participation in CRC screening programmes is scarce. Moreover, it is even scarcer when social inequalities are also considered. It has been observed that certain characteristics can increase the risk of not participating in screening even though there is not a clear pattern. Nevertheless, some risk factors have been identified in this review such as being under 60 years old, obese, smoker and sedentary as risk factors for not participating in the CRC screening programme. As far as age is concerned, it has emerged as a risk factor for not participating in the studies in which the analysis is not stratified; indeed, in some of the studies in which the analysis is stratified by gender, 22,23 the risk is doubled and even quintupled with ORs ranging from 2.0 to 5.58. 23 Obesity is also a factor to be considered, especially the BMI over 35, as it affects not only participation but also adherence. Smoking has also been shown to be a risk factor for not participating in screening, namely, Senore et al. 21 in their clinical trial concluded that smoking prevented people from participating . Finally, physical activity has also been shown to be a factor, mainly with regard to adherence to screening. 21,24 In addition, it is necessary to highlight how gender influences the other factors. It should be noted that some of the risk factors or preventive activities influence men and women differently, as can be seen from the stratified analysis by gender. 20,[22][23][24]26,28 In fact, Senore et al., 20 in the only clinical trial included in this review, conclude that despite several factors affecting participation these are considerably reduced in the stratified analysis. This should be taken into consideration when developing specific strategies to promote participation by decreasing the influence that gender may have. In accordance with our results, several studies have stated that people who attend primary healthcare infrequently, generally have poor concern about their health and do not participate in preventive activities. 17,31,32 Visiting the GP or specialist increases the likelihood of participation, indeed, Zapka et al. 17 stated that people who consider their health as good/excellent are less likely to participate than those who consider it acceptable/poor. What is more, our results are consistent with other studies conducted from a qualitative approach in which the socio-cultural context of the place where the invited person lives increases the risk of non-participation. 33,34 The type of work , precariousness, not being able to go to the doctor during working hours, etc. may pose barriers to participation. 35 The main limitation of this review is the heterogeneity of studies, regarding the lack of interventional studies, substantial variability for the studied factors, characteristics of the screening programmes, location, health system, employment, etc. Concerning the countries where the studies have been carried out, it should be noted that the characteristics of the health system and the screening programme itself and the socio-economic level of the people involved might influence the variability of the results. In fact, participation in Norway is much higher than in France, where screening is organized. A similar situation exists in the USA and Greece, where participation is lower maybe due to opportunistic screening. With these disparities, in addition to the factors mentioned, there could be other factors, cultural or related to trust in the health system, which also influence participation and have not been taken into account in these studies. Nevertheless, this heterogeneity allows us to design studies in which the same or similar variables can be studied, in order to compare them. On the other hand, it should be borne in mind that the data had been collected from self-administered questionnaires in several studies. With regard to the quality of the included studies, they do not indicate whether the factors that may influence participation were prior to screening which could help to establish a cause-effect relationship as well as an association. A more detailed description of the rationale for sample size and effect size is also missing. Studies that improve these two methodological aspects would be very useful for designing policies or interventions to increase participation in this and other screenings. In turn, the main strength of this review is that it provides a wide overview of the factors that may be involved in non-participation, including individual characteristics and habits related to health and social factors. Considering that characteristics of the health system, universal coverage and an organized screening programme, facilitate participation. 36,37 and that having unhealthy behaviours increases the risk of non-participation, basing screening on Primary Healthcare may be the key to improving participation as well as modifying unhealthy lifestyle habits. In fact, it is in Primary Healthcare where it is possible to address the determinants of health by implementing specific policies and actions and to empower people to improve their health by tackling risk factors, behaviours and inequalities that we have seen influence participation in CRC screening. 38 Primary Healthcare could play a key role in encouraging patients to adopt healthier behaviours by identifying risk factors and promoting access to the health system. Nevertheless, more interventional studies are needed in this area to analyze in depth how all these factors influence participation and each other and to be able to design interventions aimed at inequalities or risk factors with specific approaches. --- Conclusions In summary, the relationship of health risk factors with nonparticipation in CRC screening programmes taking into account social inequalities has not been extensively analyzed. Consequently, the results observed so far do not allow a pattern to be established. The heterogeneity on the design of the screening programmes in different countries as well as the differences in their health systems make it difficult to compare the results. Being under 60 years old, obese, smoker and sedentary have been considered as risk factors for not participating in the colon cancer screening programme. Gender had not shown statistically significant differences. Regardless of the type of health system, visiting a doctor implies an awareness of self-care and is therefore related to a lower risk of nonparticipation in cancer screening programmes. Inequalities affect lifestyles and risk factors, despite this, it has not been possible to establish their influence on participation by isolating the effect of other factors and hence more studies are needed to identify them. --- Supplementary data Supplementary data are available at EURPUB online. --- Statements of ethical approval Ethical approval for this study was obtained from the Research with Drugs Ethics Committee of the Basque Country . --- Conflicts of interest: The authors report no conflicts of interest. --- Key points • Unhealthy behaviours , age and obesity have a negative influence on the participation in CRC screening programmes • Worse health is often associated with non-participation in screening programmes • Interventional studies can be very helpful in determining the factors that need to be addressed to improve participation --- Efforts needed for preventing breast and colorectal cancer through changes in dietary patterns Patrick Mullie 1,2 , Gilles Guillot Background: Prospective cohort studies on diet and cancer report risk associations as hazard ratios. But hazard ratios do not inform on the number of people who need to alter their dietary behaviours for preventing cancer. The objective of this study is to estimate the number of people that need to alter their diet for preventing one additional case of female breast or colorectal cancer. Methods: Based on the largest prospective studies done in the USA and in Europe, we computed the number of subjects who need to alter their diet. Results: For preventing one case of breast cancer, European women should increase their fruit consumption by 100 g/day during 33 000 person-years, and US women by 60 g/day during 10 600 person-years. For vegetables, European women should increase their consumption by 160 g/day during 26 900 person-years and US women by 100 g/day during 19 000 person-years. For preventing one case of colorectal cancer, European subjects should decrease their red meat consumption by 20 g/day during 26 100 person-years, and US subjects by 30 g/day during 8170 person-years. For processed meat, European subjects should decrease their consumption by 20 g/day during 17 400 person-years, and US subjects by 10 g/day during 7940 person-years. Conclusions: Large number of subjects would need to alter their intake of fruits, vegetables, red and processed meat during many years in order to prevent one additional breast or colorectal cancer.

Background: Colorectal cancer (CRC) screening programmes require high levels of participation in order to reduce mortality. To improve participation rates, it is necessary to identify the health risk factors and social inequalities associated with non-participation. Methods: A systematic review was conducted between June and September of 2019 in six databases: CINHAL, Medline, Scopus, Social Sciences Citation Index, Embase and PsycINFO. Studies assessing the relationship between health risk factors, participation in preventive activities and participation in CRC screening were included. Methodological assessment was carried out according to the Quality Assessment Tools of the National Heart, Lung and Blood Institute. Results: A total of nine studies that analyze participation in both organized and opportunistic screening programmes using any type of screening method were finally selected. Data were mainly self-reported although in two studies medical records were also studied. We identified several variables: gender, body mass index, consultation with a doctor or a specialist, educational level, employment, health insurance, residence, ethnicity, age, marital status, income, other preventive activities, obesity, physical activity, smoking, family history of CRC and general health status. Conclusion: The scarcity of studies linking risk factors, social inequalities and participation in preventive activities for participation in screening in the same study makes it difficult to reach definitive patterns related to non-participation in CRC screening programmes. Nevertheless, being under 60, obese, smoker and sedentary have shown an association with nonparticipation as well as not visiting a doctor.